Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

$MSFT TESTS NEW CHIP COOLING TECH Microfluidic cooling etched onto silicon cut GPU temps by 65% -- suggesting hyperscalers might push cooling innovation in-house instead of buying full rack systems. $VRT slid 7% on the headline 👀

StockSavvyShay's profile picture

Shay Boloor

310,588 subscribers

107,125 görüntüleme • 10 ay önce •via X (Twitter)

Bilim & TeknolojiHaberler & PolitikaFinans
LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

🔥 Gujarat start-up - From Earthquake to Innovation: The Story of Mitticool 🔥 After losing everything in the 2001 Gujarat earthquake, Mansukhbhai Prajapati didn’t just rebuild his life, he reinvented refrigeration. Inspired by a newspaper headline about a “poor man’s fridge,” he created Mitticool: a clay refrigerator that keeps food fresh for days without a single watt of electricity. 🏺 Made from simple clay, Mitticool uses evaporative cooling to preserve milk, vegetables, and fruits naturally. It’s eco-friendly, affordable, and sustainable, proving that sometimes the best innovations come from tradition + ingenuity. 🌱 From a potter’s workshop to a full clay appliance brand, Mansukhbhai’s journey is a masterclass in resilience, creativity, and impact. Innovation isn’t always high-tech. Sometimes, it’s just clay, water, and a dream.
1:27

🔥 Gujarat start-up - From Earthquake to Innovation: The Story of Mitticool 🔥 After losing everything in the 2001 Gujarat earthquake, Mansukhbhai Prajapati didn’t just rebuild his life, he reinvented refrigeration. Inspired by a newspaper headline about a “poor man’s fridge,” he created Mitticool: a clay refrigerator that keeps food fresh for days without a single watt of electricity. 🏺 Made from simple clay, Mitticool uses evaporative cooling to preserve milk, vegetables, and fruits naturally. It’s eco-friendly, affordable, and sustainable, proving that sometimes the best innovations come from tradition + ingenuity. 🌱 From a potter’s workshop to a full clay appliance brand, Mansukhbhai’s journey is a masterclass in resilience, creativity, and impact. Innovation isn’t always high-tech. Sometimes, it’s just clay, water, and a dream.

JIX5A

28,280 görüntüleme • 5 ay önce

Morgan Stanley just dropped numbers that should make every investor pay attention (Save this). Hyperscalers spent $261B in 2024, they're now projected to spend $1.4T in 2028, a 5x increase in four years and that doesn't even include OpenAI or Anthropic. For the first time, Morgan Stanley is classifying SpaceX as a legitimate hyperscaler alongside Google, Amazon, Microsoft, and Meta. All of it flows into chips, data centers and memory and the supply chain companies sitting beneath the hyperscalers are the ones that will compound quietly. The most underrated play is memory. Micron is the only American HBM supplier, giving it a structural edge in government AI contracts that Samsung and SK Hynix cannot touch. Its entire 2026 HBM4 production is already sold out, revenue nearly tripled to $23.9B, and the memory prices are roughly doubling every year. The cooling problem is one of the most profitable bottlenecks in this entire trade. Vertiv makes the power management, liquid cooling systems and racks that keep GPU clusters from melting and it's up over 100% year to date in 2026 with a $15B+ backlog and guidance raised to $13.5–$14B in full year revenue. Arista Networks (ANET) is the networking infrastructure play, every AI data center needs ultra high speed networking fabric to connect thousands of GPUs together And Arista just doubled its 2026 AI revenue target as the industry shifts from proprietary InfiniBand to Open Ethernet, a shift that plays directly into Arista's strengths. Astera Labs solves the interconnect bottleneck inside data centers, the problem of getting data between chips fast enough to keep up with the GPUs. Revenue grew 93% year over year, it's already profitable, and its customers are Microsoft and Amazon directly. The hyperscalers are the miners and the real money is in the companies selling them the shovels, the electricity, the memory, the cooling, the networking, and the custom silicon. Make sure to follow me Melvin for more underrated infrastructure plays.
3:58

Morgan Stanley just dropped numbers that should make every investor pay attention (Save this). Hyperscalers spent $261B in 2024, they're now projected to spend $1.4T in 2028, a 5x increase in four years and that doesn't even include OpenAI or Anthropic. For the first time, Morgan Stanley is classifying SpaceX as a legitimate hyperscaler alongside Google, Amazon, Microsoft, and Meta. All of it flows into chips, data centers and memory and the supply chain companies sitting beneath the hyperscalers are the ones that will compound quietly. The most underrated play is memory. Micron is the only American HBM supplier, giving it a structural edge in government AI contracts that Samsung and SK Hynix cannot touch. Its entire 2026 HBM4 production is already sold out, revenue nearly tripled to $23.9B, and the memory prices are roughly doubling every year. The cooling problem is one of the most profitable bottlenecks in this entire trade. Vertiv makes the power management, liquid cooling systems and racks that keep GPU clusters from melting and it's up over 100% year to date in 2026 with a $15B+ backlog and guidance raised to $13.5–$14B in full year revenue. Arista Networks (ANET) is the networking infrastructure play, every AI data center needs ultra high speed networking fabric to connect thousands of GPUs together And Arista just doubled its 2026 AI revenue target as the industry shifts from proprietary InfiniBand to Open Ethernet, a shift that plays directly into Arista's strengths. Astera Labs solves the interconnect bottleneck inside data centers, the problem of getting data between chips fast enough to keep up with the GPUs. Revenue grew 93% year over year, it's already profitable, and its customers are Microsoft and Amazon directly. The hyperscalers are the miners and the real money is in the companies selling them the shovels, the electricity, the memory, the cooling, the networking, and the custom silicon. Make sure to follow me Melvin for more underrated infrastructure plays.

Melvin

36,068 görüntüleme • 27 gün önce

Efficiency is the real bottleneck in AI. As AI adoption accelerates, data centers are becoming some of the most energy-hungry assets in the world. $NBIS' approach is clear: sustainability isn’t an afterthought, it’s a design principle. The company designs its servers, racks, and data centers in-house, allowing: - Higher thermal & power efficiency - Faster cluster deployment - Lower failure rates and easier maintenance - Longer hardware lifespan Essentially, more compute per unit of energy. In Finland, $NBIS operates hybrid cooling systems that combine outside air with direct liquid-to-chip cooling, using a closed-loop system with zero water withdrawal. Even more interesting: waste heat isn’t wasted. In 2024 alone, Nebius’ Finnish data center: - Helped heat ~2,500 households - Avoided ~3,200 tons of CO₂ emissions This is circular efficiency in practice. On the software side, $NBIS' full-stack cloud platform avoids overprovisioning by sharing buffers across hundreds of customers, while offering deep observability, down to tracking every kilowatt used to train large models across compute and storage. All in all, $NBIS is building AI infrastructure optimized for peak performance, cost efficiency, resilience, and long-term sustainability.
3:29

Efficiency is the real bottleneck in AI. As AI adoption accelerates, data centers are becoming some of the most energy-hungry assets in the world. $NBIS' approach is clear: sustainability isn’t an afterthought, it’s a design principle. The company designs its servers, racks, and data centers in-house, allowing: - Higher thermal & power efficiency - Faster cluster deployment - Lower failure rates and easier maintenance - Longer hardware lifespan Essentially, more compute per unit of energy. In Finland, $NBIS operates hybrid cooling systems that combine outside air with direct liquid-to-chip cooling, using a closed-loop system with zero water withdrawal. Even more interesting: waste heat isn’t wasted. In 2024 alone, Nebius’ Finnish data center: - Helped heat ~2,500 households - Avoided ~3,200 tons of CO₂ emissions This is circular efficiency in practice. On the software side, $NBIS' full-stack cloud platform avoids overprovisioning by sharing buffers across hundreds of customers, while offering deep observability, down to tracking every kilowatt used to train large models across compute and storage. All in all, $NBIS is building AI infrastructure optimized for peak performance, cost efficiency, resilience, and long-term sustainability.

M. V. Cunha

44,039 görüntüleme • 7 ay önce

REPORT: Microsoft’s Stargate AI data center just drained half a billion gallons of water from drought-stricken Abilene, Texas—while residents are being told to cut back on showers. And this is only the beginning. By 2030, AI data centers are projected to consume 7% of Texas’ entire water supply. As families are forced to conserve, Big Tech is torching billions of gallons using outdated cooling systems that waste every drop—and towns like Abilene are already paying the price. This isn’t just a Texas problem—it’s coming for every state. Doc Pete Chambers, former Green Beret and candidate for Governor, warns that Texas is barreling toward a manmade disaster. If nothing changes, the cattle industry could collapse, food prices could soar, and Texans may be forced to rely on imported beef riddled with disease. He’s calling for urgent action: coastal desalination, water recycling, and strict limits on where these facilities can be built. Are we really going to drain our land, kill our food supply, and surrender to Silicon Valley—just to keep AI online? Watch the full interview to hear what the current Governor of Texas refuses to admit.
13:10

REPORT: Microsoft’s Stargate AI data center just drained half a billion gallons of water from drought-stricken Abilene, Texas—while residents are being told to cut back on showers. And this is only the beginning. By 2030, AI data centers are projected to consume 7% of Texas’ entire water supply. As families are forced to conserve, Big Tech is torching billions of gallons using outdated cooling systems that waste every drop—and towns like Abilene are already paying the price. This isn’t just a Texas problem—it’s coming for every state. Doc Pete Chambers, former Green Beret and candidate for Governor, warns that Texas is barreling toward a manmade disaster. If nothing changes, the cattle industry could collapse, food prices could soar, and Texans may be forced to rely on imported beef riddled with disease. He’s calling for urgent action: coastal desalination, water recycling, and strict limits on where these facilities can be built. Are we really going to drain our land, kill our food supply, and surrender to Silicon Valley—just to keep AI online? Watch the full interview to hear what the current Governor of Texas refuses to admit.

Vigilant Fox 🦊

77,767 görüntüleme • 1 yıl önce

NEW: Premium Inference 101 The Economics & Infrastructure Behind Running Trillion Parameter Models Rodrigo Liang, CEO & Co-Founder of SambaNova "Inference has arrived. 70-80% of those racks are running inference." "[Inference services] are generating lots of revenue, but not enough margin. In order for them to sustain, they've gotta be more profitable." "With SambaNova, that min quantum is down to 1 rack. Where if you have other service providers, [with] say, a DeepSeek model, now 1.5 trillion parameters, to run that, the min for some of the other providers might be 10-20 racks." SambaNova builds full-stack inference infrastructure. 16 chips to a 10kW air-cooled rack that runs trillion parameter models, where a GPU rack pulls 130kW. They just demonstrated the fastest MiniMax M2.7 inference in the world, as benchmarked by Artificial Analysis. The demo paired one NVIDIA H200 rack for prefill with one SambaRack SN50 for decode. Disaggregated inference: GPUs load the context, RDUs generate the tokens. Now serving JPMorgan, SoftBank, Saudi Aramco & DOE national labs, just valued at $11B on a $1B Series F led by General Atlantic. We Cover: › Why inference will need orders of magnitude more chips than training ever did › The 10kW rack vs the 130kW rack, & why air cooling decides geography › Running a 1T parameter model in one rack at full precision, no quantization › The agent latency problem: 20 agents, 2 seconds each, 40 seconds gone › Revenue per rack, & why inference providers have revenue but no margin › JPMorgan, sovereignty, & the move back to on-prem Filmed at the RAISE Summit in Paris. Thank you to Brex, MongoDB & AssemblyAI for helping make this trip & content series happen. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Rodrigo Liang , Co-Founder & CEO at SambaNova Systems (00:59) SambaNova’s Series F: $1B raise at an $11 billion valuation (03:00) The Inference problem nobody saw coming (04:52) SambaNova's chip evolution (07:19) Running a trillion-parameter model on a single rack (11:00) Do $100 billion data centers actually make sense? (14:14) What "premium inference" really means (18:28) Speed is about to become AI's biggest price tag (20:43) Starlink, edge computing, & AI reaching every corner of the planet (24:27) Working alongside NVIDIA & rival chipmakers (27:49) How customers actually measure inference performance (32:07) The biggest bottlenecks in AI's global land grab (35:12) Justifying the billion-dollar AI valuations (37:53) Why SambaNova refuses to build its own cloud (41:03) The "AI sovereignty" debate (43:48) Data privacy fears are driving the return to on-prem AI (47:55) How to actually get ROI out of AI spend (51:16) The one question every business should be asking about AI (56:09) The mentors & lessons behind a 32-year career in chips (58:02) Unveiling SambaNova's newest chip, the SN50
59:56

