Loading video...

Video Failed to Load

Go Home

The real metric that matters? GPU utilization. With Aethir, customers like DCENT - Driving DePIN growth with Sustainable Edge see utilization rates reaching up to 100%, making ROI clearer & GPU scaling more efficient. DCENT’s Founder & CEO, Hidde, explains how they quantify performance and cost 👇📹

AethirEco's profile picture

Aethir Ecosystem

258,920 subscribers

21,281 views • 7 months ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

"Aethir is the fabric between DCENT and its clients" DCENT struggled with GPU scaling and utilization… until they moved to Aethir. DCENT - Driving DePIN growth with Sustainable Edge’s founder & CEO, Hidde, explains the shift perfectly, a real look at the impact Aethir delivers. Worth watching 👇🏻
1:22

"Aethir is the fabric between DCENT and its clients" DCENT struggled with GPU scaling and utilization… until they moved to Aethir. DCENT - Driving DePIN growth with Sustainable Edge’s founder & CEO, Hidde, explains the shift perfectly, a real look at the impact Aethir delivers. Worth watching 👇🏻

Aethir Ecosystem

15,253 views • 8 months ago

IN NEWS: Crusoe acquires Atero We spoke with Chase Lochmiller (CEO of Crusoe) and Alon Yariv (CEO of Atero) about what the deal unlocks. “Our investment is about accelerating performance, enhancing GPU utilization, and driving reliability. By optimizing storage and usage, we can drive down the cost per token, allowing customers to maximize value.” He added: “Our goal is to invest in the platform and drive massive performance gains to enable our customers to create more value using Crusoe Cloud.”
2:49

IN NEWS: Crusoe acquires Atero We spoke with Chase Lochmiller (CEO of Crusoe) and Alon Yariv (CEO of Atero) about what the deal unlocks. “Our investment is about accelerating performance, enhancing GPU utilization, and driving reliability. By optimizing storage and usage, we can drive down the cost per token, allowing customers to maximize value.” He added: “Our goal is to invest in the platform and drive massive performance gains to enable our customers to create more value using Crusoe Cloud.”

TBPN

12,721 views • 11 months ago

Long Live Outputmaxxing: AI compute grids, Anthropic’s coding takeoff, data center backlash, & frontier systems AMP PBC founder Anjney Midha explains why 95% GPU utilization was considered an outage at Google, why the AI race is no longer just about buying more GPUs, how AMP is trying to make FLOPs flow like megawatts, why data center backlash could become one of AI’s biggest bottlenecks, how Anthropic cracked coding through culture and preparation, why DeepMind research hoarding creates a market failure, and why the next frontier may belong to teams that can “output max” across compute, capital, culture, and science.
0:25

Long Live Outputmaxxing: AI compute grids, Anthropic’s coding takeoff, data center backlash, & frontier systems AMP PBC founder Anjney Midha explains why 95% GPU utilization was considered an outage at Google, why the AI race is no longer just about buying more GPUs, how AMP is trying to make FLOPs flow like megawatts, why data center backlash could become one of AI’s biggest bottlenecks, how Anthropic cracked coding through culture and preparation, why DeepMind research hoarding creates a market failure, and why the next frontier may belong to teams that can “output max” across compute, capital, culture, and science.

Latent.Space

30,621 views • 1 month ago

🎙️ Want a full breakdown on why cost per token is a key metric for AI infrastructure TCO? We’re operating in a power-constrained world. Demand for intelligence is scaling faster than infrastructure can keep up—so metrics like cost per GPU hour or FLOPs per dollar don’t tell the full story. When it comes to TCO, Cost per Token is the metric that captures end-to-end performance across GPUs, CPUs, networking, software, and ecosystem making it the key driver of real profitability and scalability in AI. If you’re not measuring efficiency at the token level, you’re missing what actually drives cost. Watch the deep-dive conversation with Dr. Gerro Prinsloo , Nader Khalil and Carter Abdallah, live on the NVIDIA YouTube channel.
1:05

🎙️ Want a full breakdown on why cost per token is a key metric for AI infrastructure TCO? We’re operating in a power-constrained world. Demand for intelligence is scaling faster than infrastructure can keep up—so metrics like cost per GPU hour or FLOPs per dollar don’t tell the full story. When it comes to TCO, Cost per Token is the metric that captures end-to-end performance across GPUs, CPUs, networking, software, and ecosystem making it the key driver of real profitability and scalability in AI. If you’re not measuring efficiency at the token level, you’re missing what actually drives cost. Watch the deep-dive conversation with Dr. Gerro Prinsloo , Nader Khalil and Carter Abdallah, live on the NVIDIA YouTube channel.

