NEW YORK/LONDON (Reuters) -Technology shares around the world slid on Monday as a surge in popularity of a Chinese discount artificial intelligence model shook investors' faith in the AI sector's voracious demand for high-tech chips.
Startup DeepSeek has rolled out a free assistant it says uses lower-cost chips and less data, seemingly challenging a widespread bet in financial markets that AI will drive demand along a supply chain from chipmakers to data centres.
- Nvidia down 16.3%; Microsoft off 3.8%, Meta Platforms down 0.8% and Alphabet shed 3.4%
- European tech stocks slid 3.25%, set for their worst day since October. Chip maker ASML down 7%, and Siemens Energy, which provides electric hardware for AI infrastructure, slid around 20%.
- Japan's Nikkei shed nearly 1%, weighed on by heavyweight tech names. AI-focused startup investor SoftBank Group fell over 8%.
- The yield on the U.S. 10-year Treasury note was down 7.9 basis points 4.544% as investors moved into safer US government debt.
"Bear in-mind that DeepSeek’s AI Assistant is focused on mobile phones, PCs and laptop devices, NOT the massive Datacenters services that the Hyperscalers provide like - AMZN’s AWS, MSFT’s Azure and Alphabet’s Google cloud platform."
"So, I see very little threat here. The real money in AI is providing the chips for the datacenters from the likes of -NVDA, AMD, & AVGO."
"I can see some threat from DeepSeek’s AI models to MSFT’s ChatGPT, META and Alphabet’s Gemini. But not to the data center."
"Overall, I view the AI tech sell-off today as an opportunity to add high quality tech shares on weakness."
“I think that everyone in the market loves to hate on the U.S. concentration of the top 10 stocks, but these stocks are the exact reason that we hit all-time highs last week. Some may see it as a risk, as I see it as an opportunity because these mega-cap companies exhibit operating leverage which can continue to drive operating margin expansion. No matter how you slice it, the U.S. market is dependent on these stocks, but that's not always a bad thing.
"Does this undermine valuations broadly in US equities? When a narrative is driving the market, there may be increased volatility until we see tangible results from an ROIC perspective on the AI spend. This does not mean that the narrative has been disproven, and I think that this is a buying opportunity, as U.S. stocks still yield the largest amount of growth with the lowest regulations throughout the entire world.
SHAMS, AFZAL, MANAGING DIRECTOR CARNEGIE INVESTMENT COUNSEL, TOLEDO, OH
“DeepSeek R1’s emergence has shown that necessity is still the mother of invention. The fractional compute intensity at DeepSeek compared to U.S. counterparts is a result of some breakthroughs that would likely not have happened if the engineers had $80 billion of discretionary capex at their disposal. As LLM scholars and practitioners dissect the DeepSeek papers, here is what we know so far: “DeepSeek’s model does not fire on all cylinders at all times but rather gets activated at different phases of the total compute cycle. It predicts which part of the model would need to activate and thus keeps the computational workload at a minimum. This is quite different than Meta’s, Google’s, and Open AI’s LLMs that are on full tilt from the start. The reserved computational output is made possible by one of their major breakthroughs, the KV cache compression, meaning that their engineers have been able to extract higher levels of throughput in memory constrained applications. This is primarily why DeepSeek’s inference costs are much lower.
“This has implications on the existing sanctions on high end Nvidia chips and the significant capex targets by the Mag 7 names, especially Meta, Microsoft and Google. In the near term, this will negatively affect Nvidia’s valuation especially if the lower priced H800 chips may end up being potent enough for many AI applications. It does not appear that DeepSeek circumvented export restrictions based on an early assessment that the H800 and not the latest H100 chips were used to train the models. The question for national security experts would be if Nvidia was right to have been allowed to modify the H100 chips for export purposes.
“The development hurts lofty valuations of chip companies in the short term, and it likely hurts data center REITS long term if the final assessment is less demand for high intensity compute. The downstream effects will be seen in lowering expectations for power equipment players like Eaton, cooling solutions providers like Vertiv and other verticals in the data center space.
“Ultimately, if AI is a positive for humanity, breakthroughs like KV cache compression and predictive computational power deployment in LLMs are net benefits to the overall AI endeavor. How this changes the over $225 billion in capex targets announced by a half dozen mega tech names remain to be seen.”
BEN LAIDLER, HEAD OF EQUITY STRATEGY, BRADESCO BBI, LONDON
"The significance of the DeepSeek news seems to be very powerful, seems to be cost effective and if that's proved right over the fullness of time, that challenges the investment thesis of paying very high multiples for big tech, for that AI dominance and AI growth.
"The share price move today is a reflection of very high expectations built into those stocks. It's highly significant, this potential valuation and expectations reset. However, I wouldn't push it too much. There's more to the big tech stocks than just AI, even though it's a significant portion of the story.
"Four of the Magnificent 7 are reporting this week. We’ll get the management to say about their AI progress and the potential threat from China. I would struggle to become overly bearish here. We're going to see guidance which will probably be reasonably reassuring.
"Chinese equities is the second biggest equity market and one of the cheapest equity markets in the world. This news shines a broader light on the tech prowess within China. We may begin to see investors revisiting Chinese stocks, which certainly international investors haven't done for a very long time."