NEW: Premium Inference 101 The Economics & Infrastructure Behind Running Trillion Parameter Models Rodrigo Liang, CEO & Co-Founder of SambaNova "Inference has arrived. 70-80% of those racks are running inference." "[Inference services] are generating lots of revenue, but not enough margin. In order for them to sustain, they've gotta be more profitable." "With SambaNova, that min quantum is down to 1 rack. Where if you have other service providers, [with] say, a DeepSeek model, now 1.5 trillion parameters, to run that, the min for some of the other providers might be 10-20 racks." SambaNova builds full-stack inference infrastructure. 16 chips to a 10kW air-cooled rack that runs trillion parameter models, where a GPU rack pulls 130kW. They just demonstrated the fastest MiniMax M2.7 inference in the world, as benchmarked by Artificial Analysis. The demo paired one NVIDIA H200 rack for prefill with one SambaRack SN50 for decode. Disaggregated inference: GPUs load the context, RDUs generate the tokens. Now serving JPMorgan, SoftBank, Saudi Aramco & DOE national labs, just valued at $11B on a $1B Series F led by General Atlantic. We Cover: › Why inference will need orders of magnitude more chips than training ever did › The 10kW rack vs the 130kW rack, & why air cooling decides geography › Running a 1T parameter model in one rack at full precision, no quantization › The agent latency problem: 20 agents, 2 seconds each, 40 seconds gone › Revenue per rack, & why inference providers have revenue but no margin › JPMorgan, sovereignty, & the move back to on-prem Filmed at the RAISE Summit in Paris. Thank you to Brex, MongoDB & AssemblyAI for helping make this trip & content series happen. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Rodrigo Liang , Co-Founder & CEO at SambaNova Systems (00:59) SambaNova’s Series F: $1B raise at an $11 billion valuation (03:00) The Inference problem nobody saw coming (04:52) SambaNova's chip evolution (07:19) Running a trillion-parameter model on a single rack (11:00) Do $100 billion data centers actually make sense? (14:14) What "premium inference" really means (18:28) Speed is about to become AI's biggest price tag (20:43) Starlink, edge computing, & AI reaching every corner of the planet (24:27) Working alongside NVIDIA & rival chipmakers (27:49) How customers actually measure inference performance (32:07) The biggest bottlenecks in AI's global land grab (35:12) Justifying the billion-dollar AI valuations (37:53) Why SambaNova refuses to build its own cloud (41:03) The "AI sovereignty" debate (43:48) Data privacy fears are driving the return to on-prem AI (47:55) How to actually get ROI out of AI spend (51:16) The one question every business should be asking about AI (56:09) The mentors & lessons behind a 32-year career in chips (58:02) Unveiling SambaNova's newest chip, the SN50

Molly O’Shea

134,960 görüntüleme • 24 gün önce

Etched came out of stealth at $800M and by lunch X had NVIDIA in the ground We do this every few months. A chip launches, the deck says killer, the timeline holds a funeral, and NVIDIA closes green anyway Etched hardwires the transformer into silicon. That is where the speed comes from, nearly the whole die on one job instead of the ~30% a GPU uses. It is also the trap. The day that chip tapes out is the best it will ever be. You cannot patch it. You burned progress into a wafer and now pray the field stops moving NVIDIA made the opposite bet. Same board, faster every quarter in software. Dynamo is pulling more tokens per watt out of the same rack, on version 1.0 The depreciation risk the bears aimed at NVIDIA for two years does not live at NVIDIA. It lives here, on the chip built to bury it Etched is not a fraud. It is a niche tool priced like a general one, and $800M is not enough to run a frontier supply chain. The rest get bought on the next down cycle Bury the lead, not the leader. Full case with Jack Farley and Max Wiethe on MTS And special thanks for Baseten for the cool T Shirt! Chapters 00:00 Switching from bonds to semis 00:33 What Etched actually is 01:31 Faster and cheaper, but how much HBM 02:56 Maturing market, not an NVIDIA killer 05:01 $1B in contracts and a Taiwan factory 05:20 Why these startups all get absorbed 06:48 Tiered inference and the obsolescence trap 09:39 Etched vs TPUs and Trainium 12:17 Is the CUDA moat weakening 13:21 Co-design, squeezing every token per watt 14:37 NVIDIA is a software company that sells a chip 14:58 Who is NVIDIA's most dangerous competitor 16:23 The NVIDIA killers, ranked 18:19 A rich man's game 18:43 AMD's MI500 vs Rubin Ultra 20:19 The neocloud business decision 22:46 Lightning round, Rambus the toll on HBM 25:11 The CXL run-up on Astera, Marvell, Credo 26:22 Use AI less, go to the booth 28:10 EDA is not dead
30:42

Etched came out of stealth at $800M and by lunch X had NVIDIA in the ground We do this every few months. A chip launches, the deck says killer, the timeline holds a funeral, and NVIDIA closes green anyway Etched hardwires the transformer into silicon. That is where the speed comes from, nearly the whole die on one job instead of the ~30% a GPU uses. It is also the trap. The day that chip tapes out is the best it will ever be. You cannot patch it. You burned progress into a wafer and now pray the field stops moving NVIDIA made the opposite bet. Same board, faster every quarter in software. Dynamo is pulling more tokens per watt out of the same rack, on version 1.0 The depreciation risk the bears aimed at NVIDIA for two years does not live at NVIDIA. It lives here, on the chip built to bury it Etched is not a fraud. It is a niche tool priced like a general one, and $800M is not enough to run a frontier supply chain. The rest get bought on the next down cycle Bury the lead, not the leader. Full case with Jack Farley and Max Wiethe on MTS And special thanks for Baseten for the cool T Shirt! Chapters 00:00 Switching from bonds to semis 00:33 What Etched actually is 01:31 Faster and cheaper, but how much HBM 02:56 Maturing market, not an NVIDIA killer 05:01 $1B in contracts and a Taiwan factory 05:20 Why these startups all get absorbed 06:48 Tiered inference and the obsolescence trap 09:39 Etched vs TPUs and Trainium 12:17 Is the CUDA moat weakening 13:21 Co-design, squeezing every token per watt 14:37 NVIDIA is a software company that sells a chip 14:58 Who is NVIDIA's most dangerous competitor 16:23 The NVIDIA killers, ranked 18:19 A rich man's game 18:43 AMD's MI500 vs Rubin Ultra 20:19 The neocloud business decision 22:46 Lightning round, Rambus the toll on HBM 25:11 The CXL run-up on Astera, Marvell, Credo 26:22 Use AI less, go to the booth 28:10 EDA is not dead

Ben Pouladian

29,969 görüntüleme • 1 ay önce

We sat down with Philip Johnston, co-founder and CEO of Starcloud, at MIT to discuss why the future of data centers might be in space. After graduating Y Combinator less than 2 years ago, Starcloud just raised an impressive $170M Series A at a $1.1B valuation led by Benchmark and EQT Ventures & Growth. The conversation covers everything from solar physics and cooling systems to GPU economics, radiation hardening, launch costs, and satellite design. Philip also shares what it takes to build a unicorn deeptech startup. We discuss his experience with YC, the skepticism around their demoday launch, and the crazy last minute race to get Starcloud’s first satellite onboard their scheduled Falcon flight. Full episode is here on X and at any of the links below (see comment). Timestamps: 00:00 - Intro 01:12 - What is Starcloud? 02:44 - Why do data centers need to go to space? 06:15 - Can’t we just build more solar panels on earth? 11:10 - Economic analysis of Starcloud 19:56 - How does Starcloud’s cooling work? 28:26 - Training an LLM in space 32:07 - Addressing critics on space Twitter 34:23 - Is Starcloud overfunded? 35:59 - Will demand for data centers keep going up? 38:11 - GPU lifespan and disposal in space 39:47 - Bus structures 41:43 - Starcloud’s origin and founders 49:29 - Fundraising, Competition, and Meeting Expectations 53:29 - Satellite size and collisions 56:29 - Manufacturing Bottlenecks 1:00:20 - Starcloud 1 tests 1:01:57 - Acceleration after YC 1:03:43 - Testing on Earth 1:05:06 - Motivations for Starcloud 1:06:45 - Data centers on the Moon 1:08:12 - Interacting with AI companies 1:08:18 - What’s next for Starcloud? 1:14:01 - Other uses for Starcloud satellites 1:17:56 - Lunar hotels and space elevators 1:24:28 - Complementary business ideas to Starcloud 1:29:51 - Philip’s competitive twin 1:32:18 - Philip and Mike’s thoughts on YC 1:36:04 - Advice for young entrepreneurs Elon Musk Scott Manley Kyle Hill Hank Green
1:37:33

We sat down with Philip Johnston, co-founder and CEO of Starcloud, at MIT to discuss why the future of data centers might be in space. After graduating Y Combinator less than 2 years ago, Starcloud just raised an impressive $170M Series A at a $1.1B valuation led by Benchmark and EQT Ventures & Growth. The conversation covers everything from solar physics and cooling systems to GPU economics, radiation hardening, launch costs, and satellite design. Philip also shares what it takes to build a unicorn deeptech startup. We discuss his experience with YC, the skepticism around their demoday launch, and the crazy last minute race to get Starcloud’s first satellite onboard their scheduled Falcon flight. Full episode is here on X and at any of the links below (see comment). Timestamps: 00:00 - Intro 01:12 - What is Starcloud? 02:44 - Why do data centers need to go to space? 06:15 - Can’t we just build more solar panels on earth? 11:10 - Economic analysis of Starcloud 19:56 - How does Starcloud’s cooling work? 28:26 - Training an LLM in space 32:07 - Addressing critics on space Twitter 34:23 - Is Starcloud overfunded? 35:59 - Will demand for data centers keep going up? 38:11 - GPU lifespan and disposal in space 39:47 - Bus structures 41:43 - Starcloud’s origin and founders 49:29 - Fundraising, Competition, and Meeting Expectations 53:29 - Satellite size and collisions 56:29 - Manufacturing Bottlenecks 1:00:20 - Starcloud 1 tests 1:01:57 - Acceleration after YC 1:03:43 - Testing on Earth 1:05:06 - Motivations for Starcloud 1:06:45 - Data centers on the Moon 1:08:12 - Interacting with AI companies 1:08:18 - What’s next for Starcloud? 1:14:01 - Other uses for Starcloud satellites 1:17:56 - Lunar hotels and space elevators 1:24:28 - Complementary business ideas to Starcloud 1:29:51 - Philip’s competitive twin 1:32:18 - Philip and Mike’s thoughts on YC 1:36:04 - Advice for young entrepreneurs Elon Musk Scott Manley Kyle Hill Hank Green

632nm

46,599 görüntüleme • 4 ay önce

🚨 I DON'T KNOW WHAT PEOPLE ARE WAITING FOR! This guy just deployed a full GPU compute farm in a data center Not renting cloud instances, not paying per-token, just racks of consumer GPUs on warehouse shelving doing AI inference 24/7 The setup is wild: open-frame rigs with rows of white fans, PCIe riser cables, server-grade PSUs, and a PDU for power distribution. All mounted on red industrial shelving with raised-floor cooling underneath. It looks like a crypto mining op, because it probably was one Here's what's actually happening: > Ex-mining rigs are being redeployed for AI inference at scale > Each rig runs 6-8 GPUs on PCIe risers connected to a single motherboard > Fans pull cold air from the raised floor through the open frames > A PDU on each rack handles power distribution across all rigs > The whole thing runs in a colocation facility with cheap power rates The economics make sense: cloud GPU pricing runs $2-4/hr per card. Own the hardware, colocate it for $0.08-0.12/kWh, and you're printing compute at a fraction of the cost The crypto-to-AI pipeline is real. Same hardware, same facilities, same power contracts, completely different revenue model Bookmark this!
0:53

🚨 I DON'T KNOW WHAT PEOPLE ARE WAITING FOR! This guy just deployed a full GPU compute farm in a data center Not renting cloud instances, not paying per-token, just racks of consumer GPUs on warehouse shelving doing AI inference 24/7 The setup is wild: open-frame rigs with rows of white fans, PCIe riser cables, server-grade PSUs, and a PDU for power distribution. All mounted on red industrial shelving with raised-floor cooling underneath. It looks like a crypto mining op, because it probably was one Here's what's actually happening: > Ex-mining rigs are being redeployed for AI inference at scale > Each rig runs 6-8 GPUs on PCIe risers connected to a single motherboard > Fans pull cold air from the raised floor through the open frames > A PDU on each rack handles power distribution across all rigs > The whole thing runs in a colocation facility with cheap power rates The economics make sense: cloud GPU pricing runs $2-4/hr per card. Own the hardware, colocate it for $0.08-0.12/kWh, and you're printing compute at a fraction of the cost The crypto-to-AI pipeline is real. Same hardware, same facilities, same power contracts, completely different revenue model Bookmark this!

