Nvidia to spend hundreds of billions on US supply chain over next 4 years, says chief

[XYang2023]

--Nvidia plans to spend hundreds of billions of dollars on chips and other electronics manufactured in the U.S. over the next four years, as it shifts its supply chain from Asia amid President Trump's trade policies, Chief Executive Jensen Huang told the Financial Times in an interview.

--Nvidia will procure "probably half a trillion dollars worth of electronics in total" over the next four years, Huang told the FT, adding that the chipmaker is now able to manufacture its latest Blackwell systems in the U.S. through suppliers, such as TSMC and Foxconn Technology.

--Huang denied reports that Nvidia was involved in discussions to form a consortium with the likes of TSMC to invest in Intel, the FT said.

--Huang said that the U.S.-led efforts to constrain Huawei's advances in producing AI chips have been "done poorly," given Huawei's continued success "in every market they've engaged," the FT quoted Huang as saying.

Nvidia to spend hundreds of billions on US supply chain over next 4 years, says chief

Jensen Huang’s chip pledge follows onshoring plans by other tech companies as Donald Trump’s trade policies ripple through industry

www.ft.com

 

[XYang2023]

Referring to IFS:

“We evaluate their foundry technology on a regular basis, and we are ongoing in doing that”

 

[Daniel Nenni]

Referring to IFS:

“We evaluate their foundry technology on a regular basis, and we are ongoing in doing that”

7 of the top 10 TSMC customers are evaluating IFS. Hopefully all 7 put chips through 18A.

Huang, who in January met President Donald Trump, lauded the new administration’s support for the developing artificial intelligence sector and told the FT it was “a phenomenal result for AI in the US.” He also praised TSMC — the sole manufacturer for Nvidia’s most advanced AI chips — for expanding its investment in the US.

Samsung will never be able to match the TSMC and Intel investments in US manufacturing so it really will be a two horse race.

By the way, Lip-Bu knows Samsung Foundry as well from Cadence. He will change the foundry landscape, absolutely.

 

[kevin01]

Referring to IFS:

“We evaluate their foundry technology on a regular basis, and we are ongoing in doing that”

“I have every confidence that Intel has the ability to do it,” said Huang, referring to Intel’s ability to be competitive in advanced chip technologies.

He added that the “success and welfare of Intel” was important. “But it takes a while to convince yourself and each other that a new supply chain ought to get built up.”

 

[kevin01]

“I have every confidence that Intel has the ability to do it,” said Huang, referring to Intel’s ability to be competitive in advanced chip technologies.

He added that the “success and welfare of Intel” was important. “But it takes a while to convince yourself and each other that a new supply chain ought to get built up.”

Reading between the lines, seems to me that, in absence of some external forces (like even a small tariff), it is kinda hard to see NVDA adopting IFS. Maintaining the status quo is the most comfortable thing to do, after all.

 

[Daniel Nenni]

“I have every confidence that Intel has the ability to do it,” said Huang, referring to Intel’s ability to be competitive in advanced chip technologies.

He added that the “success and welfare of Intel” was important. “But it takes a while to convince yourself and each other that a new supply chain ought to get built up.”

Very wise words. It is all about reducing the risk of failure. Right now TSMC is the safe foundry bet. Hopefully Lip-Bu can change that.

 

[Artificer60]

Referring to IFS:


7 of the top 10 TSMC customers are evaluating IFS. Hopefully all 7 put chips through 18A.

Huang, who in January met President Donald Trump, lauded the new administration’s support for the developing artificial intelligence sector and told the FT it was “a phenomenal result for AI in the US.” He also praised TSMC — the sole manufacturer for Nvidia’s most advanced AI chips — for expanding its investment in the US.

Samsung will never be able to match the TSMC and Intel investments in US manufacturing so it really will be a two horse race.

By the way, Lip-Bu knows Samsung Foundry as well from Cadence. He will change the foundry landscape, absolutely.

Intel really needs the customer feedback from these engagements to get up to speed as a foundry even if they don't land many early contracts. From everything I've read Lip-Bu Tan will be actively seeking that feedback and acting on it to improve IFS's ability to offer competitive services.

 

[Daniel Nenni]

Intel really needs the customer feedback from these engagements to get up to speed as a foundry even if they don't land many early contracts. From everything I've read Lip-Bu Tan will be actively seeking that feedback and acting on it to improve IFS's ability to offer competitive services.