BRIAN JACOBSEN, CHIEF ECONOMIST, ANNEX WEALTH MANAGEMENT, MENOMONEE FALLS, WISCONSIN
"If it’s true that DeepSeek is the proverbial 'better mousetrap,' that could disrupt the entire AI narrative that has helped drive the markets over the last two years. It could mean less demand for chips, less need for a massive buildout of power production to fuel the models, and less need for largescale datacenters. However, it could also mean that AI becomes more accessible and help kickstart the development of a wide array of useful applications.
"For the last two years we have not only been monitoring the AI revolution and investing in it in different ways, but we have also been focused on not focusing solely on investing in the conventional AI story. Concentration risk—having too much in one stock or one theme—can feel good when those few names or ideas are on the ascent, but it is even more dangerous when disruptions take place.
"We urge people to not overreact to the market moves. It is possible that the news out of China could be overstated and then we could see a reversal of the recent market moves. It is also possible that the news is true, but then that would present new investment opportunities. Those are the various investment opportunities we have been positioning for—focusing less on the chips and hardware and more on the possible profitable use cases of the technology."
CHRIS LARKIN, MANAGING DIRECTOR, TRADING AND INVESTING, E*TRADE FROM MORGAN STANLEY, NEW YORK
"What was shaping up to be a big week in the markets got even bigger with the disruption in the AI space. That could make this week’s megacap tech earnings even more critical to market sentiment. And while no one expects a rate cut rate on Wednesday, everyone will be looking for some signs of longer-term dovishness from the Fed. But don’t forget surprises out of Washington. The Colombia tariff spat may increase concerns that the Trump administration will be more aggressive on this front than previously thought."
BOB SAVAGE, HEAD OF MARKETS STRATEGY AND INSIGHTS, BNY (in a note)
“The central question today for markets is where is the bottom? … The catalyst of a foreign competitor to US led AI dominance begs other questions about trade and semiconductor chips and energy needs … The markets are unsettled, and volatility is higher after last week’s strong returns. Also adding to today’s moves is the lack of Asian liquidity given the Lunar New Year holidays”
JON WITHAAR, SENIOR PORTFOLIO MANAGER, PICTET ASSET MANAGEMENT, SINGAPORE:
"We still don’t know the details and nothing has been 100% confirmed in regards to the claims, but if there truly has been a breakthrough in the cost to train models from $100 million+ to this alleged $6 million number this is actually very positive for productivity and AI end users as cost is obviously much lower meaning lower cost of access."
"Is it negative for Nvidia in the short term? Yes, as expectations are sky high on Blackwell (chips)and positioning is long in anything AI supply chain related, but ultimately anything that makes AI cheaper to implement is positive for those selling AI related products and applications and using AI related tools - an ever growing group."
"But let’s see the devil is in the detail and as you can imagine a Chinese model will be controversial for many uses. Still it is a cold shower and a dose of reality for a sector that probably needed it."
DANIEL TAN, PORTFOLIO MANAGER, GRASSHOPPER ASSET MANAGEMENT, SINGAPORE:
"The selloff in Japan and U.S. tech names should not be a surprise given high valuations based on P/E and P/B ratios. With the current rate of P/E priced for some U.S. and Japan tech names, the market is expecting future earnings to continue to justify the high prices of these tech names. That is definitely a high expectation to meet."
"However, DeepSeek has shown over the weekend with its updated AI model that is cost-effective with OpenAI's technology while running on reduced-capability chips, raising questions over the dominance of U.S. tech firms such as Nvidia Corp."
RICHARD CLODE, TECH PORTFOLIO MANAGER, JANUS HENDERSON INVESTORS, LONDON:
"DeepSeek appears to bringing some genuine innovation to the architecture of general purpose and reasoning models. Given compute restrictions in China it is not surprising that necessity drives innovation as we saw in Russia in the 1990s when limited access to PCs drove coding creativity and a generation of novel Russian coders."
"As Yann LeCun (Meta's chief AI scientist) has noted, this is a victory for the open source model of driving community innovation with DeepSeek leveraging Meta’s Llama and Alibaba’s Qwen open source models."
"However, DeepSeek is leveraging distillation techniques that are reliant on the work of others. How reliant is a key question the market is grappling with currently. Ultimately, we think that creates more questions around the direct monetisation of LLMs (large language models)and open source models than it does about AI capex. This announcement will also draw more scrutiny from the US around chip restrictions as well as the proliferation open source AI models."
ALEXANDR WANG, CEO SCALE AI, (TWITTER POST)
"DeepSeek is a wake up call for America, but it doesn’t change the strategy:
- USA must out-innovate & race faster, as we have done in the entire history of AI
- Tighten export controls on chips so that we can maintain future leads
Every major breakthrough in AI has been American."
GEORGE LAGARIAS, INVESTMENT STRATEGIST, FORVIS MAZARS, LONDON:
"China and DeepSeek say, at the very least, that they can deliver what ChatGPT can deliver today at a fraction of the cost," said George Lagarias, investment strategist at Forvis Mazars.
"It makes sense that markets question the narrative that has been underpinning the whole market ... It's a very frothy market so it doesn't really take that much for investors to take some profit."
BEN BARRINGER, TECHNOLOGY ANALYST, QUILTER CHEVIOT, LONDON:
"DeepSeek's success will serve to intensify the US/China AI war, particularly following the recent announcement of the Stargate project in the U.S."
"It also provides a wake up call and somewhat of a question mark on how much needs to be spent in order to build a model, and whether quite the level of CapEx we have been seeing is really required."
"However, bringing the price of model training down is no bad thing as it will help to lower the entry point and this price elasticity could help drive usage and volume."
(Compiled by the Global Finance & Markets Breaking News team)