cristal

16,305 görüntüleme • 1 ay önce

$MU $SNDK $LITE $VRT NVIDIA and Groq: 2nd and 3rd Order Strategic Infrastructure Effects and Market Implications Public reporting indicates NVIDIA has agreed to acquire Groq for approximately $20,000,000,000 in cash, while excluding Groq’s nascent cloud business from the transaction perimeter. The reported carve-out materially constrains the immediate, direct linkage from the acquisition to incremental, NVIDIA-controlled data center capacity build-out because GroqCloud appears to be the principal channel through which Groq hardware is currently monetized at scale as a service. The infrastructure-market implications therefore depend primarily on post-close product strategy: whether NVIDIA (1) commercializes Groq silicon as a distinct inference product line and drives broad deployment through OEM/ODM channels and partners, (2) uses the acquisition mainly to absorb IP and talent while de-emphasizing standalone Groq hardware volumes, or (3) uses Groq technology to reshape NVIDIA’s own inference systems and networking roadmaps. The dominant transmission mechanism into memory, networking, and facility infrastructure markets is the degree to which NVIDIA shifts incremental inference deployments away from GPU architectures that are tightly coupled to external high-bandwidth memory (HBM) and toward Groq’s current architecture, which emphasizes large on-chip SRAM, deterministic compiler-scheduled execution, and direct chip-to-chip connectivity. Independent and company-published materials describe Groq’s current-generation approach as having no external memory, keeping weights and KV cache on-chip during processing, and requiring model sharding across multiple chips due to limited on-chip SRAM per device. That architectural choice is directionally HBM-negative on a per-accelerator basis and ambiguous for DRAM, NAND, networking, power, and cooling on a per-token basis because the design can reduce memory wall losses and tail-latency overhead while potentially increasing the number of chips and interconnect endpoints required to serve large models and long-context workloads. HBM implications are the most mechanically straightforward but should be framed as second-derivative rather than absolute. If Groq-class inference silicon meaningfully displaces NVIDIA GPU-based inference deployments, incremental HBM bit demand tied to inference growth could be reduced relative to a GPU-only baseline because Groq’s current approach does not appear to attach HBM stacks to each accelerator. However, current market structure suggests HBM remains supply-constrained and is being pulled by multiple vectors including continued GPU training scale and high-capacity inference configurations, with leading suppliers signaling tight conditions extending beyond 2026. In that environment, reduced inference-driven HBM intensity could primarily reallocate scarce HBM supply toward higher-end training and premium inference GPUs rather than creating an outright volume collapse, preserving high utilization of HBM capacity while potentially affecting the slope of pricing power and capacity expansion urgency over a multi-year horizon. The key downside scenario for the HBM complex would be a durable architectural bifurcation where “good-enough” inference shifts disproportionately to HBM-less ASICs across a broad swath of deployments (latency-sensitive, batch-1, cost-per-token optimized), while training remains GPU-HBM dominated; such a split would reduce the portion of future inference compute that naturally monetizes through HBM content and could compress the incremental HBM-per-AI-dollar ratio. The key upside/neutral scenario for HBM is that the supply chain remains fully allocated regardless, with NVIDIA using any “freed” HBM to ship more high-end GPUs into training and long-context inference, especially as roadmaps increase HBM per GPU, sustaining robust aggregate bit demand even if inference becomes more heterogeneous. Conventional DRAM implications split into 2 channels: (1) DRAM wafer capacity diversion into HBM and (2) DDR content per server in AI clusters. Supplier commentary indicates that AI-driven memory demand is supporting elevated DRAM markets more broadly, and HBM production is resource-intensive versus conventional DRAM, tightening supply for DDR products in parallel. A meaningful NVIDIA pivot to an inference architecture that reduces HBM dependence could, at the margin, ease the most acute HBM-driven bottlenecks and allow memory manufacturers more flexibility in balancing DRAM mix, which could be modestly DDR-positive on the supply side (less crowding-out) even if it is DDR-neutral or slightly negative on the demand side (if per-node CPU/DDR requirements decline due to more efficient accelerator utilization). The dominant practical outcome is likely that DDR demand remains supported by broad AI server proliferation and increasing memory footprints at the system level (CPUs, networking stacks, caching layers, retrieval-augmented pipelines), while HBM remains the premium profit pool; therefore, any HBM displacement that increases total server volumes could indirectly keep DDR demand resilient even if DDR per accelerator is not rising materially. NAND flash implications are comparatively indirect and volume-driven rather than architecture-driven. Inference clusters require SSD capacity for model storage, container images, logging, and increasingly for fast local retrieval indices and embedding stores, but the storage footprint per unit of compute is typically smaller than in training pipelines that stage large datasets and checkpoints. If NVIDIA uses Groq to lower inference cost and latency enough to expand the total number of inference deployment locations (regional colocation, enterprise on-prem, sovereign footprints), aggregate SSD attach could rise through geographic fragmentation and replication of model artifacts across more sites, even if per-site storage is modest. The NAND effect is therefore likely to be demand-broadening and mix-positive (datacenter SSDs) but not a primary swing factor versus the macro AI capex cycle and consumer/device cycles. Hard disk drive (HDD) markets should see negligible direct sensitivity because nearline HDD demand is driven by bulk storage and cloud archiving economics, while inference acceleration choices primarily reshape compute and network layers; any HDD benefit would be a tertiary function of overall data center square footage expansion rather than a direct consequence of Groq silicon displacing GPUs. Optical networking implications require separating (1) intra-cluster back-end fabrics that connect accelerators and (2) front-end / data center interconnect (DCI) that connects sites and regions. Groq’s own positioning and third-party reporting suggest scaling beyond a single node or rack relies on high-bandwidth fabrics and, in some described configurations, optical interconnect scaling across hundreds of chips. If NVIDIA commercializes Groq at scale, 2 offsetting forces emerge: lower cost-per-token and improved latency could expand inference throughput and drive more east-west traffic, increasing demand for high-speed switching and optics; conversely, if Groq delivers materially higher utilization and tokens per unit of network bandwidth for certain workloads, the network required per served token could decline. Public NVIDIA materials already indicate an aggressive photonics roadmap aimed at scaling AI factories, including co-packaged optics (CPO) switches and explicit collaboration with Coherent and Lumentum in the silicon photonics supply chain. That linkage is important because it suggests that, independent of Groq, NVIDIA is already pushing optics integration deeper into the switch package to reduce power and increase resiliency; Groq increases the strategic incentive to reduce network power and latency if inference becomes even more distributed and latency-sensitive. For Lumentum and Coherent specifically, the net implication is less about “more optics versus fewer optics” and more about a shift in optics form factor and value capture. Co-packaged optics can reduce reliance on pluggable transceivers in some switch architectures while increasing demand for integrated photonic engines, lasers, fiber attach, packaging processes, and component-level supply. NVIDIA’s own announcements explicitly position Coherent and Lumentum as collaborators in creating the integrated silicon/optics process and supply chain for photonics switches. If Groq accelerates the transition to very large-scale fabrics (more endpoints, higher port speeds, tighter power envelopes), that tends to pull forward CPO adoption and amplifies demand for the underlying photonics components even if the conventional pluggable module TAM is structurally pressured over time. If Groq instead pushes inference toward smaller, more localized pods (closer to users, more regional colocation), that can be optics-positive for DCI and metro connectivity because more sites must be interconnected at high bandwidth with low latency, favoring coherent optics and high-speed interconnect between facilities. The principal risk for optics suppliers is timing and margin structure: a faster move to NVIDIA-driven integrated photonics could concentrate bargaining power and compress margins for commoditized transceiver modules while favoring suppliers with differentiated lasers, integration capability, and qualification depth in NVIDIA’s CPO ecosystem. AEC and copper interconnect implications hinge on whether Groq deployment increases the density of short-reach links inside racks and rows. High-speed copper remains structurally advantaged at very short distances on cost, power, and serviceability, but reaches become constrained as lane speeds and aggregate bandwidth rise, creating a role for active electrical cables (AECs), retimers, and signal-conditioning silicon. Credo explicitly positions its AEC products as enabling reliable lossless 800G connectivity for AI clusters, and the company has highlighted participation at NVIDIA GTC with content focused on extending PCIe/CXL using AECs, indicating relevance to next-generation system topologies that require longer reach and higher signal integrity than passive copper can deliver. If NVIDIA turns Groq into a widely deployed inference card or chassis product, the likely near-term effect is AEC-positive because (1) more inference throughput tends to increase top-of-rack connectivity requirements, (2) distributing inference across more racks and sites increases short-reach links per unit of delivered service, and (3) PCIe-attached accelerator architectures tend to require robust signal conditioning as systems move to PCIe 6.x and beyond. Groq workshop materials explicitly reference GroqCard and GroqNode form factors, reinforcing that PCIe-attached deployment has been central to Groq’s current packaging strategy. The main countervailing risk is that Groq’s deterministic chip-to-chip fabric could be implemented primarily through backplanes and direct board-level connectivity that reduces the need for merchant AECs inside the box; in that case, incremental AEC demand would concentrate more in rack-to-switch and node-to-fabric links rather than within-chassis chip fabrics. Astera Labs implications are connectivity-architecture sensitive and, on balance, skew positive if NVIDIA increases heterogeneity and disaggregation in AI systems. NVIDIA has publicly positioned NVLink Fusion as a pathway for partners to build semi-custom AI infrastructure and has explicitly identified Astera Labs as a partner in that ecosystem, with Astera describing NVLink-related solutions expanding its connectivity platform across PCIe, CXL, and Ethernet plus fleet observability software. A Groq acquisition increases the probability that NVIDIA offers a broader menu of accelerators (training GPUs, inference-focused ASICs) and therefore increases the importance of scalable, high-reliability connectivity, retiming, switching, and telemetry across mixed topologies. If Groq silicon remains PCIe-attached in many deployments, PCIe 6.x retimers/switches and active cable modules become more central, aligning with Astera’s core portfolio. If NVIDIA instead integrates Groq concepts into scale-up fabrics (NVLink-like domains) or uses Groq to expand into inference “appliances” that must be rapidly deployed in colocation environments, the need for standard-compliant, serviceable connectivity with strong RAS/telemetry increases, again aligning with Astera’s positioning. Power equipment and cooling implications for Vertiv and adjacent suppliers should be viewed through the lens of rack power density, cooling modality (air vs liquid), and site deployment model (hyperscale campuses vs distributed colocation/enterprise). Groq claims its LPU and rack designs are “air-cooled by design” and require no complex cooling and power infrastructure, and third-party reporting has described Groq’s approach as relying on parallelism across many lower-power units rather than extreme per-chip performance. If NVIDIA scales Groq as a mainstream inference platform, the mix of data center cooling spend could shift modestly away from the highest-density liquid-cooled racks toward more air-cooled or hybrid deployments, particularly for inference pods placed in existing facilities that cannot easily retrofit for very high rack heat flux. That would be a mix headwind for suppliers most levered exclusively to high-end liquid cooling attachments per rack, but it is not necessarily a volume headwind for Vertiv given the company’s broad exposure to both power and cooling infrastructure and the likelihood that total AI deployment locations expand. Vertiv’s own industry commentary emphasizes that AI racks require higher power-density UPS, batteries, power distribution equipment, and switchgear capable of handling rapid load transients, and that hybrid cooling systems will evolve across deployment environments. Those statements align with a world where inference growth increases the count of powered racks and raises the operational complexity of power delivery even if per-rack density is lower than the most extreme training clusters. The most material infrastructure impact may occur outside the rack and upstream of the data hall: grid interconnects, substations, transformers, switchgear, generators, and utility-scale generation additions. Recent regulatory actions in the U.S. highlight that projected data center demand is already driving large planned increases in electricity generation capacity, underscoring that power availability is a binding constraint. In that context, an inference architecture that lowers joules per token could reduce the power required per unit of inference delivered, but it can also accelerate demand by lowering cost and improving latency, increasing the total volume of inference served (a classic rebound effect). The net outcome is likely continued, elevated demand for power infrastructure even if efficiency improves, with the key swing factor being whether AI capex remains on a multi-year growth trajectory or enters a digestion phase. Other data center infrastructure implications include server/ODM mix, facility design standardization, and networking architecture choices. If NVIDIA positions Groq-based inference as a broadly distributable “standard server + accelerator” solution rather than as an integrated, liquid-cooled rack like GB200 NVL72, spend could shift toward more conventional air-cooled server designs, higher unit volumes of mainstream racks, and faster deployment in colocation footprints, increasing demand for modular power rooms, busways, and rapidly deployable cooling solutions. If NVIDIA instead integrates Groq into its “AI factory” paradigm, the primary effect is likely acceleration of dense back-end fabric build-outs and a faster push toward photonics switching, increasing demand for fiber plant, connectors, and integrated optics supply chains while potentially compressing the lifecycle of transitional architectures based on pluggable optics and mid-reach copper. NVIDIA’s stated roadmap toward co-packaged optics and silicon photonics switches is already oriented toward scaling to very large GPU counts; adding a high-end inference ASIC increases the strategic importance of power-efficient, low-latency fabrics because inference economics become increasingly sensitive to network overhead as compute cost declines. Across the covered segments, the most defensible base case is limited near-term dislocation and a medium-term increase in uncertainty around memory intensity per unit of inference growth. HBM faces the clearest relative risk from an HBM-less inference platform, but supply tightness and GPU training roadmaps reduce the probability of an absolute demand shock over the next 12–24 months. Optical, AEC/copper, and power/cooling are more likely to remain volume-supported because they scale with endpoint count, deployment fragmentation, and total data center footprint, and those tend to rise when inference becomes cheaper and more widely deployed. The highest-conviction second-order effect is a shift in infrastructure mix: incrementally more distributed inference deployments (favoring colocation power/cooling standardization, DCI optics, and serviceable short-reach interconnect) and a gradual migration from pluggable optics toward integrated photonics in back-end fabrics (favoring suppliers positioned in the CPO ecosystem).
9:00

