For US chip giant Intel, the favorite of the computer era before collapsing in the more difficult era of the AI era, things may be different.
About seven years ago, the company had the opportunity to buy shares in OpenAI, then a non-profit research organization working in a field called generative artificial intelligence, four people with direct knowledge of the discussions told Reuters.
Over several months in 2017 and 2018, executives at both companies discussed various options, including Intel buying a 15% stake for $1 billion in cash, the three people said. They also discussed Intel taking an additional 15% stake in OpenAI if it makes hardware for the startup at cost, the two people said.
Intel ultimately decided against the deal, in part because CEO Bob Swan didn’t think the generative AI model would be marketed in the future and thus repay the chipmaker’s investment, according to three sources, who all requested anonymity to discuss confidentiality. case.
(Uncover the complexity of our digital world on The Interface podcast, where business leaders and scientists share the insights that shape tomorrow’s innovations. The Interface is also available on YouTube, Apple Podcasts and Spotify.)
OpenAI is interested in the investment from Intel because it will reduce its reliance on Nvidia chips and allow the startup to build its own infrastructure, the two people said. The deal also fell through because Intel’s data center unit didn’t want to make the product cost-effective, the people added.
An Intel spokeswoman did not respond to questions about the potential deal. Swan did not respond to a request for comment and OpenAI declined to comment.
Intel’s decision not to invest in OpenAI, which later launched ChatGPT in 2022 and is now reportedly worth around $80 billion, was previously unannounced.
It was one of a series of strategic mishaps that saw the company, which was on the cutting edge of computer chips in the 1990s and 2000s, stumble into the AI era, according to a Reuters interview with nine people familiar with the matter including a former Intel executive. executives and industry experts.
Last week, Intel’s second-quarter earnings sent its stock price down by more than a quarter in its worst trading day since 1974.
For the first time in 30 years, a tech company is worth less than $100 billion. The former market kingpin — whose marketing slogan “Intel Inside” long represented the gold standard of quality — is still struggling to get blockbuster AI chip products to market.
Intel is now being overtaken by its $2.6 trillion rival Nvidia, which has moved from video game graphics to the AI chips needed to build, train and operate large generative AI systems like OpenAI’s GPT4 model and Meta Platforms’ Llama. Intel is also behind AMD by $218 billion.
Asked about AI progress, an Intel spokesman cited recent comments from CEO Pat Gelsinger, who said the company’s third-generation Gaudi AI chip, which it plans to launch in the third quarter of this year, will outpace its competitors.
Gelsinger said the company has “20-plus” customers for the second- and third-generation Gaudi and the next-generation Falcon Shores AI chips will be launched by the end of 2025.
“We are nearly completing a history of design and process technology innovation, and we are encouraged by the product pipeline we are building to capture more share of the AI market going forward,” the spokesperson told Reuters.
Game chips sweep the AI
In front of OpenAI, Microsoft entered to make investments in 2019, pushing itself forward in the era of AI triggered by the 2022 release of ChatGPT and the excitement of activities among the world’s largest companies to deploy AI.
Although in hindsight, the prospective deal is a missed opportunity for Intel, the company has been losing the battle for AI supremacy for more than a decade, according to former executives and industry experts interviewed.
“Intel failed in AI because it didn’t deliver a cohesive product strategy to customers,” said Dylan Patel, founder of semiconductor research group SemiAnalysis.
For more than two decades, Intel has believed that CPUs, or central processing units, like those that power desktop and laptop computers, can more effectively handle the processing tasks needed to build and run AI models, according to four former Intel executives with direct knowledge. of the company plan.
Intel engineers see the graphics processing unit (GPU) video game chip architecture, used by rivals Nvidia and Advanced Micro Devices, as relatively “ugly,” one of the people said.
However, in the mid-2000s, researchers discovered that gaming chips were more efficient than CPUs at handling the intensive data crunching required to build and train large AI models. Because GPUs are designed for game graphics, they can perform an enormous number of calculations in parallel.
Nvidia engineers have spent years tweaking their GPU architecture to make it work for AI, and creating the software needed to harness those capabilities.
“When it comes to AI … Intel just didn’t have the right processor at the right time,” said Lou Miscioscia, an analyst at Japanese investment bank Daiwa.
Nervana and Habana
Since 2010, Intel has made at least four attempts to produce viable AI chips, including acquiring two startups and at least two major homegrown efforts. Neither Nvidia nor AMD is in trouble in the fast-growing and lucrative market, according to three people with direct knowledge of the company’s internal workings.
Intel’s entire data center business is expected to generate $13.89 billion in sales this year – including the company’s AI chips but also many other designs – while analysts expect Nvidia to generate $105.9 billion in data center revenue.
In 2016, Intel CEO Brian Krzanich sought to buy his way into the AI business by acquiring Nervana Systems for $408 million. Intel executives were attracted to Nervana’s technology, which is similar to tensor processing unit (TPU) chips made by Google, according to two former executives.
TPU – designed specifically for building, or training, large generative AI models – ditches conventional GPU features useful for video games and focuses exclusively on optimizing AI calculations.
Nervana enjoyed some success with customers including Meta Platforms for processors, though not enough to prevent Intel from switching horses and abandoning the project.
In 2019, Intel bought a second chip startup, Habana Labs, for $2 billion before shutting down Nervana’s efforts in 2020.
Krzanich did not respond to a request for comment for this article.
