How Intel Missed Out on AI Supremacy
Once the crown jewel of the computer age, U.S. chip giant Intel has been overshadowed in the AI era, Reuters reported. Over recent years, Intel's rival Nvidia, now valued at $2.6 trillion, pivoted from video game graphics to AI chips that power leading generative AI systems like OpenAI’s GPT-4 and Meta's Llama. AMD, valued at $218 billion, has also surged ahead.
Consider what might have been: Seven years ago, Intel had the opportunity to invest in OpenAI, then a little-known nonprofit focused on generative artificial intelligence. Discussions in 2017 and 2018 considered Intel purchasing a 15% stake for $1 billion, with a further 15% contingent on providing hardware at cost. Several sources familiar with the matter revealed that then-CEO Bob Swan decided against it, doubting the near-term market viability of generative AI models. An Intel spokesperson didn’t address questions regarding the potential deal, and neither Swan nor OpenAI commented.
OpenAI sought Intel’s investment to diversify from Nvidia’s chips and develop its infrastructure. However, Intel's data center unit resisted providing products at cost, leading to the deal’s collapse.
Intel's judgment, which has not been publicly revealed before, is one of a series of misjudgments that have contributed to the company's decline in AI prowess, once a significant player in computer chips. Recently, disappointing second-quarter earnings led to Intel's stock dropping by over 25% on its worst trading day since 1974. For the first time in 30 years, the company's value has fallen below $100 billion.
Facing obstacles, Intel strives to present a breakthrough AI chip. CEO Pat Gelsinger highlighted progress on launching the third-generation Gaudi AI chip this quarter, which he claims will outperform competitors, and the upcoming Falcon Shores AI chipset for 2025.
Former executives and industry analysts cite a more profound issue: successive strategic errors over the last decade have eroded Intel's leadership in AI. Instead of timely embracing GPU technology, which excels in parallel data processing for AI, Intel focused on traditional CPUs.
In 2016, former CEO Brian Krzanich acquired Nervana Systems for $408 million, whose technology resembled Google’s TPU chip tailored for AI model training. Nevertheless, Intel abandoned Nervana for its second AI chip acquisition, Habana Labs, for $2 billion in 2019.
Despite such efforts, none have effectively challenged Nvidia or AMD. Expected to generate $13.89 billion from its entire data center business, Intel’s AI chip sales pale before Nvidia's projected $105.9 billion in revenue. Intel faces the pressing challenge of re-establishing its competitive edge.
Earlier, SSP reported that Apple's scheduled release of AI features will be delayed.