$MU $SNDK $LITE $VRT NVIDIA and Groq: 2nd and 3rd Order Strategic Infrastructure Effects and Market Implications Public reporting indicates NVIDIA has agreed to acquire Groq for approximately $20,000,000,000 in cash, while excluding Groq’s nascent cloud business from the transaction perimeter. The reported carve-out materially constrains the immediate, direct linkage from the acquisition to incremental, NVIDIA-controlled data center capacity build-out because GroqCloud appears to be the principal channel through which Groq hardware is currently monetized at scale as a service. The infrastructure-market implications therefore depend primarily on post-close product strategy: whether NVIDIA (1) commercializes Groq silicon as a distinct inference product line and drives broad deployment through OEM/ODM channels and partners, (2) uses the acquisition mainly to absorb IP and talent while de-emphasizing standalone Groq hardware volumes, or (3) uses Groq technology to reshape NVIDIA’s own inference systems and networking roadmaps. The dominant transmission mechanism into memory, networking, and facility infrastructure markets is the degree to which NVIDIA shifts incremental inference deployments away from GPU architectures that are tightly coupled to external high-bandwidth memory (HBM) and toward Groq’s current architecture, which emphasizes large on-chip SRAM, deterministic compiler-scheduled execution, and direct chip-to-chip connectivity. Independent and company-published materials describe Groq’s current-generation approach as having no external memory, keeping weights and KV cache on-chip during processing, and requiring model sharding across multiple chips due to limited on-chip SRAM per device. That architectural choice is directionally HBM-negative on a per-accelerator basis and ambiguous for DRAM, NAND, networking, power, and cooling on a per-token basis because the design can reduce memory wall losses and tail-latency overhead while potentially increasing the number of chips and interconnect endpoints required to serve large models and long-context workloads. HBM implications are the most mechanically straightforward but should be framed as second-derivative rather than absolute. If Groq-class inference silicon meaningfully displaces NVIDIA GPU-based inference deployments, incremental HBM bit demand tied to inference growth could be reduced relative to a GPU-only baseline because Groq’s current approach does not appear to attach HBM stacks to each accelerator. However, current market structure suggests HBM remains supply-constrained and is being pulled by multiple vectors including continued GPU training scale and high-capacity inference configurations, with leading suppliers signaling tight conditions extending beyond 2026. In that environment, reduced inference-driven HBM intensity could primarily reallocate scarce HBM supply toward higher-end training and premium inference GPUs rather than creating an outright volume collapse, preserving high utilization of HBM capacity while potentially affecting the slope of pricing power and capacity expansion urgency over a multi-year horizon. The key downside scenario for the HBM complex would be a durable architectural bifurcation where “good-enough” inference shifts disproportionately to HBM-less ASICs across a broad swath of deployments (latency-sensitive, batch-1, cost-per-token optimized), while training remains GPU-HBM dominated; such a split would reduce the portion of future inference compute that naturally monetizes through HBM content and could compress the incremental HBM-per-AI-dollar ratio. The key upside/neutral scenario for HBM is that the supply chain remains fully allocated regardless, with NVIDIA using any “freed” HBM to ship more high-end GPUs into training and long-context inference, especially as roadmaps increase HBM per GPU, sustaining robust aggregate bit demand even if inference becomes more heterogeneous. Conventional DRAM implications split into 2 channels: (1) DRAM wafer capacity diversion into HBM and (2) DDR content per server in AI clusters. Supplier commentary indicates that AI-driven memory demand is supporting elevated DRAM markets more broadly, and HBM production is resource-intensive versus conventional DRAM, tightening supply for DDR products in parallel. A meaningful NVIDIA pivot to an inference architecture that reduces HBM dependence could, at the margin, ease the most acute HBM-driven bottlenecks and allow memory manufacturers more flexibility in balancing DRAM mix, which could be modestly DDR-positive on the supply side (less crowding-out) even if it is DDR-neutral or slightly negative on the demand side (if per-node CPU/DDR requirements decline due to more efficient accelerator utilization). The dominant practical outcome is likely that DDR demand remains supported by broad AI server proliferation and increasing memory footprints at the system level (CPUs, networking stacks, caching layers, retrieval-augmented pipelines), while HBM remains the premium profit pool; therefore, any HBM displacement that increases total server volumes could indirectly keep DDR demand resilient even if DDR per accelerator is not rising materially. NAND flash implications are comparatively indirect and volume-driven rather than architecture-driven. Inference clusters require SSD capacity for model storage, container images, logging, and increasingly for fast local retrieval indices and embedding stores, but the storage footprint per unit of compute is typically smaller than in training pipelines that stage large datasets and checkpoints. If NVIDIA uses Groq to lower inference cost and latency enough to expand the total number of inference deployment locations (regional colocation, enterprise on-prem, sovereign footprints), aggregate SSD attach could rise through geographic fragmentation and replication of model artifacts across more sites, even if per-site storage is modest. The NAND effect is therefore likely to be demand-broadening and mix-positive (datacenter SSDs) but not a primary swing factor versus the macro AI capex cycle and consumer/device cycles. Hard disk drive (HDD) markets should see negligible direct sensitivity because nearline HDD demand is driven by bulk storage and cloud archiving economics, while inference acceleration choices primarily reshape compute and network layers; any HDD benefit would be a tertiary function of overall data center square footage expansion rather than a direct consequence of Groq silicon displacing GPUs. Optical networking implications require separating (1) intra-cluster back-end fabrics that connect accelerators and (2) front-end / data center interconnect (DCI) that connects sites and regions. Groq’s own positioning and third-party reporting suggest scaling beyond a single node or rack relies on high-bandwidth fabrics and, in some described configurations, optical interconnect scaling across hundreds of chips. If NVIDIA commercializes Groq at scale, 2 offsetting forces emerge: lower cost-per-token and improved latency could expand inference throughput and drive more east-west traffic, increasing demand for high-speed switching and optics; conversely, if Groq delivers materially higher utilization and tokens per unit of network bandwidth for certain workloads, the network required per served token could decline. Public NVIDIA materials already indicate an aggressive photonics roadmap aimed at scaling AI factories, including co-packaged optics (CPO) switches and explicit collaboration with Coherent and Lumentum in the silicon photonics supply chain. That linkage is important because it suggests that, independent of Groq, NVIDIA is already pushing optics integration deeper into the switch package to reduce power and increase resiliency; Groq increases the strategic incentive to reduce network power and latency if inference becomes even more distributed and latency-sensitive. For Lumentum and Coherent specifically, the net implication is less about “more optics versus fewer optics” and more about a shift in optics form factor and value capture. Co-packaged optics can reduce reliance on pluggable transceivers in some switch architectures while increasing demand for integrated photonic engines, lasers, fiber attach, packaging processes, and component-level supply. NVIDIA’s own announcements explicitly position Coherent and Lumentum as collaborators in creating the integrated silicon/optics process and supply chain for photonics switches. If Groq accelerates the transition to very large-scale fabrics (more endpoints, higher port speeds, tighter power envelopes), that tends to pull forward CPO adoption and amplifies demand for the underlying photonics components even if the conventional pluggable module TAM is structurally pressured over time. If Groq instead pushes inference toward smaller, more localized pods (closer to users, more regional colocation), that can be optics-positive for DCI and metro connectivity because more sites must be interconnected at high bandwidth with low latency, favoring coherent optics and high-speed interconnect between facilities. The principal risk for optics suppliers is timing and margin structure: a faster move to NVIDIA-driven integrated photonics could concentrate bargaining power and compress margins for commoditized transceiver modules while favoring suppliers with differentiated lasers, integration capability, and qualification depth in NVIDIA’s CPO ecosystem. AEC and copper interconnect implications hinge on whether Groq deployment increases the density of short-reach links inside racks and rows. High-speed copper remains structurally advantaged at very short distances on cost, power, and serviceability, but reaches become constrained as lane speeds and aggregate bandwidth rise, creating a role for active electrical cables (AECs), retimers, and signal-conditioning silicon. Credo explicitly positions its AEC products as enabling reliable lossless 800G connectivity for AI clusters, and the company has highlighted participation at NVIDIA GTC with content focused on extending PCIe/CXL using AECs, indicating relevance to next-generation system topologies that require longer reach and higher signal integrity than passive copper can deliver. If NVIDIA turns Groq into a widely deployed inference card or chassis product, the likely near-term effect is AEC-positive because (1) more inference throughput tends to increase top-of-rack connectivity requirements, (2) distributing inference across more racks and sites increases short-reach links per unit of delivered service, and (3) PCIe-attached accelerator architectures tend to require robust signal conditioning as systems move to PCIe 6.x and beyond. Groq workshop materials explicitly reference GroqCard and GroqNode form factors, reinforcing that PCIe-attached deployment has been central to Groq’s current packaging strategy. The main countervailing risk is that Groq’s deterministic chip-to-chip fabric could be implemented primarily through backplanes and direct board-level connectivity that reduces the need for merchant AECs inside the box; in that case, incremental AEC demand would concentrate more in rack-to-switch and node-to-fabric links rather than within-chassis chip fabrics. Astera Labs implications are connectivity-architecture sensitive and, on balance, skew positive if NVIDIA increases heterogeneity and disaggregation in AI systems. NVIDIA has publicly positioned NVLink Fusion as a pathway for partners to build semi-custom AI infrastructure and has explicitly identified Astera Labs as a partner in that ecosystem, with Astera describing NVLink-related solutions expanding its connectivity platform across PCIe, CXL, and Ethernet plus fleet observability software. A Groq acquisition increases the probability that NVIDIA offers a broader menu of accelerators (training GPUs, inference-focused ASICs) and therefore increases the importance of scalable, high-reliability connectivity, retiming, switching, and telemetry across mixed topologies. If Groq silicon remains PCIe-attached in many deployments, PCIe 6.x retimers/switches and active cable modules become more central, aligning with Astera’s core portfolio. If NVIDIA instead integrates Groq concepts into scale-up fabrics (NVLink-like domains) or uses Groq to expand into inference “appliances” that must be rapidly deployed in colocation environments, the need for standard-compliant, serviceable connectivity with strong RAS/telemetry increases, again aligning with Astera’s positioning. Power equipment and cooling implications for Vertiv and adjacent suppliers should be viewed through the lens of rack power density, cooling modality (air vs liquid), and site deployment model (hyperscale campuses vs distributed colocation/enterprise). Groq claims its LPU and rack designs are “air-cooled by design” and require no complex cooling and power infrastructure, and third-party reporting has described Groq’s approach as relying on parallelism across many lower-power units rather than extreme per-chip performance. If NVIDIA scales Groq as a mainstream inference platform, the mix of data center cooling spend could shift modestly away from the highest-density liquid-cooled racks toward more air-cooled or hybrid deployments, particularly for inference pods placed in existing facilities that cannot easily retrofit for very high rack heat flux. That would be a mix headwind for suppliers most levered exclusively to high-end liquid cooling attachments per rack, but it is not necessarily a volume headwind for Vertiv given the company’s broad exposure to both power and cooling infrastructure and the likelihood that total AI deployment locations expand. Vertiv’s own industry commentary emphasizes that AI racks require higher power-density UPS, batteries, power distribution equipment, and switchgear capable of handling rapid load transients, and that hybrid cooling systems will evolve across deployment environments. Those statements align with a world where inference growth increases the count of powered racks and raises the operational complexity of power delivery even if per-rack density is lower than the most extreme training clusters. The most material infrastructure impact may occur outside the rack and upstream of the data hall: grid interconnects, substations, transformers, switchgear, generators, and utility-scale generation additions. Recent regulatory actions in the U.S. highlight that projected data center demand is already driving large planned increases in electricity generation capacity, underscoring that power availability is a binding constraint. In that context, an inference architecture that lowers joules per token could reduce the power required per unit of inference delivered, but it can also accelerate demand by lowering cost and improving latency, increasing the total volume of inference served (a classic rebound effect). The net outcome is likely continued, elevated demand for power infrastructure even if efficiency improves, with the key swing factor being whether AI capex remains on a multi-year growth trajectory or enters a digestion phase. Other data center infrastructure implications include server/ODM mix, facility design standardization, and networking architecture choices. If NVIDIA positions Groq-based inference as a broadly distributable “standard server + accelerator” solution rather than as an integrated, liquid-cooled rack like GB200 NVL72, spend could shift toward more conventional air-cooled server designs, higher unit volumes of mainstream racks, and faster deployment in colocation footprints, increasing demand for modular power rooms, busways, and rapidly deployable cooling solutions. If NVIDIA instead integrates Groq into its “AI factory” paradigm, the primary effect is likely acceleration of dense back-end fabric build-outs and a faster push toward photonics switching, increasing demand for fiber plant, connectors, and integrated optics supply chains while potentially compressing the lifecycle of transitional architectures based on pluggable optics and mid-reach copper. NVIDIA’s stated roadmap toward co-packaged optics and silicon photonics switches is already oriented toward scaling to very large GPU counts; adding a high-end inference ASIC increases the strategic importance of power-efficient, low-latency fabrics because inference economics become increasingly sensitive to network overhead as compute cost declines. Across the covered segments, the most defensible base case is limited near-term dislocation and a medium-term increase in uncertainty around memory intensity per unit of inference growth. HBM faces the clearest relative risk from an HBM-less inference platform, but supply tightness and GPU training roadmaps reduce the probability of an absolute demand shock over the next 12–24 months. Optical, AEC/copper, and power/cooling are more likely to remain volume-supported because they scale with endpoint count, deployment fragmentation, and total data center footprint, and those tend to rise when inference becomes cheaper and more widely deployed. The highest-conviction second-order effect is a shift in infrastructure mix: incrementally more distributed inference deployments (favoring colocation power/cooling standardization, DCI optics, and serviceable short-reach interconnect) and a gradual migration from pluggable optics toward integrated photonics in back-end fabrics (favoring suppliers positioned in the CPO ecosystem).

TheValueist

76,179 görüntüleme • 7 ay önce

WAREHOUSES THE SIZE OF A WALMART WITH 4,000 GPUS WERE EARNING $320K A MONTH MINING ETH, THE SAME RACKS NOW EARN $4M A MONTH RENTING TO AI STARTUPS ON VAST AI the warehouse in the clip is one of hundreds across the US, the operator at 6'6" tall for scale, electricity contracts and cooling already in place ethereum's merge in september 2022 killed gpu mining for everyone without a power plant, by 2025 alt coin mining was barely covering the electricity bill the same racks now host RTX 3090s and 4090s on vast ai, indie AI developers priced out of AWS rent them by the hour for inference and image generation a 4090 nets $500 to $1,000 a month renting to AI startups, the same card was clearing $80 to $150 mining, 4 to 7 times the revenue with zero new hardware core scientific, hut 8 and riot have all signed multi billion dollar AI hosting deals in 2025, the warehouses you scrolled past on tiktok are quietly becoming the inference backbone of every AI tool you use bookmark this and read the article below
1:19

