The recent advancements by Chinese AI startup DeepSeek have sent shockwaves through the tech industry. DeepSeek’s open-source models, which nearly match the performance of leading American AI at a fraction of the cost, have sparked optimism among AI chip companies aiming to challenge Nvidia’s dominance. Andrew Feldman, CEO of Cerebras, a next-gen chip company valued at $4 billion, is celebrating the development.
“We’re sort of rejoicing,” Feldman said. “These are great days. We can’t answer the phones fast enough right now.” He believes that as AI becomes more performant and cost-effective, the market will expand rather than shrink.
Cerebras focuses on making AI usage more efficient through a process called “inference,” which involves running an AI model to allow it to “think” and reason. This aspect of AI has enabled smaller startups to compete effectively with Nvidia, which dominates the training segment of the AI market.
DeepSeek disrupts AI performance costs
Other industry players share Feldman’s optimism. Rodrigo Liang, CEO of SambaNova, valued at $5.1 billion, noted, “DeepSeek flipped the AI script for open source and inference.” Sunny Madra, COO of Groq, reported a spike in signups and usage after integrating DeepSeek’s R1 model into its GroqCloud platform. DeepSeek’s claim of training its V3 model, consisting of 671 billion parameters, for just $5.58 million over two months has been met with skepticism.
However, American rivals remain impressed. Liang summed it up: “We don’t know if it was $6 million or $60 million, but it wasn’t $600 million.”
Despite the recent market impact, Nvidia remains a formidable competitor. CEO Jensen Huang has been emphasizing the company’s strengths in inference, and Nvidia reiterated its capabilities in a statement, noting that “Inference requires significant numbers of Nvidia GPUs and high-performance networking.”
The advancements by DeepSeek are seen as a way to streamline the industry’s focus, leading to more efficient training processes and ultimately more powerful models.
This shift could be advantageous for the broader technology landscape. As Feldman reflects on the changing tides, he adds, “For those of us who are underdogs, it ought to fire you up.”