The race to invest in high-end AI chips | DW Business

They are tiny, they are mighty - and they power the modern world. Computer chips run everything from our smartphones to our dishwashers. Over the years they've become ever more powerful - and today, it is their potential to steer Artificial Intelligence that is generating the most attention. But there's a problem. Producing these super advanced chips is extremely expensive. And that is why Sam Altman - the founder of Open AI - that's the company behind Chat GPT - is calling for a $7 trillion investment in the industry. DW Business speaks to Chris Miller, Associate Professor of International History at The Fletcher School and the author of a book called "Chip War."

Follow DW on social media:

#AI #chips

[jadecarson]

Intel and AMD will definitely have their share of the market. TSMC is at max capacity and investing in other semiconductor companies will be an absolute power move, I keep increasing my shares manageably. Different chips are good at different things and Nvidia has been very specialised, which leaves other aspects of Al open.

[ScentlessSun]

To me she always looks like she just baked you some cookies and she can’t wait for you to try one.

[tamikovacs]

I still prefer the ones made from potatoes

[jamesrock]

i just want to run video games at 1080p on medium settings

[jcguy]

$7 Trillion chip question? You are forgetting about the (PHC) planetary hadron collider, the $10 quintillion question. It’s a particle collider that loops around the Earth at the equator line for 40, 075km.

[wisecynoz]

Okay but that doesn't change the fact that there are 48 million kangaroos in Australia and 3, 457, 380 people in Uruguay. So if the kangaroos decide to invade Uruguay, each Uruguayan must fight 14 kangaroos.

[paulgermano]

I'm glad I purchased shares of NVIDIA and AMD a while back.

[Snookyboo]

She suggests slowing down so regulators can catch up, but that would imply that these companies care about regulations. If anything, they most likely prefer when Regulators are not able to keep up so they can continue doing things unregulated

[raz-sensei]

Moore’s Law says that roughly every 2 years, the transistor count in an IC will double - this doesn’t mean that it’s also necessarily double performant. Please don’t be rewriting Moore’s Law.

[Daniel-xaogjeyh]

I didn't know there are 'computer' chips in the dishwasher, thank you DW!

[Continentalmunkey]

0:45 Penguins manage to live without AI

[peterbedford]

One of Sams co-founders has stated, recently, that computing power is not a bottleneck in AI.

[wildfotoz]

Moore's Law doesn't really apply here. That's related to the amount of transistors that can be put on the chip but with GPU's there's a lot of other ways to increase the performance. The performance increase from the A100 to the H100 was something like 9x and from, the H100 to the new GH100 or what I think they've renamed to H200 is something like 12x. That's probably why Jensen Huag is saying that it's not going to take 7T because the performance increase over each version is much more than just 2x.

[denuncimesmo]

As an admirer of the work of Deutsche Welle (DW), I have recently noticed a growing inclination towards what appears to be "Chinaphobia" in its coverage. Although DW has been a reliable source, I notice a tendency to negatively highlight issues related to China, sometimes exceeding the bounds of impartiality. I hope DW maintains its commitment to objectivity, avoiding unfair generalizations and stereotypes while continuing to offer a balanced and informative perspective on global issues.

[vt]

I first heard "..and that is why some old man, the founder of OpenAI..."

[Pifagorass]

Optrinics/analog computing will run AI in couple decades. But yes the suboptimal digital computing is now much more popular.

[Continentalmunkey]

2:56 ; silicon industry slice within CUDA(mvidia) not cuba or scuba and probably first throttle-hold on hitting €1trillion per millisecond

[youcantata]

$7, 000 B is huge money. Sales revenue of semiconductor industry woldwide is about $6-700B in 2022. Estimated investment is $130-150B a year. I am skeptical that it will double in the next 10 years. An $2, 000 B in the next 20 years (or $100 B a year) will be a more reasonable investment estimate in AI chip at maximum.

[dtesdff]

Kate reporting always feels exciting and enthusiastic. put smiles on my face too.

[Gepprocurement]

As technology continues to evolve and artificial intelligence becomes increasingly integral across sectors like healthcare, finance and transportation, the urgency of investing in chip development has never been more apparent.

Our AI-powered solutions offer unprecedented step changes in efficiency, accuracy and innovation throughout the source-to-pay process. This has revolutionized how businesses optimize their procurement operations and drive value in their supply chains.