NVIDIA

21,970 views • 3 months ago

Neoclouds like $CRWV & $NBIS are the most durable businesses according to Gavin Baker. The reason CoreWeave can charge a significant premium for its GPU hours is that not all GPU hours are the same. CoreWeave's GPUs are being utilized two to three times more per hour on average than a lower-quality provider's. That utilization gap is the product for $CRWV. In 2005, all you had to do to build a $50 billion market cap retailer was run 1,000 stores across 50 states with different climates and different consumer preferences. That's all you gotta do but in all of American history, roughly only ten companies have managed it. Running a GPU cluster at CoreWeave's utilization rates is harder than that. And don't forget about their impressive customers. $CRWV have over $63 billion in disclosed customer commitments this year alone: - OpenAI alone committed over $21B through 2031 - Meta committed $35B+ - Nvidia signed a $6.3B deal as a paying customer - Jane Street dropped $6B in April. - Anthropic signed a multi-billion dollar deal as well
3:56

Neoclouds like $CRWV & $NBIS are the most durable businesses according to Gavin Baker. The reason CoreWeave can charge a significant premium for its GPU hours is that not all GPU hours are the same. CoreWeave's GPUs are being utilized two to three times more per hour on average than a lower-quality provider's. That utilization gap is the product for $CRWV. In 2005, all you had to do to build a $50 billion market cap retailer was run 1,000 stores across 50 states with different climates and different consumer preferences. That's all you gotta do but in all of American history, roughly only ten companies have managed it. Running a GPU cluster at CoreWeave's utilization rates is harder than that. And don't forget about their impressive customers. $CRWV have over $63 billion in disclosed customer commitments this year alone: - OpenAI alone committed over $21B through 2031 - Meta committed $35B+ - Nvidia signed a $6.3B deal as a paying customer - Jane Street dropped $6B in April. - Anthropic signed a multi-billion dollar deal as well

Milk Road Stocks

83,624 views • 18 days ago

Airbnb CEO Brian Chesky envisions Airbnb becoming a marketplace where people can monetize any asset they own, turning homes, cars, boats, and more into income-generating services. "One way to think of Airbnb is just getting more capacity utilization out of assets. Anything that's empty could be further monetized, and the world could be a bit more efficient." "It could be cheaper to own a car. It could be cheaper to own a house if you could defer the cost by sharing with other people. That was basically how [Airbnb] started. I couldn't afford to pay rent, but that doesn't limit it to homes. It could be anything."
0:52

Airbnb CEO Brian Chesky envisions Airbnb becoming a marketplace where people can monetize any asset they own, turning homes, cars, boats, and more into income-generating services. "One way to think of Airbnb is just getting more capacity utilization out of assets. Anything that's empty could be further monetized, and the world could be a bit more efficient." "It could be cheaper to own a car. It could be cheaper to own a house if you could defer the cost by sharing with other people. That was basically how [Airbnb] started. I couldn't afford to pay rent, but that doesn't limit it to homes. It could be anything."

TBPN

20,981 views • 2 months ago

Hey everyone, today I want to introduce a project that’s aiming to redefine how we access compute for AI — it’s called GPUAI. 🔶 GPUAI: Unlocking Global GPU Power for the AI Era GPUAI isn’t just another GPU marketplace or leasing service. It’s a fully decentralized protocol that connects idle GPU resources around the world — from gaming PCs to data center clusters — and transforms them into a high-performance compute network for AI workloads. 🧠 Why does it matter? Right now, the biggest bottleneck in AI isn’t algorithms — it’s access to compute. Training and running models requires massive GPU power, but it’s locked up in centralized cloud platforms, expensive and hard to access for smaller teams. With GPUAI, anyone can tap into a global GPU pool that’s: ✅ Fully decentralized ✅ Reputation-based and smart contract coordinated ✅ Encrypted and secure ✅ Token-incentivized — meaning contributors get rewarded in $GPUAI 📈 For developers, it’s a flexible way to access GPU compute for training, inference, and more — without cloud lock-in. 💰 For GPU owners, it’s a chance to monetize idle hardware that would otherwise go unused. The protocol is live, the apps are active, and the ecosystem is growing fast. 🌐 Try it yourself at 📖 Learn more on 🎮 Play our community games at This is real infrastructure for the future of AI, not hype. Follow them and explore their mission of decentralized computing at Tell me what you think - if you have a GPU, you can start profiting now. #GPUAI #Web3Infrastructure #AIComputing #DePIN #Decentralization
1:25

