OpenAI CEO Sam Altman Wants to Be the King of Chip-Making but Challenges Await – HPCwire
Since 1987 – Covering the Fastest Computers in the World and the People Who Run Them
Since 1987 – Covering the Fastest Computers in the World and the People Who Run Them
January 24, 2024
OpenAI’s CEO Sam Altman survived an attempt to oust him as the AI company’s chief. Now, the maverick wants to establish his own chip-making factories.
Altman’s intent, which was reported by Bloomberg, is to build a network of factories to produce AI chips.
OpenAI may already be designing its own AI chips to pair with its large-language models, such as GPT-4, Reuters reported late last year.
Altman is reportedly talking to SoftBank, based in Japan, and G42, based in Abu Dhabi, to raise funds for the AI chip factories. Microsoft, an investor in OpenAI, is also part of the conversations.
The involvement of G42 is intriguing because of the company’s potential ties to China. The US Department of Commerce is investigating G42, which is building AI supercomputers with chips from suppliers Nvidia and Cerebras Systems, about potentially providing China access to its powerful AI infrastructure. The US has banned exports of the latest AI chips from US companies to China.
Designing a chip makes sense, but establishing chip factories is an unprecedented step. OpenAI rivals with their transformer models, including Google and Meta, have their own AI chips but outsource manufacturing to contract fabricators such as TSMC. Intel is taking on manufacturing to be a second-source manufacturer of such chips.
OpenAI’s plan for its chip design can deliver an advantage, but Altman’s plans to manufacture AI chips are much more ambitious. Typically, models used by OpenAI, Google, and Meta don’t work on hardware out of the box. Like operating systems, the models need to be tuned specifically to work quickly and power-efficiently on AI chips. But Google, Meta, and Microsoft just design chips and do not manufacture them.
Microsoft is using OpenAI’s models for its Bing search on Nvidia’s GPUs and is tuning the models for its upcoming Maia 100 AI accelerator, which was introduced in November. The Maia AI accelerator is targeted at inferencing, which is more related to providing answers to queries from customers. Similarly, Google has built its entire AI infrastructure around its TPUs, which also drives the development of future AI models.
OpenAI’s models also rely on Nvidia GPUs, which dominate the AI market. But like Microsoft, Altman wants to cut its reliance on Nvidia and diversify its chip profile.
But Altman is dipping his feet into semiconductor manufacturing, a highly volatile industry that is embroiled in geopolitical intrigue. Altman also believes that demand for AI chips will exceed supply in the coming years, which is why more manufacturing capacity was necessary, the Bloomberg story said.
Altman is also looking to establish a presence in a market that has had more exits than new entrants over the last few decades.
Altman is emerging as a modern-day Steve Jobs, and his reputation is at a peak. He is credited with bringing AI to the masses with ChatGPT. But he has issues – he was temporarily ousted as OpenAI’s CEO for being unclear about the direction of his AI products. He was reinstated as OpenAI’s CEO after a messy battle with the company’s board that played out in public.
But Jobs was smart enough not to make risky investments. In 2008, Apple bought PA Semiconductor – and the talent to build chips for the iPhone and now Mac – and left the manufacturing to third parties. Investing in a chip factory is the riskiest investment for any tech company.
The semiconductor manufacturing industry is also being weaponized by countries as currency to limit trade. The US has banned the export of AI and advanced chip-making equipment to China to choke the country’s AI strategy.
Altman is looking to succeed where China failed. China responded by investing billions in establishing chip factories of its own but hasn’t been successful yet.
The US election is around the corner, and that policy could change if pro-China president Donald Trump takes over. A political change could open a path for US companies to return chip-making to China at a lower cost.
Altman will need piles of cash – far more than what he is seeking – to establish factories. He will need a smart CFO to avoid overinvestment and measure the sustainability of chip demand. Altman will need to figure out if the current demand for chips is just a blip and how to survive the downturn that will come.
In the 1990s, many chip companies had their manufacturing facilities. However, companies such as AMD and Motorola ultimately divested their manufacturing assets because of the volatility and high maintenance costs. Typically, semiconductor demand surges in up-cycles in which manufacturers ramp up manufacturing and flood the market with chips such as CPUs and memory. The demand declines during the downcycle, in which manufacturers scale down production, sell off excess inventory, and leave factories underutilized.
The market has now narrowed to a handful of manufacturers, including Samsung, Texas Instruments, and GlobalFoundries, which spun off from AMD.
Companies such as Nvidia and Apple have seen success in chips through a fabless model in which internal chip designs are sent off to TSMC for fabrication. Intel failed in its old business model in which its factories were used only to make homegrown chips and is establishing a business model that is similar to TSMC.
Companies previously were conservative in investing in new factories, especially when capacity was underutilized, but things have changed in recent years. The investments for design and manufacturing are higher on more sophisticated nodes, which currently stand at about the 3-nm node and are approaching atomic levels.
Last year, the US passed the CHIPS and Science Act, which opens up billions of dollars for companies to establish or upgrade new factories. TSMC and Intel are expected to be recipients of new factories on US soil.
Intel is also investing $20 billion to establish a new advanced node factory in Columbus and is creatively funding new factories. For example, Intel dealt away a 49% stake in its Arizona fab expansions to asset management company Brookfield Asset Management, which is providing $15 billion in capital for the infrastructure upgrades.
