A sea of red on Wall Street was the first impact following news of the apparent AI breakthrough of China’s DeepSeek. For example, NVIDIA’s market cap dropped by around $600bn.
DeepSeek R1 launched earlier this month. The firm claims that its performance compares with one of OpenAI’s latest models when used for certain tasks.
Go deeper with GlobalData
Sector commentators react to news of the apparent DeepSeek breakthrough.
Rahul Bhushan, Managing Director, ARK Invest Europe
DeepSeek’s recent breakthroughs serve as a pivotal reminder that the AI opportunity is expanding far beyond the narrow focus on semiconductors and infrastructure. For over a year, we’ve been emphasising to investors that concentrating too heavily on GPUs risks missing the transformative opportunities emerging in software, platforms, and open-source innovation.
DeepSeek’s V3 model, which matches the performance of GPT-4 using just 5% of the GPU compute, and its R-1 model, delivered at 1/13th of the cost of GPT o1, underscore an important truth: AI’s future is not just about throwing more GPUs at the problem. These advancements demonstrate how necessity is driving invention, with resource constraints fostering breakthrough efficiencies that are redefining what’s possible in AI development.
Moreover, the fact that DeepSeek’s innovations are open source cannot be overstated. This move opens the door to widespread adoption and decentralisation, a trend that could democratise AI access and accelerate progress far beyond traditional players in the West. It also hints at China’s growing strategic ingenuity in shaping the AI landscape under constrained circumstances. We strongly urge investors to re-evaluate their AI funds and positions.
Focusing solely on semiconductors risks being materially underexposed to where the real opportunities are emerging: scalable, efficient AI solutions and the open-source ecosystems enabling them. The paradigm is shifting—AI portfolios need to shift with it.
How well do you really know your competitors?
Access the most comprehensive Company Profiles on the market, powered by GlobalData. Save hours of research. Gain competitive edge.
Thank you!
Your download email will arrive shortly
Not ready to buy yet? Download a free sample
We are confident about the unique quality of our Company Profiles. However, we want you to make the most beneficial decision for your business, so we offer a free sample that you can download by submitting the below form
By GlobalDataBoris Kovacevic, Global Macro Strategist, Convera
Euro profits from AI shock
Optimism is rising among euro buyers as the currency benefits from AI market shifts and improving European data. The “DeepSeek shock”—an AI model from Zhejiang University that outperforms competitors at lower costs—caused a sell-off in US stocks.
Typically, such risk-off waves push investors to safe-havens like the Swiss franc and yen, both gaining against the euro. Yet, EUR/USD defied this trend on Monday as capital flowed away from the dollar, challenging the “US equity exceptionalism” notion and pushing EUR/USD above $1.05 for the first time this year.
George Vessey, Lead FX Strategist, Convera
Tariffs, AI spook markets
After its worst weekly decline in months last week on softer tariff talk, the stage looks set for a dollar rebound after Donald Trump’s Treasury Secretary Scott Bessent stated he will push for new universal tariffs on US imports to start at 2.5% and gradually rise. The ongoing uncertainty will continue hitting risk sentiment and boosting the USD, along with volatility, which was already elevated amidst the announcement from DeepSeek. Technology stocks fell Monday after Chinese AI start-up DeepSeek launched a competitive large-language AI model at lower computing costs. This has restrained risk-taking, worsened by Trump’s tariff threats—promising tariffs “much bigger” than 2.5% targeting sectors like semiconductors and pharmaceuticals.
Looking Ahead
Investors risk mispricing Trump by first betting on immediate tariffs, then on potential cuts to those measures. While the dollar’s haven dynamics are active, Trump’s tariff threats are boosting its value today.
Nigel Green, CEO of deVere Group
This is a wake-up call for markets. The assumption that tariffs could contain China’s technological ambitions is being dismantled in real time.
DeepSeek’s breakthrough is proof that innovation will always find a way forward, regardless of economic barriers.
By restricting China’s access to high-end semiconductors, Washington sought to slow its progress in AI.
Instead, it has fueled an acceleration in domestic innovation, forcing Chinese firms to find alternatives. DeepSeek’s achievement is a direct result of this shift.
Rather than being crippled by US sanctions, Beijing has cultivated AI models that require significantly less computing power, diminishing its reliance on American technology and eroding US leverage over global supply chains. Investors should be paying close attention to this shift. If China’s AI firms no longer require cutting-edge US chips, a core pillar of Washington’s strategy crumbles.
The market response we’re seeing is just the beginning of what could be a larger recalibration of AI investment flows.