Hey everyone, today I want to introduce a project that’s aiming to redefine how we access compute for AI — it’s called GPUAI. 🔶 GPUAI: Unlocking Global GPU Power for the AI Era GPUAI isn’t just another GPU marketplace or leasing service. It’s a fully decentralized protocol that connects idle GPU resources around the world — from gaming PCs to data center clusters — and transforms them into a high-performance compute network for AI workloads. 🧠 Why does it matter? Right now, the biggest bottleneck in AI isn’t algorithms — it’s access to compute. Training and running models requires massive GPU power, but it’s locked up in centralized cloud platforms, expensive and hard to access for smaller teams. With GPUAI, anyone can tap into a global GPU pool that’s: ✅ Fully decentralized ✅ Reputation-based and smart contract coordinated ✅ Encrypted and secure ✅ Token-incentivized — meaning contributors get rewarded in $GPUAI 📈 For developers, it’s a flexible way to access GPU compute for training, inference, and more — without cloud lock-in. 💰 For GPU owners, it’s a chance to monetize idle hardware that would otherwise go unused. The protocol is live, the apps are active, and the ecosystem is growing fast. 🌐 Try it yourself at 📖 Learn more on 🎮 Play our community games at This is real infrastructure for the future of AI, not hype. Follow them and explore their mission of decentralized computing at Tell me what you think - if you have a GPU, you can start profiting now. #GPUAI #Web3Infrastructure #AIComputing #DePIN #Decentralization

The Crypto GEMs

69,984 views • 1 year ago

In the latest Professionally Curious 🧠, Joanna Cook @ SOON📍🇺🇸 🗽 (AI 🦞Arc) had a founder-to-founder talk with Franklyn Wang (founder of Liquid) about his path from Math Olympian → Quant Fund Trading → Founding an AI-native platform that aims to be the Future of Trading. They cover why “conventional wisdom” doesn’t hold up, Robinhood’s impact on mobile-first trading, and why the future of trading will look fundamentally different — with plenty of 🌶️ takes:) Also: building a Gen Z-heavy team, raising from Paradigm Haun Ventures General Catalyst, and scaling from seed to growth. Highlights – How to Raise $25m in this Challenging Market – How a Tweet about Meta & Athropic got Franklyn on TBPN – Why Conventional Wisdom is 💀 – Received the World’s Largest GPU as a Birthday 🎁
34:33

In the latest Professionally Curious 🧠, Joanna Cook @ SOON📍🇺🇸 🗽 (AI 🦞Arc) had a founder-to-founder talk with Franklyn Wang (founder of Liquid) about his path from Math Olympian → Quant Fund Trading → Founding an AI-native platform that aims to be the Future of Trading. They cover why “conventional wisdom” doesn’t hold up, Robinhood’s impact on mobile-first trading, and why the future of trading will look fundamentally different — with plenty of 🌶️ takes:) Also: building a Gen Z-heavy team, raising from Paradigm Haun Ventures General Catalyst, and scaling from seed to growth. Highlights – How to Raise $25m in this Challenging Market – How a Tweet about Meta & Athropic got Franklyn on TBPN – Why Conventional Wisdom is 💀 – Received the World’s Largest GPU as a Birthday 🎁

SOON - Solana Optimistic Network (Mainnet Arc)

11,452 views • 3 months ago

How to Create an Economic Boom in America: Bend the Fiscal Curve Toward Sustainable David Sacks explains: "The thing that probably hurt the Democrats the most was inflation." "Voters clearly do not like the diminishment of their purchasing power." "But how do you stop inflation? You have to raise interest rates." "And that's not good either because that raises the cost of a mortgage, the cost of a car payment, it's bad for investment." "So how do you get out of that box where you either have high inflation or high interest rates?" "The only way is to bend the fiscal curve to that more sustainable path." "And if you can do that, the bond markets will actually give you credit for it in advance." "Right now we're unsustainable. The bond markets know it." "Inflation remains persistently high. The Fed has not been able to cut interest rates the way that they expected to." "If Department of Government Efficiency, working with the rest of the government, can now convince the markets that the US financial picture is more sustainable, we'll get credit for that..." "... interest rates will come down, and that will lead to a boom in the economy."
0:58

How to Create an Economic Boom in America: Bend the Fiscal Curve Toward Sustainable David Sacks explains: "The thing that probably hurt the Democrats the most was inflation." "Voters clearly do not like the diminishment of their purchasing power." "But how do you stop inflation? You have to raise interest rates." "And that's not good either because that raises the cost of a mortgage, the cost of a car payment, it's bad for investment." "So how do you get out of that box where you either have high inflation or high interest rates?" "The only way is to bend the fiscal curve to that more sustainable path." "And if you can do that, the bond markets will actually give you credit for it in advance." "Right now we're unsustainable. The bond markets know it." "Inflation remains persistently high. The Fed has not been able to cut interest rates the way that they expected to." "If Department of Government Efficiency, working with the rest of the government, can now convince the markets that the US financial picture is more sustainable, we'll get credit for that..." "... interest rates will come down, and that will lead to a boom in the economy."

