Loading video...

Video Failed to Load

Go Home

Why is Google still using Broadcom as a middleman to TSMC? Dylan Patel answers: "Broadcom is the biggest networking company in the world." "The networking side of things is so important, and the technical competence of everyone around the world besides Broadcom and Nvidia in networking is just not...

tbpn's profile picture

TBPN

135,333 subscribers

128,752 views • 8 months ago •via X (Twitter)

Science & Technology
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Broadcom turns Google’s TPU IP into chips and networking, while Google earns from its IP without doing the hardware build itself. "Broadcom is making ~$37 B from 1GW of Anthropic's orders, whereas Google is capturing ~$13-14 B of that"
1:23

Broadcom turns Google’s TPU IP into chips and networking, while Google earns from its IP without doing the hardware build itself. "Broadcom is making ~$37 B from 1GW of Anthropic's orders, whereas Google is capturing ~$13-14 B of that"

Rohan Paul

72,987 views • 4 months ago

$AVGO helps hyperscalers build the custom chips and networking fabric needed to run massive AI systems. If $GOOGL, $META, OpenAI or Anthropic want to build their own inference chips then Broadcom is one of the first calls because it has the IP, packaging expertise and networking portfolio required to scale custom AI systems. Thats why Broadcom remains one of the strongest infrastructure positions in the AI economy.
1:00

$AVGO helps hyperscalers build the custom chips and networking fabric needed to run massive AI systems. If $GOOGL, $META, OpenAI or Anthropic want to build their own inference chips then Broadcom is one of the first calls because it has the IP, packaging expertise and networking portfolio required to scale custom AI systems. Thats why Broadcom remains one of the strongest infrastructure positions in the AI economy.

Shay Boloor

77,765 views • 2 months ago

Broadcom's CEO just exposed the real fight underneath Google's AI chip strategy. It is not Google versus Broadcom. It is Google and Broadcom trying to make Nvidia replaceable. Within two minutes at Bloomberg Tech, Hock Tan was asked whether Google bringing more chip design in house keeps him up at night. His exact words: "So we just compete against my own customer." Then he named the real enemy: "the real competitor facing all this is the GPU out of Nvidia." That is the part most people miss. Custom AI chips are not just cheaper GPUs. They are ownership claims. If Google owns the workload, the compiler stack, the cloud customer, and the TPU roadmap, Nvidia becomes a benchmark instead of the toll booth. But Broadcom is still in the room because independence is not binary. The hard part is not drawing a chip. The hard part is shipping generation after generation at scale, matching Nvidia's cadence, keeping networking tight, and making the whole system useful enough that developers do not care what silicon sits underneath. That is why Tan can say Google is trying to create customer owned tooling and still sound calm. Broadcom is not selling picks and shovels. It is selling the bridge out of Nvidia dependency. The numbers explain why this is suddenly a board level issue. Broadcom reported $22.2 billion of Q2 2026 revenue. Its AI semiconductor revenue hit $10.8 billion, up 143 percent year over year. For Q3, Broadcom guided AI semiconductor revenue to $16.0 billion, up more than 200 percent year over year. In the clip, Tan says Broadcom has exactly 6 custom AI accelerator customers. He says OpenAI has been engaged for over 2 years, its accelerator is already working in labs and data centers, and production is on track for late this year. That is the hidden mechanism: The AI labs are not becoming software companies with some chips attached. They are becoming capacity companies with model interfaces attached. Once your margin depends on tokens, latency, memory bandwidth, power contracts, packaging slots, networking gear, and a private accelerator schedule, the "model company" label starts to look like a costume. The precedent is Apple. Apple did not move into custom silicon because it wanted a cute chip branding story. It moved because the iPhone needed control over performance per watt, release cadence, and differentiation. A series chips in 2010. M1 in 2020. More than a decade of slowly pulling the bottleneck inside the company. But Apple still needed TSMC. That is the useful analogy for Google, OpenAI, and the other AI giants. They want Nvidia's margin pool. They want Nvidia's roadmap power. They want Nvidia's ability to decide who gets capacity first. But the first supplier they replace becomes the supplier they cannot live without. Broadcom is the customs officer at the border of private silicon. Second order consequence: AI company valuation will shift from model demos to infrastructure custody. Who owns the workload? Who controls the accelerator roadmap? Who has memory secured? Who can afford to keep a bad first generation alive long enough to get to the second and third? My bet: by the end of 2027, at least one major AI lab will be judged more by its custom chip execution than by its model benchmark lead. The model race is public. The margin race is being negotiated in silicon.
0:55

