For U.S. chip giant Intel, once the darling of the computer age, the landscape has changed dramatically in the AI era. What could have been a different story unfolded about seven years ago when Intel had the opportunity to invest in OpenAI, a then-nascent non-profit focused on generative artificial intelligence.

During discussions in 2017 and 2018, Intel executives considered various options, including a $1 billion investment for a 15% stake in OpenAI. They even discussed a potential additional stake if Intel provided hardware at cost. However, Intel ultimately decided against the deal. The then-CEO Bob Swan believed that generative AI models would not reach the market soon enough to justify the investment.

(Explore the complexities of our digital world on The Interface podcast, where business leaders and scientists share insights shaping tomorrow’s innovations.)

OpenAI sought Intel’s investment to reduce its reliance on Nvidia’s chips and to build its own infrastructure. However, the deal fell through as Intel’s data center unit was unwilling to produce products at cost.

Intel’s decision not to invest in OpenAI, which later launched the revolutionary ChatGPT in 2022 and is now valued at approximately $80 billion, has remained largely undisclosed until now. This decision is part of a series of strategic missteps that have seen Intel stumble in the AI era, as revealed by interviews with industry experts and former Intel executives.

Recently, Intel’s second-quarter earnings led to a significant drop in its stock price, marking its worst trading day since 1974. For the first time in three decades, Intel’s market value has fallen below $100 billion, a stark contrast to its previous dominance in the chip market.

In the current AI landscape, Intel finds itself overshadowed by Nvidia, valued at $2.6 trillion, which has successfully transitioned from video game graphics to AI chips essential for large generative AI systems. Additionally, AMD, valued at $218 billion, has also surpassed Intel.

When asked about its AI progress, an Intel spokesperson referred to comments made by CEO Pat Gelsinger, who expressed optimism about the upcoming third-generation Gaudi AI chip, expected to outperform competitors.

Gelsinger mentioned that Intel has secured over 20 customers for its second and third-generation Gaudi chips and anticipates launching its next-generation Falcon Shores AI chip in late 2025. He stated, “We are nearing the completion of a historic pace of design and process technology innovation, and we are encouraged by the product pipeline we’re building to capture a greater share of the AI market going forward.”

Gaming Chips and AI

Meanwhile, Microsoft stepped in to invest in OpenAI in 2019, positioning itself at the forefront of the AI revolution spurred by the release of ChatGPT. Although Intel’s missed opportunity with OpenAI is evident in hindsight, the company has been gradually losing ground in the AI sector for over a decade.

Industry experts, including Dylan Patel from SemiAnalysis, noted that Intel’s failure in AI stemmed from a lack of a cohesive product strategy. For years, Intel believed that CPUs were sufficient for AI tasks, dismissing the potential of GPUs, which have proven to be far more efficient for AI model training.

As the AI landscape evolved, Nvidia adapted its GPU architecture for AI applications, while Intel struggled to keep pace. Lou Miscioscia, an analyst at Daiwa, remarked, “When AI hit … Intel just didn’t have the right processor at the right time.”

Nervana and Habana

Since 2010, Intel has made several attempts to develop a competitive AI chip, including acquiring startups and launching internal projects, yet none have succeeded against Nvidia or AMD in the lucrative AI market. Analysts predict Intel’s data center business will generate $13.89 billion this year, while Nvidia is expected to bring in $105.9 billion.

In 2016, Intel CEO Brian Krzanich aimed to enter the AI market by acquiring Nervana Systems for $408 million, attracted by its technology similar to Google’s TPU chip. However, despite some initial success, Intel eventually abandoned the project. In 2019, Intel acquired Habana Labs for $2 billion, but shut down Nervana’s efforts in 2020.

Krzanich did not respond to requests for comments regarding these developments.