But there are positives to an outsider entering chip manufacturing. If Altman decides to start a chip factory, it could bring a fresh approach that breaks decades of old business practices that TSMC, Intel, and Samsung instilled.
Altman could seek a shorter path to success. UAE’s investment fund Mubadala, which has invested in G42, already has a stake in fab company GlobalFoundries and could bring some of that experience to a deal with Altman. SoftBank could leverage a controlling stake in ARM, which went public last year.
More Off The Wire
Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!
June 27, 2024
With the deployment of generative artificial intelligence (GenAI) happening at a rapid pace, organizations of all sizes are tasked with navigating the challenges around implementation, especially regarding ethics and Read more…
June 27, 2024
Everyone is aware of the inflationary model of the early universe in which the volume of space expands exponentially then slows down. AI-augmented HPC (AHPC for short) has started to expand creating new space in the scie Read more…
June 26, 2024
It’s a truism of data analytics: when it comes to data, more is generally better. But the explosion of AI-powered large language models (LLMs) like ChatGPT and Google Gemini (formerly Bard) challenges this conventional Read more…
June 25, 2024
Last week, the Defense Advanced Research Projects Agency (DARPA) issued an interim progress update from the second phase of its Quantum Benchmark (QB) program. Begun in 2021 the QB effort has the ambitious “goal of rei Read more…
June 24, 2024
Europe officially announced its second exascale system, Alice Recoque, and you can expect to see that name on the Top500 supercomputer list in a few years. Alice Recoque is the new name for a supercomputer with the opera Read more…
June 21, 2024
As AI continues to reach into every domain of life, the question remains as to what kind of software these tools will run on. The choice in software stacks – or collections of software components that work together to Read more…
June 27, 2024
Everyone is aware of the inflationary model of the early universe in which the volume of space expands exponentially then slows down. AI-augmented HPC (AHPC for Read more…
June 25, 2024
Last week, the Defense Advanced Research Projects Agency (DARPA) issued an interim progress update from the second phase of its Quantum Benchmark (QB) program. Read more…
June 21, 2024
As AI continues to reach into every domain of life, the question remains as to what kind of software these tools will run on. The choice in software stacks – Read more…
June 20, 2024
The HPE Discover 2024 conference is currently in full swing, and the keynote address from Hewlett-Packard Enterprise (HPE) CEO Antonio Neri on Tuesday, June 18, Read more…
June 19, 2024
Samsung may have unintentionally revealed its intent to develop a RISC-V CPU, which a presentation slide showed may be used in an AI chip. The company plans to Read more…
June 18, 2024
In his opening keynote at D-Wave’s annual Qubits 2024 user meeting, being held in Boston, yesterday and today, CEO Alan Baratz again made the compelling pitch Read more…
June 17, 2024
The world is currently experiencing two of the largest societal upheavals since the beginning of the Industrial Revolution. One is the rapid improvement and imp Read more…
June 13, 2024
As managing editor of the major global HPC news source, the term “news fire hose” is often mentioned. The analogy is quite correct. In any given week, there are Read more…
May 21, 2024
Atos – via its subsidiary Eviden – is the second major supercomputer maker outside of HPE, while others have largely dropped out. The lack of integrators and Atos’ financial turmoil have the HPC market worried. If Atos goes under, HPE will be the only major option for building large-scale systems. Read more…
October 30, 2023
With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…
May 30, 2024
Consider the GPU. An island of SIMD greatness that makes light work of matrix math. Originally designed to rapidly paint dots on a computer monitor, it was then Read more…
August 17, 2023
The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…
March 18, 2024
Nvidia’s latest and fastest GPU, codenamed Blackwell, is here and will underpin the company’s AI plans this year. The chip offers performance improvements from Read more…
December 11, 2023
Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…
May 15, 2024
The makers of the Aurora supercomputer, which is housed at the Argonne National Laboratory, gave some reasons why the system didn’t make the top spot on the Top Read more…
February 8, 2024
Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…
June 10, 2024
Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…
May 17, 2024
On Tuesday May 14th, Google announced its sixth-generation TPU (tensor processing unit) called Trillium. The chip, essentially a TPU v6, is the company’s l Read more…
April 30, 2024
It’s a long wait for customers hanging on for Intel’s next-generation GPU, Falcon Shores, which will be released in late 2025. “Then we have a rich, a very Read more…
June 3, 2024
In the world of AI, there’s a desperate search for an alternative to Nvidia’s GPUs, and AMD is stepping up to the plate. AMD detailed its updated GPU roadmap, w Read more…
May 7, 2024
We have all thought about it. No one has done it, but now, thanks to HPC, we see what it looks like. Hold on to your feet because NASA has released videos of wh Read more…
October 5, 2023
When discussing GenAI, the term “GPU” almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word “GPU” is assumed to mean “Nvidia” products. (As an aside, the popular Nvidia hardware used in GenAI are not technically… Read more…
March 27, 2024
Pictures of Nvidia’s new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…
May 8, 2024
Chip companies, once seen as engineering pure plays, are now at the center of geopolitical intrigue. Chip manufacturing firms, especially TSMC and Intel, have b Read more…
© 2024 HPCwire. All Rights Reserved. A Tabor Communications Publication
HPCwire is a registered trademark of Tabor Communications, Inc. Use of this site is governed by our Terms of Use and Privacy Policy.
Reproduction in whole or in part in any form or medium without express written permission of Tabor Communications, Inc. is prohibited.
