5d ago

Despite Meeting With Nvidia CEO, Trump Sticks With Plan to Tariff Foreign Chips

www.pcmag.com Despite Meeting With Nvidia CEO, Trump Sticks With Plan to Tariff Foreign Chips

Trump says he had a 'good meeting' with Jensen Huang but still plans to tariff foreign-manufactured semiconductors. Tariffs on China and Mexico also go into effect tomorrow.

All that bending over with a leather jacket on and nothing to slow for it.

PC Gaming @lemmy.ca

inclementimmigrant @lemmy.world

5d ago

Despite Meeting With Nvidia CEO, Trump Sticks With Plan to Tariff Foreign Chips

www.pcmag.com /news/despite-meeting-with-nvidia-ceo-trump-sticks-with-plan-to-tariff-foreign

116 12

You're viewing a single thread.

10 comments

Nvidia doesn't care. Tariffs do not affect their profits at all. If anything, it increases their profits.
The reason is that Nvidia has a monopoly on high end gpus and AI hardware. Tariffs can only influence a country if there is a competitor that could undercut them.
It's not like Intel is making 5090's so a 25% tariff would allow Intel to be competitive.
- Remember, the tariff is 25-100%. Let's just say it ends up at 50%. How many gamers will pay 3k for a 5090? It is a mediocre upgrade this generation anyway. Also, both AMD and Intel GPUs are made in Taiwan (TSMC). Nvidia's profits lately have been coming via ai chips. The Chinese just showed us that those aren't needed. Jensen and his cousin at AMD are 2 of the smartest CEOs in the country, they know that tariffs like this will crash demand, give China an opening, and possibly wreck the US economy.
  
  They sold out at $2k. They'd sell out at $3k.
  It doesn't matter that it's a weak release. There is no alternative product.
  Gpus represent only a small part of Nvidia's revenue. They make the bulk of their money on data center products. If OpenAI needs $10B more B200's, they'll open an office in Vancouver and avoid the tariff.
  Nvidia won't be hurt. Corporations won't be hurt. They only people to suffer will be American consumers.
  
  They sold out because there weren't any. UFD tech waited in line at micro center, until they found out how few were available. I believe Steve has a video on calling it a paper launch. The alternative is to buy 4080's and even 3080's, until the price becomes too high on used and older GPUs. We saw that happen during covid. Yes, Nvidia makes all kinds of chips, GPUs and especially high end ones are a very small part. A very large and growing segment of Nvidea is also vehicles, which in America is now tariffed on Mexican and Canadian production an additional 25%. Almost all portable electronics have chips from Taiwan or China in them. We are already in a trade war with China, so now we are staring one with the alternative? OpenAI can move to Vancouver all they want, but I bet the 25% tariffs from there will hurt also. Plus, If AI firms don't need as many of the AI chips, as recently shown, then that market goes down. Google makes their own server chips with cpus produced in... Taiwan.
  These tariffs, if not removed, will crash the US economy and possibly the global economy with it. Good luck everyone! or as POTUS says " Jesus didn't need electricity".
  
  Where are we going to buy TSMC chips from? Samsung and Intel are behind in manufacturing. There is no alternative so American companies will pay the tariff. It isn't a tax on foreign companies. It's a tax on American consumers.
  Consumers will see 25% higher prices. It will crash the economy. But again, it is Americans who pay tariffs. It is Americans who will suffer.
  If a company imports Nvidia chips into Canada they don't pay American tariffs. Tariffs are only paid when those chips are sold into the US. Trump is giving a competitive advantage to Canadian tech companies.
  
  You are correct. I know it is a tax that will be paid by us. These will probably collapse the US economy. I give it 8 months. Then another 6 to start wiping out the rest of the worldwide markets. Maybe two years before we become a warring shithole with nukes.
- Nvidia, and every GPU manufacturer, got quite some competition in regards to AI, with specialized chips.
  E.g. Cerebras - https://www.forbes.com/sites/craigsmith/2024/08/27/cerebras-speeds-ai-by-putting-entire-foundation-model-on-its-giant-chip/
  I don't think, that Nvidia will keep its dominance in that branch for much longer. Especially when specialized chips can be run much cheaper and with less electricity cost.
  Edit: more up to date article with actual working model on their chip:
  https://cerebras.ai/press-release/cerebras-launches-worlds-fastest-deepseek-r1-distill-llama-70b-inference
  
  Never head of Cerebras. shrug
  The world buys Nvidia. 98% of datacenter GPUs are made by Nvidia.
  Source: https://www.hpcwire.com/2024/06/10/nvidia-shipped-3-76-million-data-center-gpus-in-2023-according-to-study/
  I see no reason for that to change. Everybody is using them, knows how to use them and pretty much every AI software in existence runs on them.
  
  Well, the reason to change is, that specifically made AI chips perform better or use less power for the same performance.
  But this is a pretty new field, and because they didn't exist (besides some intel AI sticks I have somewhere laying around), we used GPUs as the next best thing, because their architecture is well layed out for parallel computations.
  So, when now new specific AI chips enter the market, Nvidia will get quite some competition, as their product is still just a GPU, and not actually designed for AI inference.
  Although there have been some GPUs sold as AI specific, as far as I remember they were still the same chip and only didn't sport the outputs - but this is just from the back of my head...
  Edit: couldn't find the article I was looking for, but here are at least some graphs showing the difference:
  https://www.msn.com/en-us/news/technology/deepseek-on-steroids-cerebras-embraces-controversial-chinese-chatgpt-rival-and-promises-57x-faster-inference-speeds/ar-AA1y8g3N
  Which shows much lower latency and higher tokens per second output
  Yeah, they will need to proof that, when customers use them, but it seems a specific chip design is much more suited for the task - who could have guessed ..

10 comments