The All-In Podcast

134,744 views • 1 year ago

Time for this week’s #Filecoin announcements! Catch the latest updates from the past 7 days 👇 1. State of Filecoin Q3 by Messari is live! Dive into a comprehensive analysis of Filecoin’s recent growth: ⬆️ New storage deals surged by 16% and storage utilization rose from 26% to 30%. ⬆️ 2. Previando LABITCONF with Filecoin Orbit and Zondax on Oct 30: Gearing up for LaBitconf with #FilecoinOrbit, Zondax, and Aaron Stanley from Filecoin Foundation! An event crafted for blockchain builders and enthusiasts. Where: Crypster Club 3. FILBangkok on Nov 11: DePIN Meets AI. Hosted by Filecoin Foundation in collaboration with CoinDesk. Explore how decentralized infrastructure and AI innovation are reshaping the future. 4. DeStor & Kabat join forces! Bringing energy-efficient, secure, and scalable decentralized cloud storage to businesses across Europe. Stay tuned for more updates, and get involved in building the future of decentralized storage!
0:36

Time for this week’s #Filecoin announcements! Catch the latest updates from the past 7 days 👇 1. State of Filecoin Q3 by Messari is live! Dive into a comprehensive analysis of Filecoin’s recent growth: ⬆️ New storage deals surged by 16% and storage utilization rose from 26% to 30%. ⬆️ 2. Previando LABITCONF with Filecoin Orbit and Zondax on Oct 30: Gearing up for LaBitconf with #FilecoinOrbit, Zondax, and Aaron Stanley from Filecoin Foundation! An event crafted for blockchain builders and enthusiasts. Where: Crypster Club 3. FILBangkok on Nov 11: DePIN Meets AI. Hosted by Filecoin Foundation in collaboration with CoinDesk. Explore how decentralized infrastructure and AI innovation are reshaping the future. 4. DeStor & Kabat join forces! Bringing energy-efficient, secure, and scalable decentralized cloud storage to businesses across Europe. Stay tuned for more updates, and get involved in building the future of decentralized storage!

Filecoin

769,902 views • 1 year ago

Announcing: DePIN Summit Africa in Kenya and Tanzania in July 2025 Hosting DePIN Summit has been one of my favorite highlights of building Escape Velocity over the past few years. DePIN, or Decentralized Physical Infrastructure Networks, use crypto to scale critical infrastructure like wireless, electricity, or compute in a peer-to-peer manner. By cutting out the middle-man, DePINs provide cheaper and more reliable services than the current incumbents. In other words, DePIN is all about putting power back into the hands of users. 2023's Summit showed that DePIN is a MOVEMENT: for the very first time, 100+ DePIN founders and investors sat in the same room, in Menlo Park, all with a shared vision for the future of what decentralized infrastructure networks will become. 2024's Summit showed that DePIN is INEVITABLE: with more than 350 founders, investors, and operators coming together on a summer Saturday, in NYC, to discuss humanity’s urgent need for digital infrastructure networks that can scale to serve exponentially-growing demand decades in the future. In July 2025, DePIN Summit Africa will show that DePIN is GLOBAL by bringing the community together in Mombasa, Kenya and Zanzibar, Tanzania. We are excited to be collaborating with our on-the-ground partners SHARE and ThreeFold to make the event truly special, bringing together DePIN entrepreneurs and investors from abroad, alongside local industry and government leaders who are eager to understand how DePIN can serve their communities. DePIN has been mostly a US-based phenomenon to date, by almost any metric you can pick: number of projects, total capital raised, etc. A few protocols are bucking the trend and showing how DePIN can serve users in emerging markets: for example, local ISPs in Ghana are using DAWN to provide internet services, one of the largest telcos in Mexico is offloading its customers traffic onto Helium🎈, and grid-scale solar farms in India are mining on Glow Foundation. But by and large, DePIN activity is focused on developed markets, and specifically the US.... DePIN in the US is a huge opportunity, but it’s not the only one! Emerging markets are a different ballgame: the end-user demand for more reliable, performant, scalable digital infrastructure in growing economies is enormous across virtually every sector of DePIN: wireless, energy, compute, sensors, logistics, identity, etc. In East Africa, consumers spend 10-20% of their monthly income on internet service for only 3-4 GB of capacity. The user behaviors, unit economics, and regulatory hurdles are fundamentally different than other markets. With DePIN Summit Africa 2025, we hope to bridge that gap and help plant the seeds for DePIN to expand its reach globally, and specifically into East Africa going forward. We’re excited to be joined by headline speakers including: - Kristof De Spiegeleer, co-founder of ThreeFold's decentralized cloud computing network - Neil Chatterjee, co-founder of DAWN's decentralized fixed wireless internet - Jose Aycart, co-founder of SHARE's decentralized internet and compute infrastructure - Alireza Ghods, co-founder of NATIX Network's decentralized digital mapping network - Robin Wingardh, co-founder of wingbits's decentralized flight tracking to Africa - Fredrik Ahlgren, co-founder Sourceful Energy ⚡️'s decentralized virtual power plants - Raullen Chai, co-founder of IoTeX's modular DePIN L1 and development platform If you’d like to join us and help accelerate the growth of DePIN in emerging markets, please apply below 👇
0:38

