- As US tech companies prep for earnings this week, there’s one major topic on analysts’ minds.
- China’s DeepSeek introduced an AI model that’s similarly effective to US ones but at a lower cost.
- Big Tech is expected to address DeepSeek’s potential impact on AI spending during earnings calls.
DeepSeek just made this earnings season a lot more interesting.
A slew of Big Tech names are due to report this week and next. Tesla, Meta, and Microsoft take center stage on January 29, with Apple, Microsoft, and Amazon hot on their heels.
Investors were already scrutinizing the companies, which have been spending multiple billions of dollars on building AI models and infrastructure, for signs about whether the outlay on staff, chips, and data centers is delivering returns.
Now DeepSeek has made justifying that spending to investors and analysts a lot trickier for execs, PR, and IR.
“I expect pretty much everyone to get questions on it,” Morningstar analyst William Kerwin said.
DeepSeek’s new AI model, R1, shook the tech industry this week, wiping more than $1 trillion off US stocks in market cap in one day. It offers a model that seems similarly effective to ones made by US companies. Moreover, the Chinese company said it cost less than $6 million to train.
In comparison, Meta has said it plans to invest over $60 billion in AI infrastructure in 2025. Earlier this month, President Donald Trump announced a $500 billion partnership involving OpenAI, Oracle, and SoftBank to build AI infrastructure in the US.
Markets stabilized a little by Tuesday, but the panic shows how DeepSeek’s impact will be top of mind during earnings calls, analysts told Business Insider.
Investors will want to know if R1 changes companies’ views on future infrastructure spending, Kerwin said.
That could “temper the demand” for chips from Nvidia, Forrester analysts said in a Tuesday note. Tech companies like Meta and Elon Musk’s xAI have been buying and using hundreds of thousands of specialist Nvidia chips for their AI models, helping its stock price rocket. Monday’s tech rout already wiped about 17% off Nvidia’s market cap.
However, having access to cheaper models could pay off in the longer term because of “creation opportunities for many, many more startups and enterprises alike, thereby increasing demand,” per Forrester.
Apple, which reports after the bell on Thursday, stands to gain from a decline in cost for AI training. The iPhone maker took a lower-cost AI strategy than competitors, and it would benefit from integrating cost-effective training for Apple Intelligence, analysts told BI.
Although Apple’s stock gained on Monday, CEO Tim Cook should still expect DeepSeek chatter, analysts said.
Nvidia won’t report earnings until February 26, yet will likely still have to answer DeepSeek questions.
And across the board, “Beyond cost concerns, tech leaders will need to address whether this signals a need to shift toward more efficient AI development approaches that could reshape strategic planning,” Jacob Bourne, analyst at BI sister company EMARKETER, said.
