Research
Interview Synopsis

AI Chips – Trends, Supply Chain & Geopolitics

  • Public Equity
  • TMT
  • North America

We are in the “Wild West” period of AI, with Nvidia announcing at the end of May that its GH200 Grace Hopper Superchip is in full production. “This is really going to get interesting very quickly. I think the thing that may be a bit of a surprise is the speed at which this is going to start to just fan out,” a former VP at Onto Innovation Inc told Third Bridge Forum. 

Over the next five years, the impact of large language models such as ChatGPT (view a special report here) is expected to add USD 4.4tn of value to the global economy, the specialist said, with cloud servers/hyperscalers enjoying the lion’s share of this growth. By 2024, chatbots are estimated to hit USD 142bn worth of revenue generation vs just USD 2.8bn in 2019, they added.

Thirst for capability and complexity “at the right price point” will span multiple segments. Nvidia has not yet indicated a price for its Grace Hopper chips, but estimates vary between USD 80,000 and USD 200,000 for a super chip, the expert said. “Then you look at the DGX supercomputer version of it that they’re going to model and sell, which is easily an eight-figure machine or piece of hardware.” 

The race is certainly on. In terms of competitive dynamics, Nvidia (view Nvidia’s value chain on Maps) currently accounts for 90% of higher-end computing but in five years it could lose 20-30% to hyperscalers, AMD and potentially other entrants, forcing it to lower prices, our Interview revealed.

Meanwhile, Nvidia is linking up with hyperscalers as a hardware supplier as players including Meta, Google and Microsoft develop their own chips in-house. Although there is a partnership aspect, there is also a competition aspect. Indeed, this trend, we were told, “could certainly have an impact on the demand for a company like Nvidia or AMD’s hardware followings”. 

The specialist also emphasised that “there’s no free lunch in terms of the environmental impact of all of this”. Chips are power hungry – a single Grace Hopper will consume 1,000 watts, they said. “Global energy use of semiconductors powering AI is going to become an issue.”

We also heard the US is 3-5 years behind in recognising AI’s significance and implementing actions being put into place today. However, as the US remains a leader in R&D, there is a crucial need to strengthen public-private partnerships and expand access to national AI R&D infrastructures.

Forum is Third Bridge’s analyst-led Interviews product, part of our integrated research offering that also includes: Community, investor-led InterviewsMaps, a value chain product; Primers and Tearsheets, for fundamental information on private companies; and Connections, an expert network service.

Related Transcripts and Research

The information used in compiling this document has been obtained by Third Bridge from experts participating in Forum Interviews. Third Bridge does not warrant the accuracy of the information and has not independently verified it. It should not be regarded as a trade recommendation or form the basis of any investment decision.

For any enquiries, please contact sales@thirdbridge.com