Broadcom's CEO just exposed the real fight underneath Google's AI chip strategy. It is not Google versus Broadcom. It is Google and Broadcom trying to make Nvidia replaceable. Within two minutes at Bloomberg Tech, Hock Tan was asked whether Google bringing more chip design in house keeps him up at night. His exact words: "So we just compete against my own customer." Then he named the real enemy: "the real competitor facing all this is the GPU out of Nvidia." That is the part most people miss. Custom AI chips are not just cheaper GPUs. They are ownership claims. If Google owns the workload, the compiler stack, the cloud customer, and the TPU roadmap, Nvidia becomes a benchmark instead of the toll booth. But Broadcom is still in the room because independence is not binary. The hard part is not drawing a chip. The hard part is shipping generation after generation at scale, matching Nvidia's cadence, keeping networking tight, and making the whole system useful enough that developers do not care what silicon sits underneath. That is why Tan can say Google is trying to create customer owned tooling and still sound calm. Broadcom is not selling picks and shovels. It is selling the bridge out of Nvidia dependency. The numbers explain why this is suddenly a board level issue. Broadcom reported $22.2 billion of Q2 2026 revenue. Its AI semiconductor revenue hit $10.8 billion, up 143 percent year over year. For Q3, Broadcom guided AI semiconductor revenue to $16.0 billion, up more than 200 percent year over year. In the clip, Tan says Broadcom has exactly 6 custom AI accelerator customers. He says OpenAI has been engaged for over 2 years, its accelerator is already working in labs and data centers, and production is on track for late this year. That is the hidden mechanism: The AI labs are not becoming software companies with some chips attached. They are becoming capacity companies with model interfaces attached. Once your margin depends on tokens, latency, memory bandwidth, power contracts, packaging slots, networking gear, and a private accelerator schedule, the "model company" label starts to look like a costume. The precedent is Apple. Apple did not move into custom silicon because it wanted a cute chip branding story. It moved because the iPhone needed control over performance per watt, release cadence, and differentiation. A series chips in 2010. M1 in 2020. More than a decade of slowly pulling the bottleneck inside the company. But Apple still needed TSMC. That is the useful analogy for Google, OpenAI, and the other AI giants. They want Nvidia's margin pool. They want Nvidia's roadmap power. They want Nvidia's ability to decide who gets capacity first. But the first supplier they replace becomes the supplier they cannot live without. Broadcom is the customs officer at the border of private silicon. Second order consequence: AI company valuation will shift from model demos to infrastructure custody. Who owns the workload? Who controls the accelerator roadmap? Who has memory secured? Who can afford to keep a bad first generation alive long enough to get to the second and third? My bet: by the end of 2027, at least one major AI lab will be judged more by its custom chip execution than by its model benchmark lead. The model race is public. The margin race is being negotiated in silicon.

Andrej Drats

10,572 views • 1 month ago

As we wrote in our TPUv7 article, TPUv7 will have competitive perf per TCO verus Blackwell in 2025 for large labs but TPUv8 will have worse perf per TCO verus Rubin in EOY 2026/2027. We added an subway surfer video as some people don't have the attention span to read an 10,000 word article. There will be 2 versions of TPU v8 but this will be different from the “P” (full) and “E” (lite) SKUs that featured in the then 4th and 5th generations. Rather, it is a dual track with one SKU co-designed with Broadcom (TPU 8AX ) and one SKU co-designed with MediaTek (TPU 8X). The strategy behind Google choosing to partner with MediaTek is to reduce the margin that they pay to their silicon design partners or specifically Broadcom. Broadcom charges Google for the whole System in Package which they stack a healthy margin on. This includes HBM. This is despite Google being largely responsible for both the front-end and back-end design of the compute elements of the chip, with Broadcom contributing various PHYs (most importantly Broadcom’s best in class SerDes) and controllers. Google wants to move to a model where their silicon partner only charges for the things they add value on and get closer to paying BOM cost for their silicon.
0:28