WAREHOUSES THE SIZE OF A WALMART WITH 4,000 GPUS WERE EARNING $320K A MONTH MINING ETH, THE SAME RACKS NOW EARN $4M A MONTH RENTING TO AI STARTUPS ON VAST AI the warehouse in the clip is one of hundreds across the US, the operator at 6'6" tall for scale, electricity contracts and cooling already in place ethereum's merge in september 2022 killed gpu mining for everyone without a power plant, by 2025 alt coin mining was barely covering the electricity bill the same racks now host RTX 3090s and 4090s on vast ai, indie AI developers priced out of AWS rent them by the hour for inference and image generation a 4090 nets $500 to $1,000 a month renting to AI startups, the same card was clearing $80 to $150 mining, 4 to 7 times the revenue with zero new hardware core scientific, hut 8 and riot have all signed multi billion dollar AI hosting deals in 2025, the warehouses you scrolled past on tiktok are quietly becoming the inference backbone of every AI tool you use bookmark this and read the article below

starmex

427,469 görüntüleme • 1 ay önce

Germany has just erased another symbol of its industrial strength. The demolition of the Gundremmingen cooling towers marks not only the end of a nuclear facility but the continued dismantling of Germany’s economic backbone. What began under the red-green coalition two decades ago, and was later accelerated by Angela Merkel’s panic-driven phaseout, is now being completed under Friedrich Merz. The same politicians who once promised “sustainability” are now presiding over energy dependence, deindustrialization, and the collapse of what was once Europe’s most competitive economy. Nuclear power was not Germany’s problem. It was its greatest asset: clean, efficient, and far cheaper than imported gas or unstable renewables. Yet ideological blindness turned success into guilt. The country that once built the best reactors in the world now imports nuclear energy from France and coal-based electricity from Poland. The political class calls this progress while businesses shut down and energy-intensive industries relocate abroad. Under the Merz administration, the absurdity has reached its final stage. While AI data centers, chip fabs, and future manufacturing all demand enormous amounts of reliable power, Germany is blowing up the very plants that could have supplied it. Instead of securing affordable electricity for innovation, Berlin is subsidizing wind turbines that stand still half the year and gas plants that rely on foreign regimes. This is not strategy; it is surrender disguised as virtue. The destruction of Gundremmingen is not about energy policy. It is about the decline of rational governance. Germany’s leaders are sacrificing economic sovereignty for ideology, while pretending it is moral necessity. The result is predictable: higher costs, weaker industry, and growing dependence on those who still value energy realism. What fell today in Bavaria were not cooling towers, but the last remnants of Germany’s common sense.
0:23

Germany has just erased another symbol of its industrial strength. The demolition of the Gundremmingen cooling towers marks not only the end of a nuclear facility but the continued dismantling of Germany’s economic backbone. What began under the red-green coalition two decades ago, and was later accelerated by Angela Merkel’s panic-driven phaseout, is now being completed under Friedrich Merz. The same politicians who once promised “sustainability” are now presiding over energy dependence, deindustrialization, and the collapse of what was once Europe’s most competitive economy. Nuclear power was not Germany’s problem. It was its greatest asset: clean, efficient, and far cheaper than imported gas or unstable renewables. Yet ideological blindness turned success into guilt. The country that once built the best reactors in the world now imports nuclear energy from France and coal-based electricity from Poland. The political class calls this progress while businesses shut down and energy-intensive industries relocate abroad. Under the Merz administration, the absurdity has reached its final stage. While AI data centers, chip fabs, and future manufacturing all demand enormous amounts of reliable power, Germany is blowing up the very plants that could have supplied it. Instead of securing affordable electricity for innovation, Berlin is subsidizing wind turbines that stand still half the year and gas plants that rely on foreign regimes. This is not strategy; it is surrender disguised as virtue. The destruction of Gundremmingen is not about energy policy. It is about the decline of rational governance. Germany’s leaders are sacrificing economic sovereignty for ideology, while pretending it is moral necessity. The result is predictable: higher costs, weaker industry, and growing dependence on those who still value energy realism. What fell today in Bavaria were not cooling towers, but the last remnants of Germany’s common sense.

Torsten Prochnow

75,877 görüntüleme • 9 ay önce

MARKETS OPEN IN 20. every day is different. I will post unusual volume as I see it. I will give long thesis for potential 10x on small caps. We won’t miss another. Here’s another $DDD I give ideas on stocks that fly under the radar you won’t see me talking about the names 90% x are on. Be different. Be unique. Be you. Ideas are ideas. But finding another account working harder Good luck. 😤 Research 24/7 Scanning charts daily 500% runners & 10x longs I’m a real person DMs always open I’m shouting out legends every day who provide value I also just unfollowed 117 accounts who didn’t want rock with us. If you’re working hard and want to be seen and found on here. Follow and subscribe. I’m trying to find new hard workers and legends daily. Engaging with me means I’ll go straight to your account and do the same and if I like your content I make lists daily to do shouts. Plus. Alpha. Alpha. Alpha. More alpha Here’s some accounts you all should look into following BoltEdge ⚡️ Leif | Investing Tyler SolSuit d i v e r g e n t Doris Di Franklin407 Perspez GP the Fundamental Investor Eno 🥷 KURREN$Y KAPITAL Dmytro Lebid William Paulson Let’s all bank fam. Subscribe for just 1$ The community I’m building here on x will be one to remember. Follow me. Follow the people I suggest to follow. Subscribe. 🏦 BANK. $DDD is shifting into a high-value manufacturing platform positioned across: • AI data center thermal systems • semiconductor capital equipment • aerospace & defense production • medical & dental manufacturing THE NUMBERS THAT MATTER • Revenue (Q1 2026): $95.5M (+1% YoY, +11% ex-divestitures) • Healthcare segment: $50.1M (+21% YoY) • Gross margin: ~35.9–36.1% (expanding YoY) • Adjusted EBITDA: + $2.1M (from -$23.9M last year) • Cash: ~$86.5M • Market cap: ~$350M–$500M range (micro-cap re-rating zone) WHY THIS STORY IS DIFFERENT NOW 1) AI DATA CENTER COOLING DEMAND IS EXPLODING AI racks are moving into: • 30–140kW+ per rack • liquid cooling becoming mandatory • thermal engineering becoming critical infrastructure DDD already works in: • semiconductor wafer-stage thermal control • precision fluid manifolds • vacuum/cleanroom-compatible metal parts • ultra-high accuracy cooling systems That directly translates to: • cold plates • heat exchangers • direct-to-chip liquid cooling hardware 2) SEMICONDUCTOR + PRECISION ENGINEERING MOAT DDD’s advantage is not “printing parts.” It’s precision thermal + fluid physics manufacturing. They already solve: • <4 mK thermal stability systems • microfluidic channel optimization • vibration reduction via monolithic metal design • complex copper and alloy geometries 3) DEFENSE + ONSHORE MANUFACTURING TAILWIND Exposure across: • aerospace & defense programs • U.S. Air Force initiatives • naval + advanced materials supply chains • America Makes ecosystem Trend tailwind: 👉 reshoring + secure domestic production Defense buyers prioritize: • reliability • qualification • supply chain security not lowest cost $DDD fits that model. 4) TURNAROUND METRICS ARE IMPROVING What changed vs prior cycle: • Operating losses sharply reduced • ~$55M cost cuts implemented • SG&A materially down • EBITDA inflection achieved • Healthcare now a core growth engine 5) MIX SHIFT = HIGHER QUALITY REVENUE Growth areas: • Dental + MedTech (+20%+) • Aerospace & defense (double-digit growth) • materials + recurring consumables This is a $350–500M micro-cap showing: • positive EBITDA inflection • expanding margins • AI infrastructure adjacency • defense + semicap exposure • healthcare growth engine If execution continues, the rerating case is not about “3D printing hype” — it’s about whether the market starts valuing it as critical thermal + precision manufacturing infrastructure for AI and defense. $VOO $HIVE $HYLN $VELO $HYFT $CELH $PLTR $HIMS $RDDT $AIB $FEMY $JOBY $TE $SIVE $ZS $MRVL $SNPS $SNOW $DELL $LAES
1:34

MARKETS OPEN IN 20. every day is different. I will post unusual volume as I see it. I will give long thesis for potential 10x on small caps. We won’t miss another. Here’s another $DDD I give ideas on stocks that fly under the radar you won’t see me talking about the names 90% x are on. Be different. Be unique. Be you. Ideas are ideas. But finding another account working harder Good luck. 😤 Research 24/7 Scanning charts daily 500% runners & 10x longs I’m a real person DMs always open I’m shouting out legends every day who provide value I also just unfollowed 117 accounts who didn’t want rock with us. If you’re working hard and want to be seen and found on here. Follow and subscribe. I’m trying to find new hard workers and legends daily. Engaging with me means I’ll go straight to your account and do the same and if I like your content I make lists daily to do shouts. Plus. Alpha. Alpha. Alpha. More alpha Here’s some accounts you all should look into following BoltEdge ⚡️ Leif | Investing Tyler SolSuit d i v e r g e n t Doris Di Franklin407 Perspez GP the Fundamental Investor Eno 🥷 KURREN$Y KAPITAL Dmytro Lebid William Paulson Let’s all bank fam. Subscribe for just 1$ The community I’m building here on x will be one to remember. Follow me. Follow the people I suggest to follow. Subscribe. 🏦 BANK. $DDD is shifting into a high-value manufacturing platform positioned across: • AI data center thermal systems • semiconductor capital equipment • aerospace & defense production • medical & dental manufacturing THE NUMBERS THAT MATTER • Revenue (Q1 2026): $95.5M (+1% YoY, +11% ex-divestitures) • Healthcare segment: $50.1M (+21% YoY) • Gross margin: ~35.9–36.1% (expanding YoY) • Adjusted EBITDA: + $2.1M (from -$23.9M last year) • Cash: ~$86.5M • Market cap: ~$350M–$500M range (micro-cap re-rating zone) WHY THIS STORY IS DIFFERENT NOW 1) AI DATA CENTER COOLING DEMAND IS EXPLODING AI racks are moving into: • 30–140kW+ per rack • liquid cooling becoming mandatory • thermal engineering becoming critical infrastructure DDD already works in: • semiconductor wafer-stage thermal control • precision fluid manifolds • vacuum/cleanroom-compatible metal parts • ultra-high accuracy cooling systems That directly translates to: • cold plates • heat exchangers • direct-to-chip liquid cooling hardware 2) SEMICONDUCTOR + PRECISION ENGINEERING MOAT DDD’s advantage is not “printing parts.” It’s precision thermal + fluid physics manufacturing. They already solve: • <4 mK thermal stability systems • microfluidic channel optimization • vibration reduction via monolithic metal design • complex copper and alloy geometries 3) DEFENSE + ONSHORE MANUFACTURING TAILWIND Exposure across: • aerospace & defense programs • U.S. Air Force initiatives • naval + advanced materials supply chains • America Makes ecosystem Trend tailwind: 👉 reshoring + secure domestic production Defense buyers prioritize: • reliability • qualification • supply chain security not lowest cost $DDD fits that model. 4) TURNAROUND METRICS ARE IMPROVING What changed vs prior cycle: • Operating losses sharply reduced • ~$55M cost cuts implemented • SG&A materially down • EBITDA inflection achieved • Healthcare now a core growth engine 5) MIX SHIFT = HIGHER QUALITY REVENUE Growth areas: • Dental + MedTech (+20%+) • Aerospace & defense (double-digit growth) • materials + recurring consumables This is a $350–500M micro-cap showing: • positive EBITDA inflection • expanding margins • AI infrastructure adjacency • defense + semicap exposure • healthcare growth engine If execution continues, the rerating case is not about “3D printing hype” — it’s about whether the market starts valuing it as critical thermal + precision manufacturing infrastructure for AI and defense. $VOO $HIVE $HYLN $VELO $HYFT $CELH $PLTR $HIMS $RDDT $AIB $FEMY $JOBY $TE $SIVE $ZS $MRVL $SNPS $SNOW $DELL $LAES

Hunter Allen

19,993 görüntüleme • 2 ay önce

Last week, we pushed our Hadfield MK IV engine and Darkhorse engine test cell to their limits ahead of our upcoming thrust vector control (TVC) and orbital engine test campaigns. It was thrilling to see this controlled test go sideways — literally! Orbital launch vehicles operate within narrow design margins and constrained safety factors, where excess mass in any subsystem directly impacts payload capacity or mission viability. Destructive and limit testing enable us to validate optimal mass-performance trade-offs across propulsion, pressure systems, and primary structures. Key outcomes from this test include: ✅ Structural margins validated – Darkhorse demonstrated stable operation under full thrust loads at gimbal angles exceeding design specifications ✅ Thermal performance characterized – Extended burn duration at off-nominal mixture ratios provided empirical data on regenerative cooling degradation modes and injector thermal limits ✅ Fault tolerance demonstrated – Engine maintained functionality despite progressive damage, validating robustness for anomalous flight conditions ✅ TVC readiness confirmed – Test results validate system integration for upcoming actuated TVC test series l Design optimization insights – Failure mode analysis generated actionable improvements for cooling architecture, injector design, thrust structures, and engine reusability At NordSpace, we push limits. Canada needs to get to orbit with sovereign light-lift launch by 2028 and medium-lift launch by the early 2030s. The only way this is possible is through extreme levels of testing, manufacturing, and investment. Our mission to build a Canadian end-to-end space missions capability will change the shape of our nation both on Earth and in space. If you would like to join our mission, please apply for a role at NordSpace via the Careers page on our website, and join us at the Canadian Space Launch Conference on May 5th, in Ottawa. National Defence Defence Research and Development Canada NSERC / CRSNG Canadian Space Agency Transport Canada
0:53

Last week, we pushed our Hadfield MK IV engine and Darkhorse engine test cell to their limits ahead of our upcoming thrust vector control (TVC) and orbital engine test campaigns. It was thrilling to see this controlled test go sideways — literally! Orbital launch vehicles operate within narrow design margins and constrained safety factors, where excess mass in any subsystem directly impacts payload capacity or mission viability. Destructive and limit testing enable us to validate optimal mass-performance trade-offs across propulsion, pressure systems, and primary structures. Key outcomes from this test include: ✅ Structural margins validated – Darkhorse demonstrated stable operation under full thrust loads at gimbal angles exceeding design specifications ✅ Thermal performance characterized – Extended burn duration at off-nominal mixture ratios provided empirical data on regenerative cooling degradation modes and injector thermal limits ✅ Fault tolerance demonstrated – Engine maintained functionality despite progressive damage, validating robustness for anomalous flight conditions ✅ TVC readiness confirmed – Test results validate system integration for upcoming actuated TVC test series l Design optimization insights – Failure mode analysis generated actionable improvements for cooling architecture, injector design, thrust structures, and engine reusability At NordSpace, we push limits. Canada needs to get to orbit with sovereign light-lift launch by 2028 and medium-lift launch by the early 2030s. The only way this is possible is through extreme levels of testing, manufacturing, and investment. Our mission to build a Canadian end-to-end space missions capability will change the shape of our nation both on Earth and in space. If you would like to join our mission, please apply for a role at NordSpace via the Careers page on our website, and join us at the Canadian Space Launch Conference on May 5th, in Ottawa. National Defence Defence Research and Development Canada NSERC / CRSNG Canadian Space Agency Transport Canada

