Chinese AI startup DeepSeek is threatening Nvidia’s AI dominance 

The hype around the startup’s new model, R1, which is said to rival OpenAI’s o1 in its reasoning capabilities, spiked over the weekend. The model rocketed to the top of Apple’s App Store—beating out OpenAI’s ChatGPT—and spurred a selloff in AI-linked equities.

Some people are worrying that the U.S.—which long assumed it was front-running AI development—is actually behind China. “If China is catching up quickly to the U.S. in the AI race, then the economics of AI will be turned on its head,” Kathleen Brooks, research director at XTB, told her clients. Marc Andreessen, the legendary tech investor, posted on X, “Deepseek R1 is AI’s Sputnik moment.”

In addition, the news that DeepSeek appears to have built its newest model using lower capability and less expensive chips is putting pressure on Nvidia, with some concerned that other Big Tech firms could scale back their demand for Nvidia’s more advanced offerings.

“Investors are worried that DeepSeek’s ability to work with less powerful AI chips may damage Nvidia’s dominance in the AI hardware space, particularly as its valuation has been heavily reliant on AI demand,” Kate Leaman, chief market analyst at AvaTrade, told Fortune.

So far, Nvidia has been one of the biggest beneficiaries of the AI boom. It has continuously beat Wall Street’s expectations and dodged anxieties about an AI bubble.  

The company controls the dominant share of the AI chip market and is the front-runner for developing powerful graphics processing units (GPUs) that are needed for training and maintaining AI models. A limited supply of the powerful chips has left Big Tech vying for access and given the company a massive stock boost in return.

However, if organizations begin thinking about lower-cost solutions or reconsidering their reliance on high-performance chips, this could have a ripple impact on the chip market at large, Leaman said.

DeepSeek’s impact on Big Tech

DeepSeek could also challenge the investment case Big Tech has been making for AI development: that more spending leads to better outcomes for AI models.

DeepSeek’s new AI model is mounting a serious challenge to OpenAI’s flagship reasoning model, o1, but was, by some estimates, developed at 3% to 5% of the cost.

This has sparked concern throughout Silicon Valley. Meta has responded by setting up four war rooms to analyze DeepSeek in an attempt to understand how the company—backed by hedge fund High-Flyer—cut training costs and what data it may have used, The Information reported.

Long-term impact

The long-term impact of DeepSeek is still up for debate as there is still major demand in the U.S. for advanced AI chips. Meta recently announced it was pushing its capital expenditure to $65 billion to focus on AI infrastructure while OpenAI is co-leading the $500 billion Stargate project.

“In the short term, the DeepSeek announcements could have more implications than in the long term,” Javier Correonero, equity analyst at Morningstar, told Fortune.

“If big tech firms would cut their capex outlooks, this would flow upstream in the industry, affecting all players and near-term growth rates across the whole space. However, there is also a bullish angle, as higher compute efficiency could lead to lower cost and even broader adoption of AI in the long term, something known as the Jevons paradox,” he said.

Wedbush’s Dan Ives dismissed concerns over the selloff in a note, instead calling it a “golden buying” opportunity for Nvidia.

“At the end of the day there is only one chip company in the world launching autonomous, robotics, and broader AI use cases and that is Nvidia. Launching a competitive LLM model for consumer use cases is one thing…launching broader AI infrastructure is a whole other ball game and nothing with DeepSeek makes us believe anything different,” he said.