I think Lip-Bu reduces the risk of using IFS. More so than Pat Gelsinger who was overly optimistic. It is all about trust in the foundry business. TSMC sets expectations then exceeds them. It is the Morris Chang way.

 

[Daniel Nenni]

Pat Gelsinger at GTC:

"The CPU was the king of the hill [in the mid-2000s], and I applaud Jensen for his tenacity of just saying, 'No, I am not trying to build one of those; I am trying to deliver against the workload starting in graphics," said Gelsinger. "You know, it became this broader view. And then he got lucky with AI, and one time I was debating with him, he said, 'No, I got really lucky with AI workload because it just demanded that type of architecture.' That is where the center of application development is [right now].

"I had a project that was well known in the industry called Larrabee and which was trying to bridge the programmability of the CPU with a throughput oriented architecture [of a GPU], and I think had Intel stay on that path, you know, the future could have been different," said Gelsinger during a webcast. "I give Jensen a lot of credit [as] he just stayed true to that throughput computing or accelerated [vision]."

 

[blueone]

Pat Gelsinger at GTC:

"The CPU was the king of the hill [in the mid-2000s], and I applaud Jensen for his tenacity of just saying, 'No, I am not trying to build one of those; I am trying to deliver against the workload starting in graphics," said Gelsinger. "You know, it became this broader view. And then he got lucky with AI, and one time I was debating with him, he said, 'No, I got really lucky with AI workload because it just demanded that type of architecture.' That is where the center of application development is [right now].

"I had a project that was well known in the industry called Larrabee and which was trying to bridge the programmability of the CPU with a throughput oriented architecture [of a GPU], and I think had Intel stay on that path, you know, the future could have been different," said Gelsinger during a webcast. "I give Jensen a lot of credit [as] he just stayed true to that throughput computing or accelerated [vision]."

Nauseating.

 

[fansink]

Pat Gelsinger at GTC:

"The CPU was the king of the hill [in the mid-2000s], and I applaud Jensen for his tenacity of just saying, 'No, I am not trying to build one of those; I am trying to deliver against the workload starting in graphics," said Gelsinger. "You know, it became this broader view. And then he got lucky with AI, and one time I was debating with him, he said, 'No, I got really lucky with AI workload because it just demanded that type of architecture.' That is where the center of application development is [right now].

"I had a project that was well known in the industry called Larrabee and which was trying to bridge the programmability of the CPU with a throughput oriented architecture [of a GPU], and I think had Intel stay on that path, you know, the future could have been different," said Gelsinger during a webcast. "I give Jensen a lot of credit [as] he just stayed true to that throughput computing or accelerated [vision]."

WOW, Pat has like 20/5 hindsight

 

[dkr1986]

Pat Gelsinger at GTC:

"The CPU was the king of the hill [in the mid-2000s], and I applaud Jensen for his tenacity of just saying, 'No, I am not trying to build one of those; I am trying to deliver against the workload starting in graphics," said Gelsinger. "You know, it became this broader view. And then he got lucky with AI, and one time I was debating with him, he said, 'No, I got really lucky with AI workload because it just demanded that type of architecture.' That is where the center of application development is [right now].

"I had a project that was well known in the industry called Larrabee and which was trying to bridge the programmability of the CPU with a throughput oriented architecture [of a GPU], and I think had Intel stay on that path, you know, the future could have been different," said Gelsinger during a webcast. "I give Jensen a lot of credit [as] he just stayed true to that throughput computing or accelerated [vision]."

"The CPU was the king of the hill [in the mid-2000s], and I applaud Jensen for his tenacity of just saying, 'No, I am not trying to build one of those; I am trying to deliver against the workload starting in graphics," said Gelsinger. "You know, it became this broader view. And then he got lucky with AI, and one time I was debating with him, he said, 'No, I got really lucky with AI workload because it just demanded that type of architecture.' That is where the center of application development is [right now].

Pat Gelsinger, also on the same event - "You know the great Bell comment 'The harder I worked, luckier I got'. And I really believe Nvidia created their lot of their luck, right. In that regard, it was decades of hard, continuous work in a singular direction that allowed them to bend the universe in their direction right now"