As we wrote in our TPUv7 article, TPUv7 will have competitive perf per TCO verus Blackwell in 2025 for large labs but TPUv8 will have worse perf per TCO verus Rubin in EOY 2026/2027. We added an subway surfer video as some people don't have the attention span to read an 10,000 word article. There will be 2 versions of TPU v8 but this will be different from the “P” (full) and “E” (lite) SKUs that featured in the then 4th and 5th generations. Rather, it is a dual track with one SKU co-designed with Broadcom (TPU 8AX ) and one SKU co-designed with MediaTek (TPU 8X). The strategy behind Google choosing to partner with MediaTek is to reduce the margin that they pay to their silicon design partners or specifically Broadcom. Broadcom charges Google for the whole System in Package which they stack a healthy margin on. This includes HBM. This is despite Google being largely responsible for both the front-end and back-end design of the compute elements of the chip, with Broadcom contributing various PHYs (most importantly Broadcom’s best in class SerDes) and controllers. Google wants to move to a model where their silicon partner only charges for the things they add value on and get closer to paying BOM cost for their silicon.

SemiAnalysis

133,077 views • 8 months ago

AI-scale networking is hitting a new power and density ceiling — and Dell + Broadcom are rethinking the fabric from the thermals up. At SC25, Jim Wynia breaks down why Direct Liquid Cooling (DLC) is becoming a must for next-gen AI networks. Wynia puts it simply: when the hardware runs cooler, everything runs more efficiently — and the AI fabric scales further. Watch the full conversation on how Dell and Broadcom are reclaiming power and performance with the Tomahawk 6 and Enterprise SONiC.
1:21

AI-scale networking is hitting a new power and density ceiling — and Dell + Broadcom are rethinking the fabric from the thermals up. At SC25, Jim Wynia breaks down why Direct Liquid Cooling (DLC) is becoming a must for next-gen AI networks. Wynia puts it simply: when the hardware runs cooler, everything runs more efficiently — and the AI fabric scales further. Watch the full conversation on how Dell and Broadcom are reclaiming power and performance with the Tomahawk 6 and Enterprise SONiC.

Six Five Media

66,740 views • 8 months ago

What's the next big trade in AI infrastructure after memory? We asked Meltem Demirors and Kelly Greer: "Memory is ripping now. We think the next narrative is gonna be around networking." "Jensen is doing the exact same thing now that he did with Mellanox in 2019 that turned NVIDIA from a GPU company into a systems and networking company. He's doing it again with Marvell." "Hyperscalers are literally telling you the two biggest bottlenecks for AI spend right now are networking and memory." "The tea leaves are written literally everywhere... the market's just been on the memory track. We're trying to get ahead of that."
2:51

What's the next big trade in AI infrastructure after memory? We asked Meltem Demirors and Kelly Greer: "Memory is ripping now. We think the next narrative is gonna be around networking." "Jensen is doing the exact same thing now that he did with Mellanox in 2019 that turned NVIDIA from a GPU company into a systems and networking company. He's doing it again with Marvell." "Hyperscalers are literally telling you the two biggest bottlenecks for AI spend right now are networking and memory." "The tea leaves are written literally everywhere... the market's just been on the memory track. We're trying to get ahead of that."

MTS

35,125 views • 2 months ago

Jensen Huang says $NVDA invested $2B in $MRVL because the TAM for networking and custom silicon is entering an inference boom. Nvidia is becoming the platform layer for a more distributed AI factory architecture increasingly built around its networking fabric.
3:17

Jensen Huang says $NVDA invested $2B in $MRVL because the TAM for networking and custom silicon is entering an inference boom. Nvidia is becoming the platform layer for a more distributed AI factory architecture increasingly built around its networking fabric.

