Stay up to date with free updates
Simply sign up to the Artificial Intelligence myFT Digest and have it delivered straight to your inbox.
Nvidia's rivals are mobilizing to disrupt the company's dominance in the AI chip market, raising hundreds of millions of dollars and launching new products to share in the benefits of a boom in artificial intelligence technology.
Cerebras, d-Matrix and Groq are part of a group of smaller companies aiming to grab a share of the multi-billion dollar AI chip market from Nvidia, which has so far dominated the first wave of investments in graphics processing units (GPUs).
They are riding a wave of expectations that demand for artificial intelligence “inference” – the computing power needed for models such as OpenAI's ChatGPT and Google's Gemini to generate responses to queries – will grow exponentially as chatbots and other generative AI applications grow in popularity.
Nvidia's Hopper GPUs are ideal for the highly resource-intensive task of training top AI models, making them some of the most popular commodities in the world.
Cerebras, d-Matrix and Groq are instead focusing on cheaper, more specialized chips designed to run AI models.
On Tuesday, Cerebras unveiled its new “Cerebras Inference” platform, based on its dinner-plate-sized CS-3 chip. Cerebras claims its solution is 20 times faster at AI inference than Nvidia's current-generation Hopper chip, and at a fraction of the price. Cerebras cites tests conducted by benchmarking analytics provider Artificial Analysis.
“The way to beat the 800-pound gorilla is to bring a much better product to market,” Cerebras Chief Executive Andrew Feldman told the Financial Times. “In my experience, the better product usually wins, and we've won a significant customer[from Nvidia].”
The CS-3 chip eschews the use of separate high-bandwidth memory chips used by Nvidia, instead offering an alternative architecture that embeds memory directly on the chip wafer.
Feldman said memory bandwidth limitations are a fundamental constraint on the inference speed of AI chips: Combining logic and memory onto one large chip produces results that are “orders of magnitude faster,” he said.
d-Matrix, founded by Sid Sheth in 2019, has also launched a new round of funding, less than a year after raising $110 million in a Series B funding round led by Singapore state fund Temasek. Sheth said the company is looking to raise more than $200 million later this year or early next year. d-Matrix is in the early stages of its fundraising process and the final amount is subject to change.
d-Matrix plans to fully launch its chip platform, Corsair, at the end of this year. Sheth said the company is pairing its products with open software such as Triton, which competes with Nvidia's Cuda, a widely used software platform that gives developers the tools to build AI applications and optimize chip performance.
Nvidia's largest customers are in favor of using open software such as Triton: “App developers don't like to be tied to specific tools,” Sheth said, “and on the training side, people are starting to realize that Nvidia has a monopoly on Cuda.”
Groq, another AI inference competitor led by a former founding member of Google's tensor processing unit team, raised $640 million this month from investors led by BlackRock Private Equity Partners, at a valuation of $2.8 billion.
Recommendation
Despite all the excitement in the sector, it will be difficult for semiconductor startups to break into the market, one venture capitalist warned.
Chipmaker Graphcore was acquired by SoftBank last month for just over $600 million, less than the roughly $700 million the company has raised in venture capital since it was founded in 2016, people familiar with the deal said.
Groq and Cerebras were also founded in 2016. “There's an insatiable appetite among public investors to find and back the next Nvidia,” says Peter Hébert, co-founder and managing partner at venture firm Lux Capital. “It's not just about chasing the latest trends. This momentum is also benefiting several VC-funded chip startups that have been working hard for nearly a decade.”
Video: AI: Blessing or curse for humanity? | FT Tech
Source link