Announcing: DePIN Summit Africa in Kenya and Tanzania in July 2025 Hosting DePIN Summit has been one of my favorite highlights of building Escape Velocity over the past few years. DePIN, or Decentralized Physical Infrastructure Networks, use crypto to scale critical infrastructure like wireless, electricity, or compute in a peer-to-peer manner. By cutting out the middle-man, DePINs provide cheaper and more reliable services than the current incumbents. In other words, DePIN is all about putting power back into the hands of users. 2023's Summit showed that DePIN is a MOVEMENT: for the very first time, 100+ DePIN founders and investors sat in the same room, in Menlo Park, all with a shared vision for the future of what decentralized infrastructure networks will become. 2024's Summit showed that DePIN is INEVITABLE: with more than 350 founders, investors, and operators coming together on a summer Saturday, in NYC, to discuss humanity’s urgent need for digital infrastructure networks that can scale to serve exponentially-growing demand decades in the future. In July 2025, DePIN Summit Africa will show that DePIN is GLOBAL by bringing the community together in Mombasa, Kenya and Zanzibar, Tanzania. We are excited to be collaborating with our on-the-ground partners SHARE and ThreeFold to make the event truly special, bringing together DePIN entrepreneurs and investors from abroad, alongside local industry and government leaders who are eager to understand how DePIN can serve their communities. DePIN has been mostly a US-based phenomenon to date, by almost any metric you can pick: number of projects, total capital raised, etc. A few protocols are bucking the trend and showing how DePIN can serve users in emerging markets: for example, local ISPs in Ghana are using DAWN to provide internet services, one of the largest telcos in Mexico is offloading its customers traffic onto Helium🎈, and grid-scale solar farms in India are mining on Glow Foundation. But by and large, DePIN activity is focused on developed markets, and specifically the US.... DePIN in the US is a huge opportunity, but it’s not the only one! Emerging markets are a different ballgame: the end-user demand for more reliable, performant, scalable digital infrastructure in growing economies is enormous across virtually every sector of DePIN: wireless, energy, compute, sensors, logistics, identity, etc. In East Africa, consumers spend 10-20% of their monthly income on internet service for only 3-4 GB of capacity. The user behaviors, unit economics, and regulatory hurdles are fundamentally different than other markets. With DePIN Summit Africa 2025, we hope to bridge that gap and help plant the seeds for DePIN to expand its reach globally, and specifically into East Africa going forward. We’re excited to be joined by headline speakers including: - Kristof De Spiegeleer, co-founder of ThreeFold's decentralized cloud computing network - Neil Chatterjee, co-founder of DAWN's decentralized fixed wireless internet - Jose Aycart, co-founder of SHARE's decentralized internet and compute infrastructure - Alireza Ghods, co-founder of NATIX Network's decentralized digital mapping network - Robin Wingardh, co-founder of wingbits's decentralized flight tracking to Africa - Fredrik Ahlgren, co-founder Sourceful Energy ⚡️'s decentralized virtual power plants - Raullen Chai, co-founder of IoTeX's modular DePIN L1 and development platform If you’d like to join us and help accelerate the growth of DePIN in emerging markets, please apply below 👇

EV3 Sal

163,560 views • 1 year ago

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Miyasaki sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.
1:20

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Miyasaki sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.

BeaconLayer Podcast

30,687 views • 1 month ago

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Takahara / 0699.eth sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.
1:20

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Takahara / 0699.eth sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.