Shay Boloor

111,005 views • 4 months ago

200ms of network latency... and the feet still stay planted! 👀 Early test integrating Box3D by Erin Catto into Unreal Engine as a plugin. Still a work in progress, but the stability of the foot placement under latency is already looking really promising! Next step: more features, improved networking, and more physics experiments!🚀 #UnrealEngine #GameDev #Physics #Networking #IndieDev #Box3D
0:49

200ms of network latency... and the feet still stay planted! 👀 Early test integrating Box3D by Erin Catto into Unreal Engine as a plugin. Still a work in progress, but the stability of the foot placement under latency is already looking really promising! Next step: more features, improved networking, and more physics experiments!🚀 #UnrealEngine #GameDev #Physics #Networking #IndieDev #Box3D

Antonio Lattanzio

14,444 views • 1 month ago

The wait is over! NLL is back and bigger than ever. As the ultimate showcase of talent, passion, sportsmanship and networking, this year promises thrilling football action, fierce rivalries, and unforgettable moments. All teams are set, the stage is ready, almost all teams have strengthened their ranks. More than just football, the Ntare Lions League is a celebration of networking, brotherhood and tradition. Whether you're playing on the pitch or cheering from the stands, everyone is part of this exciting journey. But the biggest question remains Who will be crowned champions in Fort Portal? Bring on #NLLXV #NLLXV #Road2FortPortal
0:57

The wait is over! NLL is back and bigger than ever. As the ultimate showcase of talent, passion, sportsmanship and networking, this year promises thrilling football action, fierce rivalries, and unforgettable moments. All teams are set, the stage is ready, almost all teams have strengthened their ranks. More than just football, the Ntare Lions League is a celebration of networking, brotherhood and tradition. Whether you're playing on the pitch or cheering from the stands, everyone is part of this exciting journey. But the biggest question remains Who will be crowned champions in Fort Portal? Bring on #NLLXV #NLLXV #Road2FortPortal

Ntare Lions League

13,445 views • 1 year ago

Attention to all globers…. There’s a difference between the theory of gravity which is derived from the theory of relativity and the law of gravitation which describes gravitation as a constant in mathematical form. Gravity in the form of spacetime warping is a theory, this is why GR is titled ”The Theory Of Relativity” and not, “The Fact Of Relativity”. This takes nothing nothing more then a simple google search to figure out, just google it.
0:37

Attention to all globers…. There’s a difference between the theory of gravity which is derived from the theory of relativity and the law of gravitation which describes gravitation as a constant in mathematical form. Gravity in the form of spacetime warping is a theory, this is why GR is titled ”The Theory Of Relativity” and not, “The Fact Of Relativity”. This takes nothing nothing more then a simple google search to figure out, just google it.

Kaleb The Socrat

15,375 views • 2 years ago

Google just admitted it can't build data centers fast enough, so it's planning on baking its AI model directly into the chip instead (Save this). Google's new Frozen v2 chip permanently embeds parts of Gemini's architecture into the silicon itself, cutting down on the calculations and data movement needed to answer a query. Engineers estimate it could process 6 to 10 times more tokens per unit of power than Google's current TPUs. The real story is why Google is building this in the first place because Frozen v2 is meant to ease a severe internal compute crunch that's caused friction between teams at Google. It reportedly pushed Google Cloud to turn away outside business because it simply doesn't have enough spare capacity to go around. If Google, one of the largest chipmakers in the world, is short enough on compute to turn away paying Cloud customers, that's confirmation this shortage isn't a scaling problem unique to smaller players, it's systemic across the entire industry. This ties into a much bigger power struggle happening right now. More AI companies are trying to cut their reliance on Nvidia by building their own chips. OpenAI rolled out a custom chip called Jalapeno alongside Broadcom last month and Anthropic is now partnering with Samsung on something similar. The reasoning is pretty simple, Nvidia effectively acts as landlord for every hyperscaler out there, and the rent isn't cheap. Nvidia hardware can account for anywhere from 20% to 60% of total AI infrastructure spend, and once a company is tied into its ecosystem, every hardware refresh forces another costly one. That's exactly why Google built TPUs and Amazon built Trainium, both trying to protect their own margins for shareholders. Bullish on Marvell + Broadcom who makes these custom chips and follow me Melvin for more infrastructure plays and check out the link below for more!
2:52