NordSpace 🇨🇦

11,704 görüntüleme • 6 ay önce

$AMD's heading to $5T MC LT| Lowest $/M tokens 🧵 The real reason why Institutions are FOMOing into AMD while other Semi stocks are underperforming ($NVDA $AVGO) Not Financial Advice! DYOR! Under Dr. Lisa Su’s leadership, AMD has transformed from a distant challenger into a formidable force in AI infrastructure, delivering the industry’s most compelling TCO story for high-volume inference. Her clear vision open ecosystems, aggressive annual roadmaps, rack-scale innovation, and relentless focus on tokens-per-dollar has positioned AMD’s Helios racks as the go-to solution for hyperscalers and AI natives struggling with exploding token costs, collapsing the cost down to $0.0003-$0.0005/M tokens. I will link various threads on this analysis to supply chain and wafer ratio if you are interested in understanding the full picture. In the last 3-4 months, explosive Agentic AI demand significantly increased Inference demand for Agentic AI models with 5-10 agents. If you are a listener of CNBC or Bloomberg, u should know enterprises and companies are complaining abt cost of token, and how it starts to spike up way too much to make sense. The fact that most data center today are run by $NVDA Chips, where the cost is way too high for Training or Inference. 1. Token cost Here are some quick comp, so u understand why $META OpenAI Anthropic $MSFT $AMZN Softbank $GOOGL and many more small to medium AI Natives are buying AMD CPUs and GPUs as much as they want, or pretty much AMD chips are sold out for the next 3-5 years. Inference (Cost per Million Tokens) ~$NVDA B200 / HGX: ~$0.02–$0.08 on optimized workloads (FP4/MXFP4, speculative decoding). Significant improvement over Hopper but still premium-priced. GB200 NVL72 rack-scale: $0.05–$0.25+ ~$AMD Helios Racks: $0.0003-$0.0005 per M tokens, dramatically lower than NVIDIA equivalents in owned infra. MI355X node-level: Up to 40% more tokens per dollar vs. competing solutions ( B200), driven by higher memory capacity (up to 288GB+ HBM), strong bandwidth, and lower acquisition costs. Training ~$NVDA Rubin Rack is estimated $0.7-$1.2/M Tokens ~$AMD Helios Rack is estimated $0.65-$1.0/M Tokens 2. Why Hyperscalers and AI Natives Are Choosing AMD Token consumption (especially Agentic) is outpacing even NVIDIA’s efficiency gains, making diversification mandatory for economic viability. Massive deals reflect this reality like $META, OpenAI, $MSFT, Softbank, $AMZN, Oracle, LumaAI, G42... Dr. Lisa Su’s Vision in Action: Since taking the helm, Su has driven AMD’s turnaround with disciplined execution, annual GPU cadence (MI300 → MI350 → MI400), full-stack software (ROCm 7), open ecosystems (UALink, OCP designs), and customer-centric rack-scale solutions like Helios. Her emphasis on “tokens per dollar” and TCO has turned AMD into the pragmatic choice for sustainable AI scaling. Power/Energy Efficiency: ~Helios Rack-level is estimated at 120kW-140kW with 50% more HBM4 where Inference and Training cost matter ~Rubin Rack-Level is estimated at 160kW-230kw AMD Helios shines in owned TCO, memory density, and energy flexibility at hyperscale. Cost to build 1GW data center 1GW Helios Rack full build is estimated $30-$35B 1GW Rubin Rack full build is estimated $45-$55B 3. Superior CPUs to pair with GPUs on massive scale 5-10-20GW Agentic AI. autonomous, multi-step workflows with orchestration, tool use, parallel agents, data movement, and enterprise integration has dramatically increased the importance of strong host CPUs alongside GPUs. This shifts the CPU-to-GPU ratio higher and makes balanced systems critical toward 1:1 to 5:1 as enterprises testing more than 5-10 agents. AMD EPYC Venice excels ~Leadership core density (up to 256 Zen 6 cores per socket) for running many agents in parallel, orchestration layers, and high-throughput control-plane tasks. ~Superior performance-per-core and power efficiency ( up to 2.1x higher perf/core and 2.26x better SPECpower vs. NVIDIA Grace in benchmarks). ~Tight integration in Helios: One Venice CPU + multiple MI450 GPUs per node, enabling efficient data feeding to GPUs ("zero-copy"), parallel execution, and full rack utilization for complex agentic loops. Hyperscalers (Meta, Microsoft, Amazon, Google, Softbank) and AI natives (OpenAI, Anthropic...) are adopting high-core EPYC at scale specifically for these agentic demands, as CPUs now handle a larger share of non-model work (orchestration, policy enforcement, tool calls). This complements AMD’s lower-cost GPUs for overall TCO wins. Conclusion: NVIDIA’s Vera Rubin cannot compete with a 2 years old EPYC Turin, but AMD under Dr. Lisa Su has engineered the lowest cost-per-million-tokens, highly competitive energy-efficient solutions, and superior CPU orchestration for agentic AI at scale with Helios. Dr. Su has championed this shift since at least 2023, foreseeing the rise of agentic workflows that demand far more orchestration, parallel agents, and balanced compute well before the industry fully embraced it. Her long-term vision of AI moving from simple prompts to always-on, multi-agent systems has driven AMD’s investments in high-core EPYC CPUs and integrated rack-scale solutions, perfectly positioning the company for today’s realities. Hyperscalers and AI natives effectively have no choice but to buy more AMD system for Agentic AI as leadership in economical, power-aware, high-volume internal + agentic use. However, due to supply constraints where Supply is far behind Demand, this makes multi-vendor reality along with in-house chips drive faster industry progress, lower overall costs, and better sustainability. Not Financial Advice! DYOR! Video source: Microsoft Build 2026
0:49

$AMD's heading to $5T MC LT| Lowest $/M tokens 🧵 The real reason why Institutions are FOMOing into AMD while other Semi stocks are underperforming ($NVDA $AVGO) Not Financial Advice! DYOR! Under Dr. Lisa Su’s leadership, AMD has transformed from a distant challenger into a formidable force in AI infrastructure, delivering the industry’s most compelling TCO story for high-volume inference. Her clear vision open ecosystems, aggressive annual roadmaps, rack-scale innovation, and relentless focus on tokens-per-dollar has positioned AMD’s Helios racks as the go-to solution for hyperscalers and AI natives struggling with exploding token costs, collapsing the cost down to $0.0003-$0.0005/M tokens. I will link various threads on this analysis to supply chain and wafer ratio if you are interested in understanding the full picture. In the last 3-4 months, explosive Agentic AI demand significantly increased Inference demand for Agentic AI models with 5-10 agents. If you are a listener of CNBC or Bloomberg, u should know enterprises and companies are complaining abt cost of token, and how it starts to spike up way too much to make sense. The fact that most data center today are run by $NVDA Chips, where the cost is way too high for Training or Inference. 1. Token cost Here are some quick comp, so u understand why $META OpenAI Anthropic $MSFT $AMZN Softbank $GOOGL and many more small to medium AI Natives are buying AMD CPUs and GPUs as much as they want, or pretty much AMD chips are sold out for the next 3-5 years. Inference (Cost per Million Tokens) ~$NVDA B200 / HGX: ~$0.02–$0.08 on optimized workloads (FP4/MXFP4, speculative decoding). Significant improvement over Hopper but still premium-priced. GB200 NVL72 rack-scale: $0.05–$0.25+ ~$AMD Helios Racks: $0.0003-$0.0005 per M tokens, dramatically lower than NVIDIA equivalents in owned infra. MI355X node-level: Up to 40% more tokens per dollar vs. competing solutions ( B200), driven by higher memory capacity (up to 288GB+ HBM), strong bandwidth, and lower acquisition costs. Training ~$NVDA Rubin Rack is estimated $0.7-$1.2/M Tokens ~$AMD Helios Rack is estimated $0.65-$1.0/M Tokens 2. Why Hyperscalers and AI Natives Are Choosing AMD Token consumption (especially Agentic) is outpacing even NVIDIA’s efficiency gains, making diversification mandatory for economic viability. Massive deals reflect this reality like $META, OpenAI, $MSFT, Softbank, $AMZN, Oracle, LumaAI, G42... Dr. Lisa Su’s Vision in Action: Since taking the helm, Su has driven AMD’s turnaround with disciplined execution, annual GPU cadence (MI300 → MI350 → MI400), full-stack software (ROCm 7), open ecosystems (UALink, OCP designs), and customer-centric rack-scale solutions like Helios. Her emphasis on “tokens per dollar” and TCO has turned AMD into the pragmatic choice for sustainable AI scaling. Power/Energy Efficiency: ~Helios Rack-level is estimated at 120kW-140kW with 50% more HBM4 where Inference and Training cost matter ~Rubin Rack-Level is estimated at 160kW-230kw AMD Helios shines in owned TCO, memory density, and energy flexibility at hyperscale. Cost to build 1GW data center 1GW Helios Rack full build is estimated $30-$35B 1GW Rubin Rack full build is estimated $45-$55B 3. Superior CPUs to pair with GPUs on massive scale 5-10-20GW Agentic AI. autonomous, multi-step workflows with orchestration, tool use, parallel agents, data movement, and enterprise integration has dramatically increased the importance of strong host CPUs alongside GPUs. This shifts the CPU-to-GPU ratio higher and makes balanced systems critical toward 1:1 to 5:1 as enterprises testing more than 5-10 agents. AMD EPYC Venice excels ~Leadership core density (up to 256 Zen 6 cores per socket) for running many agents in parallel, orchestration layers, and high-throughput control-plane tasks. ~Superior performance-per-core and power efficiency ( up to 2.1x higher perf/core and 2.26x better SPECpower vs. NVIDIA Grace in benchmarks). ~Tight integration in Helios: One Venice CPU + multiple MI450 GPUs per node, enabling efficient data feeding to GPUs ("zero-copy"), parallel execution, and full rack utilization for complex agentic loops. Hyperscalers (Meta, Microsoft, Amazon, Google, Softbank) and AI natives (OpenAI, Anthropic...) are adopting high-core EPYC at scale specifically for these agentic demands, as CPUs now handle a larger share of non-model work (orchestration, policy enforcement, tool calls). This complements AMD’s lower-cost GPUs for overall TCO wins. Conclusion: NVIDIA’s Vera Rubin cannot compete with a 2 years old EPYC Turin, but AMD under Dr. Lisa Su has engineered the lowest cost-per-million-tokens, highly competitive energy-efficient solutions, and superior CPU orchestration for agentic AI at scale with Helios. Dr. Su has championed this shift since at least 2023, foreseeing the rise of agentic workflows that demand far more orchestration, parallel agents, and balanced compute well before the industry fully embraced it. Her long-term vision of AI moving from simple prompts to always-on, multi-agent systems has driven AMD’s investments in high-core EPYC CPUs and integrated rack-scale solutions, perfectly positioning the company for today’s realities. Hyperscalers and AI natives effectively have no choice but to buy more AMD system for Agentic AI as leadership in economical, power-aware, high-volume internal + agentic use. However, due to supply constraints where Supply is far behind Demand, this makes multi-vendor reality along with in-house chips drive faster industry progress, lower overall costs, and better sustainability. Not Financial Advice! DYOR! Video source: Microsoft Build 2026

Mike

145,778 görüntüleme • 2 ay önce

Dylan Patel just mapped out the most important investment theme in AI infrastructure (Save this). "In about two years, solar plus battery will be cheaper than gas." Every new NVIDIA Blackwell rack pulls 120 kilowatts, Rubin Ultra rack pulls 600 kilowatts and the next generation hits a megawatt. The US grid cannot keep up, interconnection queues now run five years in many markets so the entire industry is being forced to solve power from first principles. The solar thesis is already happening. BloombergNEF's 2026 LCOE report, covering 800+ financed projects across 50+ markets puts solar plus 4 hour battery storage at $57 per megawatt-hour. Combined cycle gas turbines hit $102 per megawatt hour, the highest on record, up 16% year over year. In California and parts of Texas, solar plus storage is already cheaper than gas for data center power today and solar panel costs are expected to drop another 30% by 2035. Getting power from the grid into the form chips actually require is an entire industry unto itself and NVIDIA just rewrote the rules. The 800 volt DC transition is the most important infrastructure shift that's happening right now. Today's data centers run on 48 volt DC power delivery, a single next-generation GPU pulls over 2,500 watts and at 48 volts, the current required to power a megawatt rack would melt the copper wiring. The investment thesis breaks into four layers and the first layer is power semiconductors, specifically silicon carbide and gallium nitride. At 800 volts, traditional silicon based IGBTs hit their physical limits. SiC and GaN devices are the mandatory replacement. Infineon estimates $175,000 of semiconductor content per megawatt of AI rack power, versus almost nothing today and by 2030, power semiconductor content per AI cabinet grows from $15,000 to $115,000+. The names here are Infineon ($IFNNY), ON Semiconductor ($ON), Wolfspeed ($WOLF), Navitas ($NVTS), and STMicroelectronics ($STM). The second layer is power management and conversion. Vertiv ($VRT) is NVIDIA's lead architectural collaborator for the 800V transition, building the hardware that converts grid AC to 800V DC and the DC to DC power shelves for ultra dense racks. Eaton ($ETN) and Monolithic Power Systems ($MPWR) round out this layer. The third layer is grid to site infrastructure, GE Vernova ($GEV) builds the heavy electrical equipment that connects utility power to the data center campus. Orders are running at twice the rate of shipments, the classic leading indicator of sustained multi year revenue growth. The fourth layer is behind the meter power generation like your bloom energy because grid interconnection queues run five years, hyperscalers are bypassing the grid entirely, building dedicated gas, solar and battery systems on site. Make sure to follow me Melvin for more opportunities across the AI supply chain.
5:06