BlockLayer Podcast

48,944 views • 1 month ago

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Takahara / 0699.eth sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.
1:20

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Takahara / 0699.eth sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.

BlockLayer Podcast

105,815 views • 2 months ago

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Miyasaki sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.
1:20

“High-performance infrastructure only matters if it expands what users can actually do onchain — and makes that experience accessible at scale.” Sachi Miyasaki sits down with Jay, Co-Founder of Sei, to examine why parallelized execution is becoming increasingly important for the next generation of onchain applications. From trading and DeFi to high-frequency user activity that simply breaks in low-throughput environments, they discuss how lower fees and greater execution capacity can fundamentally reshape the user experience — especially for smaller participants who are otherwise priced out. They also explore how this plays out in practice through projects like Bancor’s Carbon DeFi, where Sei has emerged as the ecosystem driving the strongest activity and volume, underscoring how performance advantages translate into real adoption.

BeaconLayer Podcast

62,114 views • 2 months ago

Costco's CEO reveals the Kirkland test that makes private label feel unfair: A buyer brought him a bacon, egg, and cheese sandwich that matched the national brand. He sent it back. "How can we exceed the national brand on the bacon involved in that?" Because bacon was "the highest value attribute." The buyer came back with "40% more protein" and Costco was still able to "maintain our price." Then he explains the real edge: "They know down to the whole production of the item..." "That just makes them a better buyer when they're dealing with the big brands." The concept is the highest-value-attribute test. Do not beat competitors everywhere. Find the one attribute customers value most, overdeliver there, and use volume/process knowledge to keep price intact. Source: WSJ interview with Costco CEO Ron Vachris
0:52

Costco's CEO reveals the Kirkland test that makes private label feel unfair: A buyer brought him a bacon, egg, and cheese sandwich that matched the national brand. He sent it back. "How can we exceed the national brand on the bacon involved in that?" Because bacon was "the highest value attribute." The buyer came back with "40% more protein" and Costco was still able to "maintain our price." Then he explains the real edge: "They know down to the whole production of the item..." "That just makes them a better buyer when they're dealing with the big brands." The concept is the highest-value-attribute test. Do not beat competitors everywhere. Find the one attribute customers value most, overdeliver there, and use volume/process knowledge to keep price intact. Source: WSJ interview with Costco CEO Ron Vachris

andrew engler

1,722,060 views • 24 days ago

Even if AI will write 90% of code. But it won’t replace engineers. Windsurf’s founder explains that even if AI generates most of the code, engineers still spend time reviewing it, debugging issues, designing systems, testing, deploying, and making architectural decisions. Those parts don’t disappear. He brings up Amdahl’s Law to make the point: speeding up one part of the workflow doesn’t dramatically change the total time unless everything else speeds up too. For example, if writing code takes 30 out of 100 total time units, and AI brings it down to 3, the total time still only drops to 73. That’s just a 27% improvement. More productivity doesn’t mean fewer engineers. It just increases the ROI of building tech. And when ROI goes up, companies are more likely to build—and hire more people to do it. Take JPMorgan Chase. They have 50,000+ engineers and a $17B software budget. If each engineer can ship more software, that doesn’t reduce headcount. It raises the cost of not building. Companies like Windsurf and Anthropic are hiring more engineers—not less—despite being at the edge of AI code generation.
1:52

Even if AI will write 90% of code. But it won’t replace engineers. Windsurf’s founder explains that even if AI generates most of the code, engineers still spend time reviewing it, debugging issues, designing systems, testing, deploying, and making architectural decisions. Those parts don’t disappear. He brings up Amdahl’s Law to make the point: speeding up one part of the workflow doesn’t dramatically change the total time unless everything else speeds up too. For example, if writing code takes 30 out of 100 total time units, and AI brings it down to 3, the total time still only drops to 73. That’s just a 27% improvement. More productivity doesn’t mean fewer engineers. It just increases the ROI of building tech. And when ROI goes up, companies are more likely to build—and hire more people to do it. Take JPMorgan Chase. They have 50,000+ engineers and a $17B software budget. If each engineer can ship more software, that doesn’t reduce headcount. It raises the cost of not building. Companies like Windsurf and Anthropic are hiring more engineers—not less—despite being at the edge of AI code generation.