Google just admitted it can't build data centers fast enough, so it's planning on baking its AI model directly into the chip instead (Save this). Google's new Frozen v2 chip permanently embeds parts of Gemini's architecture into the silicon itself, cutting down on the calculations and data movement needed to answer a query. Engineers estimate it could process 6 to 10 times more tokens per unit of power than Google's current TPUs. The real story is why Google is building this in the first place because Frozen v2 is meant to ease a severe internal compute crunch that's caused friction between teams at Google. It reportedly pushed Google Cloud to turn away outside business because it simply doesn't have enough spare capacity to go around. If Google, one of the largest chipmakers in the world, is short enough on compute to turn away paying Cloud customers, that's confirmation this shortage isn't a scaling problem unique to smaller players, it's systemic across the entire industry. This ties into a much bigger power struggle happening right now. More AI companies are trying to cut their reliance on Nvidia by building their own chips. OpenAI rolled out a custom chip called Jalapeno alongside Broadcom last month and Anthropic is now partnering with Samsung on something similar. The reasoning is pretty simple, Nvidia effectively acts as landlord for every hyperscaler out there, and the rent isn't cheap. Nvidia hardware can account for anywhere from 20% to 60% of total AI infrastructure spend, and once a company is tied into its ecosystem, every hardware refresh forces another costly one. That's exactly why Google built TPUs and Amazon built Trainium, both trying to protect their own margins for shareholders. Bullish on Marvell + Broadcom who makes these custom chips and follow me Melvin for more infrastructure plays and check out the link below for more!

Melvin

63,457 views • 26 days ago

.Bridgit Mendler on the founding thesis of Northwood. "We really were recognizing the same problem, that networking for the space economy is really what drives its value. And there's a significant bottleneck, and that bottleneck actually is not in space. It's actually on Earth. The way that you actually solve that networking problem for space is through introducing a model that gets you to the scale that you need, not just for one individual company or a mission, but for the industry overall." Full "Valley of Depth" episode in comments 👇
2:30

.Bridgit Mendler on the founding thesis of Northwood. "We really were recognizing the same problem, that networking for the space economy is really what drives its value. And there's a significant bottleneck, and that bottleneck actually is not in space. It's actually on Earth. The way that you actually solve that networking problem for space is through introducing a model that gets you to the scale that you need, not just for one individual company or a mission, but for the industry overall." Full "Valley of Depth" episode in comments 👇

Mo Islam

137,904 views • 1 month ago

$NVDA networking is now running at a $60B+ annualized pace, growing ~200% YoY and nearly the size of $AVGO total 2025 revenue. As AI clusters scale, the bottleneck becomes how fast GPUs, CPUs, memory and storage move data together because even the best chips underperform if the network cannot keep up. This is why Nvidia is selling the full AI factory stack.
1:25

$NVDA networking is now running at a $60B+ annualized pace, growing ~200% YoY and nearly the size of $AVGO total 2025 revenue. As AI clusters scale, the bottleneck becomes how fast GPUs, CPUs, memory and storage move data together because even the best chips underperform if the network cannot keep up. This is why Nvidia is selling the full AI factory stack.

Shay Boloor

61,104 views • 2 months ago

A new, faster internet is here. DoubleZero’s high-performance global network is now live on mainnet-beta powered by 2Z. Welcome to the world of high-performance networking.
1:26

A new, faster internet is here. DoubleZero’s high-performance global network is now live on mainnet-beta powered by 2Z. Welcome to the world of high-performance networking.