For US policymakers, this moment demands a reassessment. Tariffs once served as a blunt instrument to enforce trade priorities, but their effectiveness is waning. If China can continue to develop advanced AI capabilities without access to cutting-edge US semiconductors, Washington’s economic arsenal will look increasingly outdated.
DeepSeek’s rise is not just a milestone for China—it is a warning for the US.
Tariffs may have worked in an earlier era, but in today’s world, economic power is determined by who innovates the fastest. If Washington keeps relying on restrictions instead of pushing ahead at home, it risks being left behind.
Derren Nathan, head of equity research, Hargreaves Lansdown
Companies in the semiconductor industry have borne the brunt of the sell-off as the emergence of a new AI model from Chinese startup DeepSeek, reportedly developed on a shoestring budget of under $6m, raised concerns about the outlook for spending on cloud infrastructure. The authenticity of this figure has been widely contested.
Nonetheless, Wall Street’s darling NVIDIA has lost its briefly held crown as the world’s most valuable company, diving 17% and losing $600bn of market value along the way, the biggest ever loss for a stock in a single day.
Other big semiconductor names caught in the crossfire include custom chip designer Broadcom and memory specialist Micron. Outside of the US, stocks that have taken a hit range from Taiwan Semiconductor Manufacturing Company through to the Dutch builder of chip printing machines ASML.
But that only tells one side of the story. For those looking to integrate AI into their business models the prospect of lower development costs could seriously boost returns on investment. Salesforce CEO Mac Benioff’s comments on social media that: ‘data is the new gold’ helped propel the shares up by 4%. And within the ‘Magnificent Seven’, Apple, Meta and Amazon were all in the green.
It’s going to take a while for the dust to settle here but it’s by no means the end of the party for AI infrastructure. Many of the recent big cheques set aside for investment in the space look to have been signed off after DeepSeek hit the scene. These include the Stargate proposal, the Bank of China’s 5-year AI investment plan and Meta’s capex plans of up to $65bn for 2025.
Steep reductions in development costs in the early years of technology shifts have been commonplace in economic history. A paradox first noted by the economist William Jevons in 1865 after observing a spike in coal consumption after the introduction of the more efficient Watt steam engine. A fall in cost can actually lead to a larger addressable market. So future demand for computing power could outstrip current expectations. That bodes well for the likes of NVIDIA, meaning that the current weakness could favour those brave enough to see through the market noise and buckle up for the longer term.
Stephen Deadmon, UK Associate Partner at global consultancy group, Sia
Deepseek’s meteoric rise to the top of the US iPhone download charts marks a significant milestone in AI development. Crucially, it signals that China is rapidly narrowing the gap with US tech leaders. This achievement is likely prompting many companies to reevaluate their high-capex AI investment strategies and could disrupt the US Government’s recently announced $500bn AI initiative, Stargate.
Deepseek’s advanced reasoning and coding capabilities challenge notions about the financial and computational resources required for cutting-edge AI. However, the model’s built-in adherence to government censorship truly highlights prevalent ethical concerns about how AI reflects social and political biases.
DeepSeek’s breakthrough has also exposed market fragilities by highlighting just how vulnerable established tech giants are to disruptive innovation from emerging players. The billions wiped off US tech stocks in response to the announcement also underscore concerns about potential overvaluation in the sector, the fragility of an AI-driven market bubble and the assumption that AI dominance will rely solely on closed-source models backed by those with the deepest pockets.
By delivering competitive performance at dramatically lower costs, Deepseek completely reshapes industry benchmarks and will certainly intensify the global race for AI dominance. This breakthrough is likely to accelerate advancements in AI development worldwide demonstrating that innovation may outweigh sheer financial clout in driving further progress. However, it also raises critical questions about geopolitical competition and proper ethical governance.
Bradford Levy, Assistant Professor of Accounting, University of Chicago Booth School of Business
DeepSeek has sent shock waves through the tech industry – directly challenging tech giants like Meta, Microsoft and Open AI.
Until now, it’s been assumed their expertise in designing and operating large-scale distributed systems are essential for training state of the art models. But the development of R1 suggests otherwise – if these models can be trained using 90% fewer chips, the implications for valuation models are massive.
This opens the door for smaller, more agile players to compete, potentially driving more innovation. With limited resources, they proved that scrappy, innovative teams can shake up the industry, even on a shoestring budget.
While impressive, we should remain sceptical of any claims made by those with a vested interest in their own success. Before jumping to conclusions about the broader AI landscape, we need more time to test these models and understand how they achieved these numbers.