Dylan Patel just mapped out the most important investment theme in AI infrastructure (Save this). "In about two years, solar plus battery will be cheaper than gas." Every new NVIDIA Blackwell rack pulls 120 kilowatts, Rubin Ultra rack pulls 600 kilowatts and the next generation hits a megawatt. The US grid cannot keep up, interconnection queues now run five years in many markets so the entire industry is being forced to solve power from first principles. The solar thesis is already happening. BloombergNEF's 2026 LCOE report, covering 800+ financed projects across 50+ markets puts solar plus 4 hour battery storage at $57 per megawatt-hour. Combined cycle gas turbines hit $102 per megawatt hour, the highest on record, up 16% year over year. In California and parts of Texas, solar plus storage is already cheaper than gas for data center power today and solar panel costs are expected to drop another 30% by 2035. Getting power from the grid into the form chips actually require is an entire industry unto itself and NVIDIA just rewrote the rules. The 800 volt DC transition is the most important infrastructure shift that's happening right now. Today's data centers run on 48 volt DC power delivery, a single next-generation GPU pulls over 2,500 watts and at 48 volts, the current required to power a megawatt rack would melt the copper wiring. The investment thesis breaks into four layers and the first layer is power semiconductors, specifically silicon carbide and gallium nitride. At 800 volts, traditional silicon based IGBTs hit their physical limits. SiC and GaN devices are the mandatory replacement. Infineon estimates $175,000 of semiconductor content per megawatt of AI rack power, versus almost nothing today and by 2030, power semiconductor content per AI cabinet grows from $15,000 to $115,000+. The names here are Infineon ($IFNNY), ON Semiconductor ($ON), Wolfspeed ($WOLF), Navitas ($NVTS), and STMicroelectronics ($STM). The second layer is power management and conversion. Vertiv ($VRT) is NVIDIA's lead architectural collaborator for the 800V transition, building the hardware that converts grid AC to 800V DC and the DC to DC power shelves for ultra dense racks. Eaton ($ETN) and Monolithic Power Systems ($MPWR) round out this layer. The third layer is grid to site infrastructure, GE Vernova ($GEV) builds the heavy electrical equipment that connects utility power to the data center campus. Orders are running at twice the rate of shipments, the classic leading indicator of sustained multi year revenue growth. The fourth layer is behind the meter power generation like your bloom energy because grid interconnection queues run five years, hyperscalers are bypassing the grid entirely, building dedicated gas, solar and battery systems on site. Make sure to follow me Melvin for more opportunities across the AI supply chain.

Melvin

107,518 görüntüleme • 1 ay önce

Chamath Palihapitiya just dropped the number that explains the entire AI infrastructure trade (Save this). A gigawatt of compute now costs $100 billion and when he started his Arizona data center project it was $4 to $5 billion, it has gone up 20x in a single investment cycle. The implication is not just that AI infrastructure is expensive but rather that the capital barrier to owning meaningful compute has become so high that only a handful of entities in the world can actually build it and the companies who got there early are sitting on what may be the most durable pricing power in the history of the technology industry. This is the neocloud trade. The neocloud market, purpose-built GPU cloud providers like CoreWeave, Nebius, and Lambda Labs was worth $35 billion in 2026 and is projected to reach $236 billion by 2031, compounding at 46% annually. For context, that is faster growth than cloud computing itself posted in its first decade. The reason is very simple, hyperscalers like AWS, Azure, and Google are building for everything, storage, databases, enterprise software, networking and their GPU pricing reflects the overhead of that full-stack infrastructure. Neoclouds build for one thing only, AI compute. The result is a 60% to 85% cost advantage on the same Nvidia silicon, bare metal H100s at $0.78 to $2.79 per GPU-hour on a neocloud versus $3.43 to $5.07 per GPU-hour on a hyperscaler. That spread does not close as AI demand scales but rather it widens, because hyperscalers have to amortize legacy infrastructure and margin expectations that neoclouds do not carry. Gartner projects that by 2030, neoclouds will capture 20% of the $267 billion AI cloud market, and Vultr's own analysis says at least 80% of GPU market share by end of 2026 will be held by a small group of scaled neocloud providers. Now zoom into Nebius specifically, because it is the most interesting publicly traded proxy for this trade. Nebius is the infrastructure arm of the former Yandex Russia's equivalent of Google rebuilt from the ground up after Russia's invasion of Ukraine by Arkady Volozh and relisted on Nasdaq in October 2024. The team that built it already knew how to run internet-scale infrastructure at the lowest possible cost, which is exactly the operational DNA a neocloud requires. In Q1 2026, Nebius reported revenue of $399 million and already generating serious cash on a young business with revenue growing nearly eightfold year-over-year. Then in March 2026, Meta signed a five-year infrastructure agreement with Nebius worth up to $27 billion, $12 billion in committed dedicated GPU capacity deployments beginning early 2027, plus up to $15 billion more tied to Meta purchasing Nebius's unsold third-party capacity. The deal will be executed on one of the first large-scale deployments of Nvidia's Vera Rubin platform, the next-generation architecture after Blackwell making Nebius one of a tiny number of operators in the world with confirmed priority access to the most advanced AI hardware available. Following the contract, Nebius guided to $7 to $9 billion in annualized recurring revenue for 2026 representing 540% year-over-year growth. Chamath Palihapitiya point about the $100 billion capital moat is the bear case for new entrants and the bull case for incumbents. No one can afford to build the next CoreWeave or Nebius from scratch at current hardware and power costs. The companies that are already built, already contracted, and already deploying Nvidia's latest silicon have a moat that compounds with every GPU generation cycle because they get allocations first, they deploy fastest, and their customers re-sign rather than wait for a new operator that does not yet exist. Come join Milk Road Pro for our full breakdown, the complete neocloud competitive landscape, how to think about Nebius's valuation versus CoreWeave and AI entire thesis. Link below.
0:20

Chamath Palihapitiya just dropped the number that explains the entire AI infrastructure trade (Save this). A gigawatt of compute now costs $100 billion and when he started his Arizona data center project it was $4 to $5 billion, it has gone up 20x in a single investment cycle. The implication is not just that AI infrastructure is expensive but rather that the capital barrier to owning meaningful compute has become so high that only a handful of entities in the world can actually build it and the companies who got there early are sitting on what may be the most durable pricing power in the history of the technology industry. This is the neocloud trade. The neocloud market, purpose-built GPU cloud providers like CoreWeave, Nebius, and Lambda Labs was worth $35 billion in 2026 and is projected to reach $236 billion by 2031, compounding at 46% annually. For context, that is faster growth than cloud computing itself posted in its first decade. The reason is very simple, hyperscalers like AWS, Azure, and Google are building for everything, storage, databases, enterprise software, networking and their GPU pricing reflects the overhead of that full-stack infrastructure. Neoclouds build for one thing only, AI compute. The result is a 60% to 85% cost advantage on the same Nvidia silicon, bare metal H100s at $0.78 to $2.79 per GPU-hour on a neocloud versus $3.43 to $5.07 per GPU-hour on a hyperscaler. That spread does not close as AI demand scales but rather it widens, because hyperscalers have to amortize legacy infrastructure and margin expectations that neoclouds do not carry. Gartner projects that by 2030, neoclouds will capture 20% of the $267 billion AI cloud market, and Vultr's own analysis says at least 80% of GPU market share by end of 2026 will be held by a small group of scaled neocloud providers. Now zoom into Nebius specifically, because it is the most interesting publicly traded proxy for this trade. Nebius is the infrastructure arm of the former Yandex Russia's equivalent of Google rebuilt from the ground up after Russia's invasion of Ukraine by Arkady Volozh and relisted on Nasdaq in October 2024. The team that built it already knew how to run internet-scale infrastructure at the lowest possible cost, which is exactly the operational DNA a neocloud requires. In Q1 2026, Nebius reported revenue of $399 million and already generating serious cash on a young business with revenue growing nearly eightfold year-over-year. Then in March 2026, Meta signed a five-year infrastructure agreement with Nebius worth up to $27 billion, $12 billion in committed dedicated GPU capacity deployments beginning early 2027, plus up to $15 billion more tied to Meta purchasing Nebius's unsold third-party capacity. The deal will be executed on one of the first large-scale deployments of Nvidia's Vera Rubin platform, the next-generation architecture after Blackwell making Nebius one of a tiny number of operators in the world with confirmed priority access to the most advanced AI hardware available. Following the contract, Nebius guided to $7 to $9 billion in annualized recurring revenue for 2026 representing 540% year-over-year growth. Chamath Palihapitiya point about the $100 billion capital moat is the bear case for new entrants and the bull case for incumbents. No one can afford to build the next CoreWeave or Nebius from scratch at current hardware and power costs. The companies that are already built, already contracted, and already deploying Nvidia's latest silicon have a moat that compounds with every GPU generation cycle because they get allocations first, they deploy fastest, and their customers re-sign rather than wait for a new operator that does not yet exist. Come join Milk Road Pro for our full breakdown, the complete neocloud competitive landscape, how to think about Nebius's valuation versus CoreWeave and AI entire thesis. Link below.

Milk Road AI

139,047 görüntüleme • 1 ay önce

This 1988 ABC News report hits different in 2026. They warned that man-made CO₂ could raise global temperatures 4 to 9°F (2.2 to 5°C) within 40–60 years — possibly by 2028 or sooner — with melting polar ice caps and calls for global sacrifices, even suggesting we could no longer act primarily as citizens of individual countries. 38 years later, NASA and NOAA data show roughly 2.1–2.3°F (1.2–1.3°C) total warming since the late 1800s, with about 1.1–1.2°F (0.6–0.7°C) since 1988. We’re nowhere near the upper-end catastrophic scenarios they forecasted for this timeframe. History is full of these overreaches. In the 1970s, major media warned of an impending “global cooling” and a new Ice Age. Al Gore famously claimed in 2006 that the Arctic could be ice-free in summer by 2014 (or “within a decade”). These dramatic predictions often grab attention but frequently miss the mark on timing and severity. The clip is a reminder that we should stay humble about long-term climate forecasts. When past high-confidence apocalyptic timelines don’t materialize, it erodes trust. What stands out to you when you watch old climate alarm reports like this?
1:46

This 1988 ABC News report hits different in 2026. They warned that man-made CO₂ could raise global temperatures 4 to 9°F (2.2 to 5°C) within 40–60 years — possibly by 2028 or sooner — with melting polar ice caps and calls for global sacrifices, even suggesting we could no longer act primarily as citizens of individual countries. 38 years later, NASA and NOAA data show roughly 2.1–2.3°F (1.2–1.3°C) total warming since the late 1800s, with about 1.1–1.2°F (0.6–0.7°C) since 1988. We’re nowhere near the upper-end catastrophic scenarios they forecasted for this timeframe. History is full of these overreaches. In the 1970s, major media warned of an impending “global cooling” and a new Ice Age. Al Gore famously claimed in 2006 that the Arctic could be ice-free in summer by 2014 (or “within a decade”). These dramatic predictions often grab attention but frequently miss the mark on timing and severity. The clip is a reminder that we should stay humble about long-term climate forecasts. When past high-confidence apocalyptic timelines don’t materialize, it erodes trust. What stands out to you when you watch old climate alarm reports like this?

Camus

84,153 görüntüleme • 4 ay önce

Before we get into that ingenious cooling stunt, let's first meet the star of the story: Yuncheng (运城), north China's Shanxi Province. This seemingly low-key city actually has a few remarkable identities that may completely change the way you see it. China's "No. 1 City of National Treasures" Yuncheng is home to 100+ cultural heritage sites under national-level protection, with an astonishing concentration of ancient architecture. Step into almost any county in the area, and you may come across a temple, pagoda, or mural that has survived a thousand years. East Asia's Oldest Paleolithic Site? It is home to one of the earliest known Paleolithic sites in East Asia, dating back approximately 1.8 million years. Some European studies even suggest that the site may be more than 2.4 million years old. Salt Lake City, meet your Chinese twin Yuncheng is home to a salt lake spanning roughly 154 square kilometers, with a salt-harvesting history stretching back some 4,000 years. Salt Lake City, meet China's Salt Lake City. Perhaps it's time to make this a sister-city relationship, Salt Lake City Government ? And its waters are actually saltier than those of the Dead Sea, buoyant enough to let a person float effortlessly on the surface. The only thing that doesn't float here is its fame. Hometown of Guan Yu (关羽) China's legendary warrior saint, Guan Yu, was born there more than 1,800 years ago. Celebrated for his unwavering loyalty and valor, and immortalized by the tragedy of a fatal misjudgment in his final years, he went on to influence Chinese culture for nearly two millennia. Even today, countless temples across China dedicated to him remain active, from Taiwan to Xinjiang. By all accounts, Yuncheng has no shortage of history and culture worthy of making headlines. Yet what recently propelled the city into the national spotlight was not its cultural heritage, though. It was the rooftop of an ordinary residential complex-where it suddenly began to "rain." This summer, a punishing heatwave has swept across Europe, while much of China has also been gripped by prolonged high temperatures. Against this backdrop, an unusual cooling system that sends a fine mist drifting down from above has captured nationwide attention. According to the property developer, the system was officially put into operation in August 2024. Each building, 100 meters tall, is equipped with more than 200 high-pressure misting nozzles installed along the rooftop. The fine droplets absorb heat as they evaporate, helping to lower the surrounding temperature. The system enters regular operation each June, with a single cycle lasting about 10 minutes. My colleagues took on-site measurements. Before the system kicked in, the ground temperature within the complex had climbed to 44 degrees Celsius. Once the system began operating, clouds of fine spray descended from above. Children opened umbrellas and played in the artificial "rain." After the spraying cycle ended, measurements showed that the ground temperature had fallen to around 33 degrees Celsius, while the temperature within the mist-covered area dropped significantly, turning a sweltering afternoon into something noticeably cooler and far more comfortable environment. According to the property management, the operating costs are fully covered by the community's shared property-management budget, meaning residents pay nothing extra. The system draws on ordinary tap water. Because the droplets are so fine, most of the water evaporates before it ever reaches the ground. Beyond cooling the air, it helps reduce dust and provides some irrigation for nearby greenery, allowing the same water to serve multiple purposes. This residential community is not the only place experimenting with creative ways to cope with extreme heat. As high temperatures continue, cities across China are introducing increasingly practical and locally adapted cooling measures. In Beijing, Turpan, and many other cities, air-conditioned service stations have opened their doors to sanitation workers, delivery riders, and other outdoor workers. These stations provide watermelon, chilled drinks, ready-to-eat food, and heatstroke-prevention supplies, all free of charge. In some locations, field workers have also been equipped with cooling vests and other protective gear designed for extreme heat. Kaifeng (开封), in central China's Henan Province, has taken an approach that carries a touch of romance all its own. About a thousand years ago, Kaifeng served as the capital of China, during a period when science and technology, urban civilization, and the visual arts flourished. Today, the city is home to several large theme parks and has become one of China's major cultural tourism destinations. Some have even dubbed it "China's Orlando." Kaifeng is also renowned for its watermelons. This year's bumper harvest, however, has left some local growers struggling to find enough buyers. In response, several theme parks have stepped in to purchase large quantities of local watermelons in bulk. The initiative helps farmers expand their sales channels, and allows the parks to hand out thousands of kilograms of free watermelon to visitors each day, offering a refreshing respite from the summer heat. From mist raining down on rooftops, to icy drinks pressed into the hands of delivery riders, to free watermelons handed out by the truckload-different cities, one shared answer. Fighting the heat doesn't always take billion-dollar mega-projects. Sometimes, it just takes a little imagination, and a lot of heart. It might be a shower of mist falling from a rooftop, a wearable cooling suit, or a slice of freshly cut, ice-cold watermelon. What's the coolest heat-relief idea in YOUR city? Drop it in the comments, let's see who wins summer. (And to our friends in the Southern Hemisphere and Alaska-tell us your winter survival tricks instead. We'll need them in December☕)
0:39