Aish

27,268 views • 1 year ago

Participants in the GPU rental market had grown accustomed to a familiar regime: relentless cost declines, falling $/token, and the assumption that each new chip generation would compress pricing further. Many onlookers extrapolated this into outright bearishness on older hardware, calling for a wave of impairments across AI clouds as Hopper-era GPUs were rendered economically obsolete. As Dylan noted, the logic seemed straightforward - newer chips dramatically outperform Hopper, so prices should fall. The rise of agentic AI have shifted the paradigm. Pricing is now repricing violently to real-time economic value under constraint. As highlighted in the transcript, even though models like GPT-5.4 are cheaper and more efficient - producing more tokens per GPU at higher quality - this does not reduce demand. Instead, it expands it, in-line with Jevons' Paradox. SemiAnalysis data shows the same dynamic: token usage is accelerating rapidly because the ROI is obvious, making compute demand increasingly price insensitive. In this new regime, supply constraints and token value dominate pricing we think this paradigm is only here to stay if not intensify.
0:24

Participants in the GPU rental market had grown accustomed to a familiar regime: relentless cost declines, falling $/token, and the assumption that each new chip generation would compress pricing further. Many onlookers extrapolated this into outright bearishness on older hardware, calling for a wave of impairments across AI clouds as Hopper-era GPUs were rendered economically obsolete. As Dylan noted, the logic seemed straightforward - newer chips dramatically outperform Hopper, so prices should fall. The rise of agentic AI have shifted the paradigm. Pricing is now repricing violently to real-time economic value under constraint. As highlighted in the transcript, even though models like GPT-5.4 are cheaper and more efficient - producing more tokens per GPU at higher quality - this does not reduce demand. Instead, it expands it, in-line with Jevons' Paradox. SemiAnalysis data shows the same dynamic: token usage is accelerating rapidly because the ROI is obvious, making compute demand increasingly price insensitive. In this new regime, supply constraints and token value dominate pricing we think this paradigm is only here to stay if not intensify.

SemiAnalysis

55,206 views • 4 months ago

Dylan Patel of SemiAnalysis says a worse GPU with better storage and memory now beats the best chip without them, so buying the newest GPU alone no longer wins inference. So, an AMD GPU with more memory can outperform Nvidia in some cases. "So what we have is we have over $80 million of compute, GPUs from Nvidia, AMD, TPUs from Google, Trainium from Amazon, and we run this benchmark constantly on the newest inference engine, newest drivers, newest PyTorch version, whatever it is." "Every day it runs on an automated CI, and we run it on all the latest Chinese models, from GLM, Zhipu, Moonshot, Kimi, Alibaba, all these models we run." "Initially, when we were benchmarking the difference between these chips and different engines, different schemes for parallelism, we were just running it fixed context length." "But now with Agent X, we've analyzed over $5 million worth of Claude Code traces. This is real production traffic that people have donated to us as well as internally generated. Now we know what the actual agent workload looks like." "And then as we implement that and run those benchmarks, it turns out yes, the chip you're using is very important, but now even more important is how are you handling this memory offload?" "And so while an Nvidia GPU is faster than an AMD GPU in most cases, because AMD GPUs have more memory, they actually end up outperforming in some cases." "Or you can have a worse GPU, but a much better storage solution, and now you can outperform what the best GPU can do without those solutions. So just buying the newest and latest GPU alone doesn't get you the best inference economics." "Actually, you need to layer in all these other innovations including storage and memory." [ Who's the top player on your chart? ] "That really is a difficult multivariable problem. And generally that means you need to have, yes, you need to have the best GPU, a GB300, but you also need to have the best storage solutions. And so I won't spoil who's the best right here, but I will say that storage solutions matter a lot and memory solutions matter a lot, as does your front-end networking. That matters a lot."
2:00

Dylan Patel of SemiAnalysis says a worse GPU with better storage and memory now beats the best chip without them, so buying the newest GPU alone no longer wins inference. So, an AMD GPU with more memory can outperform Nvidia in some cases. "So what we have is we have over $80 million of compute, GPUs from Nvidia, AMD, TPUs from Google, Trainium from Amazon, and we run this benchmark constantly on the newest inference engine, newest drivers, newest PyTorch version, whatever it is." "Every day it runs on an automated CI, and we run it on all the latest Chinese models, from GLM, Zhipu, Moonshot, Kimi, Alibaba, all these models we run." "Initially, when we were benchmarking the difference between these chips and different engines, different schemes for parallelism, we were just running it fixed context length." "But now with Agent X, we've analyzed over $5 million worth of Claude Code traces. This is real production traffic that people have donated to us as well as internally generated. Now we know what the actual agent workload looks like." "And then as we implement that and run those benchmarks, it turns out yes, the chip you're using is very important, but now even more important is how are you handling this memory offload?" "And so while an Nvidia GPU is faster than an AMD GPU in most cases, because AMD GPUs have more memory, they actually end up outperforming in some cases." "Or you can have a worse GPU, but a much better storage solution, and now you can outperform what the best GPU can do without those solutions. So just buying the newest and latest GPU alone doesn't get you the best inference economics." "Actually, you need to layer in all these other innovations including storage and memory." [ Who's the top player on your chart? ] "That really is a difficult multivariable problem. And generally that means you need to have, yes, you need to have the best GPU, a GB300, but you also need to have the best storage solutions. And so I won't spoil who's the best right here, but I will say that storage solutions matter a lot and memory solutions matter a lot, as does your front-end networking. That matters a lot."