DoubleZero

945,450 views • 10 months ago

Coatue CIO of Public Investments Jaimin Rangwalla explains why the market is rewarding AI infrastructure suppliers ..and punishing the hyperscalers buying the shortage: YTD Return: Sellers ~107% Buyers ~4% "Amazon & Google are in an unusual camp" Full list: Sellers of Products in Shortage: - Broadcom - Coherent - Marvell Technology - GE Vernova - Lumentum - Samsung Electronics - Micron Technology - NVIDIA - TSMC - SK hynix - Western Digital - Seagate Technology - SanDisk Buyers of Products in Shortage: - Amazon - Google - Meta Platforms - Oracle - Microsoft COATUE
3:25

Coatue CIO of Public Investments Jaimin Rangwalla explains why the market is rewarding AI infrastructure suppliers ..and punishing the hyperscalers buying the shortage: YTD Return: Sellers ~107% Buyers ~4% "Amazon & Google are in an unusual camp" Full list: Sellers of Products in Shortage: - Broadcom - Coherent - Marvell Technology - GE Vernova - Lumentum - Samsung Electronics - Micron Technology - NVIDIA - TSMC - SK hynix - Western Digital - Seagate Technology - SanDisk Buyers of Products in Shortage: - Amazon - Google - Meta Platforms - Oracle - Microsoft COATUE

Molly O’Shea

116,648 views • 3 months ago

What a week at #GoogleCloudNext in Las Vegas ✨ From sessions to showfloor demos to networking events, we were excited to partner with Google Cloud to advance how the AI ecosystem can build, deploy, and innovate in the cloud. Looking forward to working together to advance AI agents, physical AI, and beyond.
0:29

What a week at #GoogleCloudNext in Las Vegas ✨ From sessions to showfloor demos to networking events, we were excited to partner with Google Cloud to advance how the AI ecosystem can build, deploy, and innovate in the cloud. Looking forward to working together to advance AI agents, physical AI, and beyond.

NVIDIA

32,762 views • 3 months ago

Why networking makes you richer than talent! I interviewed a Private equity MOGUL in Palm Beach, and I asked: -Where the richest people invest their money in today’s world -What he looks for when buying a company And how important it is to follow up in business.
2:09

Why networking makes you richer than talent! I interviewed a Private equity MOGUL in Palm Beach, and I asked: -Where the richest people invest their money in today’s world -What he looks for when buying a company And how important it is to follow up in business.

The School of Hard Knocks

12,732 views • 1 month ago

Marathon's Gokul Rajaram: You can't vibe code DoorDash. Software alone can’t replicate a managed marketplace. That’s what makes them so durable. You can build the app and drive orders, but when something goes wrong in the physical world, who fixes it? In 2016, Google launched Google Food Ordering. At one point, it drove multiples of traffic to restaurants compared to DoorDash. But retention was basically zero because there was no one to contact when orders failed. You couldn't call Google. The restaurant wouldn't take responsibility. And today, Google Food Ordering no longer exists. "A managed marketplace in the physical world is the most durable thing. It is not a piece of software."
3:27

Marathon's Gokul Rajaram: You can't vibe code DoorDash. Software alone can’t replicate a managed marketplace. That’s what makes them so durable. You can build the app and drive orders, but when something goes wrong in the physical world, who fixes it? In 2016, Google launched Google Food Ordering. At one point, it drove multiples of traffic to restaurants compared to DoorDash. But retention was basically zero because there was no one to contact when orders failed. You couldn't call Google. The restaurant wouldn't take responsibility. And today, Google Food Ordering no longer exists. "A managed marketplace in the physical world is the most durable thing. It is not a piece of software."

TBPN

52,383 views • 5 months ago

SNEAKO goes OFF on Fresh for networking with Z*onist and not defending Myron 😳 "Is that a crucifix? You know they b*mbed a Church last week." "He's sh*tting on your best friend, and you're networking? What the f*ck is this!"
4:40

SNEAKO goes OFF on Fresh for networking with Z*onist and not defending Myron 😳 "Is that a crucifix? You know they b*mbed a Church last week." "He's sh*tting on your best friend, and you're networking? What the f*ck is this!"

Sneako Daily

40,374 views • 1 year ago