Before we get into that ingenious cooling stunt, let's first meet the star of the story: Yuncheng (运城), north China's Shanxi Province. This seemingly low-key city actually has a few remarkable identities that may completely change the way you see it. China's "No. 1 City of National Treasures" Yuncheng is home to 100+ cultural heritage sites under national-level protection, with an astonishing concentration of ancient architecture. Step into almost any county in the area, and you may come across a temple, pagoda, or mural that has survived a thousand years. East Asia's Oldest Paleolithic Site? It is home to one of the earliest known Paleolithic sites in East Asia, dating back approximately 1.8 million years. Some European studies even suggest that the site may be more than 2.4 million years old. Salt Lake City, meet your Chinese twin Yuncheng is home to a salt lake spanning roughly 154 square kilometers, with a salt-harvesting history stretching back some 4,000 years. Salt Lake City, meet China's Salt Lake City. Perhaps it's time to make this a sister-city relationship, Salt Lake City Government ? And its waters are actually saltier than those of the Dead Sea, buoyant enough to let a person float effortlessly on the surface. The only thing that doesn't float here is its fame. Hometown of Guan Yu (关羽) China's legendary warrior saint, Guan Yu, was born there more than 1,800 years ago. Celebrated for his unwavering loyalty and valor, and immortalized by the tragedy of a fatal misjudgment in his final years, he went on to influence Chinese culture for nearly two millennia. Even today, countless temples across China dedicated to him remain active, from Taiwan to Xinjiang. By all accounts, Yuncheng has no shortage of history and culture worthy of making headlines. Yet what recently propelled the city into the national spotlight was not its cultural heritage, though. It was the rooftop of an ordinary residential complex-where it suddenly began to "rain." This summer, a punishing heatwave has swept across Europe, while much of China has also been gripped by prolonged high temperatures. Against this backdrop, an unusual cooling system that sends a fine mist drifting down from above has captured nationwide attention. According to the property developer, the system was officially put into operation in August 2024. Each building, 100 meters tall, is equipped with more than 200 high-pressure misting nozzles installed along the rooftop. The fine droplets absorb heat as they evaporate, helping to lower the surrounding temperature. The system enters regular operation each June, with a single cycle lasting about 10 minutes. My colleagues took on-site measurements. Before the system kicked in, the ground temperature within the complex had climbed to 44 degrees Celsius. Once the system began operating, clouds of fine spray descended from above. Children opened umbrellas and played in the artificial "rain." After the spraying cycle ended, measurements showed that the ground temperature had fallen to around 33 degrees Celsius, while the temperature within the mist-covered area dropped significantly, turning a sweltering afternoon into something noticeably cooler and far more comfortable environment. According to the property management, the operating costs are fully covered by the community's shared property-management budget, meaning residents pay nothing extra. The system draws on ordinary tap water. Because the droplets are so fine, most of the water evaporates before it ever reaches the ground. Beyond cooling the air, it helps reduce dust and provides some irrigation for nearby greenery, allowing the same water to serve multiple purposes. This residential community is not the only place experimenting with creative ways to cope with extreme heat. As high temperatures continue, cities across China are introducing increasingly practical and locally adapted cooling measures. In Beijing, Turpan, and many other cities, air-conditioned service stations have opened their doors to sanitation workers, delivery riders, and other outdoor workers. These stations provide watermelon, chilled drinks, ready-to-eat food, and heatstroke-prevention supplies, all free of charge. In some locations, field workers have also been equipped with cooling vests and other protective gear designed for extreme heat. Kaifeng (开封), in central China's Henan Province, has taken an approach that carries a touch of romance all its own. About a thousand years ago, Kaifeng served as the capital of China, during a period when science and technology, urban civilization, and the visual arts flourished. Today, the city is home to several large theme parks and has become one of China's major cultural tourism destinations. Some have even dubbed it "China's Orlando." Kaifeng is also renowned for its watermelons. This year's bumper harvest, however, has left some local growers struggling to find enough buyers. In response, several theme parks have stepped in to purchase large quantities of local watermelons in bulk. The initiative helps farmers expand their sales channels, and allows the parks to hand out thousands of kilograms of free watermelon to visitors each day, offering a refreshing respite from the summer heat. From mist raining down on rooftops, to icy drinks pressed into the hands of delivery riders, to free watermelons handed out by the truckload-different cities, one shared answer. Fighting the heat doesn't always take billion-dollar mega-projects. Sometimes, it just takes a little imagination, and a lot of heart. It might be a shower of mist falling from a rooftop, a wearable cooling suit, or a slice of freshly cut, ice-cold watermelon. What's the coolest heat-relief idea in YOUR city? Drop it in the comments, let's see who wins summer. (And to our friends in the Southern Hemisphere and Alaska-tell us your winter survival tricks instead. We'll need them in December☕)

Zhai Xiang

13,477 görüntüleme • 27 gün önce

A single gigawatt of orbital compute requires roughly 200 Starship launches and Elon Musk is not satisfied with gigawatts (Save this). The target is 100 gigawatts of orbital compute per year which means SpaceX is staring down a launch requirement that no organization in human history has ever attempted at anything close to that scale. He acknowledges that scaling to gigawatts per year in orbit is a very hard challenge, but then points to something most people have missed entirely, SpaceX has already demonstrated the foundational capability, because building and launching thousands of Starlink satellites per year is the same industrial problem applied to a different payload. When you understand the orbital compute satellite as a larger version of Starlink V3 with an Nvidia GPU rack at the center instead of a communications payload, the manufacturing and launch scaling challenge stops looking like science fiction and starts looking like a production ramp. The infrastructure to support that ramp is already being built. SpaceX is currently capacitizing for thousands of launches per year, two launch towers and pads in South Texas are operational, the first pad at Cape Canaveral is nearly complete, a second is on the way at Launch Complex 37, and additional locations are already in discussion. As the CFO says it "You need to have those cost curves as you ramp up in volume and time, your costs go down." The vision he describes for what this eventually enables is striking in its specificity. He imagines asking Grok a question on his phone, the inference running on an orbital compute satellite, and the answer coming back down through Starlink direct-to-cell, a complete AI query processed entirely in space, from prompt to response, without touching a single terrestrial data center. That moment, he says, is closer than the industry thinks, with initial capability demonstrations possible as soon as next year. The bottleneck that stands between now and that moment is not the satellite design, the cooling physics, or the silicon, all of which SpaceX has already worked through.
4:23

A single gigawatt of orbital compute requires roughly 200 Starship launches and Elon Musk is not satisfied with gigawatts (Save this). The target is 100 gigawatts of orbital compute per year which means SpaceX is staring down a launch requirement that no organization in human history has ever attempted at anything close to that scale. He acknowledges that scaling to gigawatts per year in orbit is a very hard challenge, but then points to something most people have missed entirely, SpaceX has already demonstrated the foundational capability, because building and launching thousands of Starlink satellites per year is the same industrial problem applied to a different payload. When you understand the orbital compute satellite as a larger version of Starlink V3 with an Nvidia GPU rack at the center instead of a communications payload, the manufacturing and launch scaling challenge stops looking like science fiction and starts looking like a production ramp. The infrastructure to support that ramp is already being built. SpaceX is currently capacitizing for thousands of launches per year, two launch towers and pads in South Texas are operational, the first pad at Cape Canaveral is nearly complete, a second is on the way at Launch Complex 37, and additional locations are already in discussion. As the CFO says it "You need to have those cost curves as you ramp up in volume and time, your costs go down." The vision he describes for what this eventually enables is striking in its specificity. He imagines asking Grok a question on his phone, the inference running on an orbital compute satellite, and the answer coming back down through Starlink direct-to-cell, a complete AI query processed entirely in space, from prompt to response, without touching a single terrestrial data center. That moment, he says, is closer than the industry thinks, with initial capability demonstrations possible as soon as next year. The bottleneck that stands between now and that moment is not the satellite design, the cooling physics, or the silicon, all of which SpaceX has already worked through.

Milk Road AI

67,868 görüntüleme • 2 ay önce

Today we announced our new Fairwater datacenter in Atlanta, connected with our first Fairwater site in Wisconsin and our broader Azure footprint to create the world’s first AI superfactory. Fairwater exemplifies our vision for a fungible fleet: infra that can serve any workload, anywhere, on fit-for-purpose accelerators and network paths, with maximum performance and efficiency. AI workloads have evolved beyond large-scale pre-training. Today, they encompass fine-tuning, reinforcement learning (RL), synthetic data generation, evaluation pipelines, and more. Fairwater is built to support this full lifecycle: Max density: Fairwater’s two-story design and liquid cooling system lets us place racks in three dimensions and pack them with GPUs as densely as possible, minimizing cable runs and improving latency and effective bandwidth. Fleet: Each Fairwater DC can integrate hundreds of thousands of the latest NVIDIA GPUs into a single coherent cluster. This provides flexible infra that can support the full spectrum of workloads, and ensure no GPU is left unnecessarily idle. And that’s on top of the more than 100,000 GB300s coming online this quarter alone for inference across the rest of our fleet. For us, it’s all about turning every gigawatt into the maximum number of useful tokens. Not every GW is created equal! Planet-scale: Every Fairwater DC will connect through our continent-spanning AI WAN to prior generations of AI supercomputers, forming a truly fungible pool of compute. This enables developers to scale beyond the capacity of a single site and dynamically land workloads on the right infra for their needs. Together, these innovations let us bring together different generations of silicon and AI systems across DCs and geos into a single elastic system that scales seamlessly across training and inference workloads And this elastic AI capacity is all available alongside all the other cloud services (compute, storage, databases, app services) that AI agents and workloads need. This is what we mean when we talk about building a fungible fleet – a single, unified platform that pushes the limits of performance per watt and per dollar. Read more:
0:44

Today we announced our new Fairwater datacenter in Atlanta, connected with our first Fairwater site in Wisconsin and our broader Azure footprint to create the world’s first AI superfactory. Fairwater exemplifies our vision for a fungible fleet: infra that can serve any workload, anywhere, on fit-for-purpose accelerators and network paths, with maximum performance and efficiency. AI workloads have evolved beyond large-scale pre-training. Today, they encompass fine-tuning, reinforcement learning (RL), synthetic data generation, evaluation pipelines, and more. Fairwater is built to support this full lifecycle: Max density: Fairwater’s two-story design and liquid cooling system lets us place racks in three dimensions and pack them with GPUs as densely as possible, minimizing cable runs and improving latency and effective bandwidth. Fleet: Each Fairwater DC can integrate hundreds of thousands of the latest NVIDIA GPUs into a single coherent cluster. This provides flexible infra that can support the full spectrum of workloads, and ensure no GPU is left unnecessarily idle. And that’s on top of the more than 100,000 GB300s coming online this quarter alone for inference across the rest of our fleet. For us, it’s all about turning every gigawatt into the maximum number of useful tokens. Not every GW is created equal! Planet-scale: Every Fairwater DC will connect through our continent-spanning AI WAN to prior generations of AI supercomputers, forming a truly fungible pool of compute. This enables developers to scale beyond the capacity of a single site and dynamically land workloads on the right infra for their needs. Together, these innovations let us bring together different generations of silicon and AI systems across DCs and geos into a single elastic system that scales seamlessly across training and inference workloads And this elastic AI capacity is all available alongside all the other cloud services (compute, storage, databases, app services) that AI agents and workloads need. This is what we mean when we talk about building a fungible fleet – a single, unified platform that pushes the limits of performance per watt and per dollar. Read more:

Satya Nadella

907,893 görüntüleme • 9 ay önce