Intel Corporation is introducing a new iteration of its AI chip, aiming to compete with Nvidia Corporation in one of the semiconductor industry’s fastest-growing segments. The updated processor, dubbed Gaudi 3, is slated for wide release in the third quarter, as announced by Intel during an early Tuesday company event. The chip enhances performance in two critical areas: facilitating AI system training, which involves inundating them with data, and executing the finalized software.
The escalating demand for AI services has prompted tech firms to vie for these accelerator chips, with Nvidia reaping most of the rewards. Previous versions of Gaudi fell short of achieving the market share gains Intel had anticipated, according to CEO Pat Gelsinger. However, he anticipates that the new model will have a more significant impact. Intel’s shares rebounded by 1.4 percent to $38.50 as of 11:58 a.m. in New York on Tuesday, reversing an earlier decline. They had witnessed a 24 per cent drop this year through Monday’s close.
What does this AI chip mean for Nvidia?
Challenging Nvidia’s dominance won’t be straightforward. The overwhelming success of Nvidia’s H100 accelerator fueled a revenue surge, propelling its market valuation past $2 trillion. Now, Nvidia seeks to extend its lead with the recently announced Blackwell chip platform, with systems based on this product set for release later this year, as announced by the company in March. Intel asserts that Gaudi 3 will outperform the H100, offering 1.7 times faster training for certain AI models and 1.5 times better software execution. The product is expected to roughly match Nvidia’s newer H200, performing slightly better in some aspects and slightly worse in others.
Santa Clara-based Intel stated it cannot provide comparisons with Nvidia’s forthcoming Blackwell line until these products are publicly available. Meanwhile, Intel’s longstanding rival, Advanced Micro Devices Inc., is also entering the fray, unveiling its MI300 accelerator lineup in December. Gelsinger of Intel has emphasized that the goal isn’t merely to catch up with Nvidia but to harness the potential of AI for the industry, especially as the technology permeates beyond the current confines of data centres owned by companies like Microsoft Corp. and Alphabet Inc.’s Google.
With Intel’s fourth-quarter results announced in late January, Gelsinger highlighted plans to ramp up Gaudi supply to meet escalating demand. The company envisions a robust pipeline for 2024, with projections exceeding $2 billion and trending upward. Intel predicts a substantial expansion in the corporate spending on generative AI equipment, from $40 billion in 2024 to $151 billion in 2027, based on market research. Nonetheless, Nvidia retains a significant lead, boasting data center revenue exceeding $47 billion in the preceding 12 months, a figure estimated to surpass $95 billion in the current fiscal year, according to analysts’ forecasts.
See also: Elon Musk Foresees AI Surpassing Human Intelligence By 2026