Fireside Alpha

175,586 views • 18 days ago

Dylan Patel on the importance of memory and storage Two key quotes: "An $NVDA GPU is faster than an $AMD GPU in most cases, but because AMD GPUs have more memory, they can outperform Nvidia in certain workloads." “It is a difficult, multivariable problem. Generally, you need the best GPU, such as a GB300, but you also need the best storage solutions. I will not spoil who comes out on top, but storage solutions matter a lot, memory solutions matter a lot, and frontend networking also matters significantly" Full Quote: “We have over $80 million of compute: GPUs from $NVDA and $AMD, TPUs from Google, and Trainium from Amazon. We constantly run this benchmark using the newest inference engines, drivers, PyTorch versions, and other software. It runs every day through automated CI across the latest Chinese models from GLM, Zhipu, Moonshot, Kimi, Alibaba, and others. Initially, when we were benchmarking the differences between these chips, inference engines, and parallelism schemes, we used fixed context lengths. But with Agent X, we have now analyzed more than $5 million worth of Claude Code traces. This is real production traffic that users have donated to us, combined with internally generated data, so we now understand what an actual agent workload looks like. When we implement those workloads and run the benchmarks, it turns out that the chip you are using is very important, but how you handle memory offload can be even more important. An Nvidia GPU is faster than an AMD GPU in most cases, but because AMD GPUs have more memory, they can outperform Nvidia in certain workloads. Similarly, you can use a less powerful GPU with a much better storage solution and outperform the best GPU when it lacks those solutions. Simply buying the newest GPU does not necessarily give you the best inference economics. You need to layer in other innovations, including storage and memory.” Interviewer: “Who is the top player on your chart? Can you tell us?” Dylan Patel: “It is a difficult, multivariable problem. Generally, you need the best GPU, such as a GB300, but you also need the best storage solutions. I will not spoil who comes out on top, but storage solutions matter a lot, memory solutions matter a lot, and frontend networking also matters significantly.”
2:27

Dylan Patel on the importance of memory and storage Two key quotes: "An $NVDA GPU is faster than an $AMD GPU in most cases, but because AMD GPUs have more memory, they can outperform Nvidia in certain workloads." “It is a difficult, multivariable problem. Generally, you need the best GPU, such as a GB300, but you also need the best storage solutions. I will not spoil who comes out on top, but storage solutions matter a lot, memory solutions matter a lot, and frontend networking also matters significantly" Full Quote: “We have over $80 million of compute: GPUs from $NVDA and $AMD, TPUs from Google, and Trainium from Amazon. We constantly run this benchmark using the newest inference engines, drivers, PyTorch versions, and other software. It runs every day through automated CI across the latest Chinese models from GLM, Zhipu, Moonshot, Kimi, Alibaba, and others. Initially, when we were benchmarking the differences between these chips, inference engines, and parallelism schemes, we used fixed context lengths. But with Agent X, we have now analyzed more than $5 million worth of Claude Code traces. This is real production traffic that users have donated to us, combined with internally generated data, so we now understand what an actual agent workload looks like. When we implement those workloads and run the benchmarks, it turns out that the chip you are using is very important, but how you handle memory offload can be even more important. An Nvidia GPU is faster than an AMD GPU in most cases, but because AMD GPUs have more memory, they can outperform Nvidia in certain workloads. Similarly, you can use a less powerful GPU with a much better storage solution and outperform the best GPU when it lacks those solutions. Simply buying the newest GPU does not necessarily give you the best inference economics. You need to layer in other innovations, including storage and memory.” Interviewer: “Who is the top player on your chart? Can you tell us?” Dylan Patel: “It is a difficult, multivariable problem. Generally, you need the best GPU, such as a GB300, but you also need the best storage solutions. I will not spoil who comes out on top, but storage solutions matter a lot, memory solutions matter a lot, and frontend networking also matters significantly.”

Daniel Romero

38,220 views • 1 month ago