Microsoft shares edged lower in early trading, while Meta Platforms nudged higher, as investors looked to key December quarter updates from two of the world's biggest tech companies amid the ongoing angst tied to the emergence of China's DeepSeek.
The AI startup, founded in 2023, has claimed to have developed a large-language model (LLM) system that is powering its popular AI chatbot for a fraction of the cost of its U.S. rivals.
Performance benchmarks suggest the R1 model is equal to or exceeds that of Microsoft (MSFT) -backed OpenAI, which spent more than $500 million developing and training its o1 model.
DeepSeek's emergence as a low-cost, high-performance alternative to OpenAI and Meta Platforms (META) Meta AI has triggered a massive recalibration of the spending, profit and revenue assumptions of some of the world's biggest tech companies by global investors, who loped a record $593 billion in market value from chipmaker Nvidia (NVDA) in Monday's tech market meltdown.
That's put both the earnings of Meta, Microsoft and to some extent Tesla (TSLA) at the forefront today, as each of the so-called Magnificent 7 tech companies will publish December quarter earnings after the close of trading and, presumably, answer questions regarding capex plans and AI demand forecasts for the coming year.
"Tonight is a huge night for the tech world," said Wedbush analyst Dan Ives. "The tech bears came out of hibernation this week afterDeepSeek AI shot across the bow that called the entire tech AI Revolution story into question: 'if DeepSeek can build this model for $6 million why is Big Tech spending $400 billion+ of capex this year?'".
Deep questions from DeepSeek
Early indications, however, suggest that could prove challenging, given DeepSeek's relatively late entry to the market and the opacity surrounding the technical aspects of its AI system.
Bloomberg News reported that Microsoft is probing allegations that DeepSeek improperly used some of OpenAI's data, through a process called distillation. That refers to a condition in which as smaller "student" AI model is trained to mimic a larger one at a fraction of the cost and with more speed and efficiency.
Those allegations were echoed by President Donald Trump's AI czar, David Sacks, in an interview last night with Fox News.
Related: Top analyst overhauls Nvidia stock price target as DeepSeek impact lingers
“There’s substantial evidence that what DeepSeek did here is they distilled knowledge out of OpenAI models and I don’t think OpenAI is very happy about this,” he said.
Other reports have suggested that DeepSeek is using high-end chips, made by Nvidia, that are currently blocked from sale to China as a result of export restrictions put in place by former President Joe Biden.
"My understanding is that Deepseek has about 50,000 H100s, which they can't talk about obviously because it is against the export controls that the U.S. has put in place," Scale AI CEO Alexandr Wang told CNBC last week.
"I think it is true that they have more chips that people than other people expect," he added. "But on a go forward basis, they are going to be limited by the chip controls and the export controls that we have in place."
DeepSeek's AI chip demand impact
The broader impact DeepSeek may have on AI chip demand is also difficult to forecast, given that its only become known as a potential alternative over the past two weeks.
ASML, the Dutch tech giant that makes or EUV lithography machines, a key component in high-end chip design and manufacturing, posted stronger-than-expected fourth quarter earnings on Wednesday and said its near-term order book had swelled to $7.4 billion, more than double analysts' estimates.
Investors reacted by taking the stock sharply higher in Amsterdam, but pared some of those gains when amid the realization that the company's order book was built long before news of DeepSeek's innovations became public.
Group CEO Christophe Fouquet told investors that while the "growth in artificial intelligence is the key driver for growth in our industry", he isn't worried that the low-cost innovations unveiled by DeepSeek will have a negative impact on demand.
Related: Analyst revisits Microsoft stock price target as DeepSeek tests OpenAI
“For AI to be everywhere, we need to see major progress on costs and power consumption," Fouquet said. "Lowering costs also leads to more volume. So I will say any technology, whatever it is, that will contribute to cost reduction will increase the opportunity.”
Ives at Wedbush expects Meta CEO Mark Zuckerberg to "double down" on his forecast for $60 billion to $65 billion in capex this year, and sees around 70% of Microsoft's installed base using AI functionality within three year.
"We also believe Tesla could reach a $2 trillion market cap by the end of the year as its autonomous vision starts to take shape along with the very solid [vehicle] demand we expect from the core China market," he said.
Mag 7 commentary in focus
Deepwater Asset Management analyst Gene Munster, however, thinks that investors looking for clarity or support from the three Mag 7 updates could be disappointed.
"Even if you think that [DeepSeek is] embellishing some of the cost piece around [its AI model], that innovation piece is real," he told CNBC on Tuesday. "But there's a much bigger question that needs to be asked is not about what the cost to train the previous models were, but to get to that incremental step forward closer to artificial general intelligence."
More AI Stocks:
- Veteran trader discloses his top stock pick for 2025
- Analyst who bet correctly in 2024 unveils top AI stock picks for 2025
- One AI stock has become analysts’ top pick for 2025
"Most investors believe [that Meta, Microsoft and Tesla] are going to have some sort of reassuring comments that the AI trade, the hardware trade, the CapEx spending is intact," he said. "And when that becomes consensus, if that doesn't yield kind of a rebound in some of these stocks, I think you could see a little bit of a near-term sell off."
Related: Veteran fund manager issues dire S&P 500 warning for 2025
