Chinese AI start-up DeepSeek is shaking up global markets, triggering a tech valuation reset and challenging giants like Nvidia. Experts say this shift is reshaping investment strategies, creating both risks and opportunities.
DeepSeek is a Chinese AI start-up making waves with budget-friendly models that rival top Western AI systems. Its breakthrough large language model ( LLM) has introduced a new paradigm in AI development, according to George Lagarias, chief economist at global professional services network Forvis Mazars. By delivering comparable performance to leading Western models like GPT-4 at lower costs and with less reliance on high-end graphics processing units (GPUs), DeepSeek has disrupted the established narrative of what is required to lead in AI. This, in turn, has catalysed a market correction, particularly affecting tech giants tied to AI advancements. “DeepSeek’s large language model catalysed a correction for the technology sector, especially Nvidia,” explained Lagarias. Nvidia, a perfect example of AI-driven growth, saw its value drop to about 17% from recent highs last Monday. US large-cap stocks, more broadly, fell 1.5% earlier this week, George pointed out.
Lagarias attributed the correction to inflated valuations that left the sector vulnerable. “We have often said the market was susceptible to a correction. Rising geoeconomic uncertainty meant even a less significant event could have triggered this.” While valuations have taken a hit, Lagarias stressed that the broader narrative around AI’s transformative potential remains intact. “The only thing that has changed is some of the froth being removed from valuations and earnings expectations,” he added.
The “ripples” caused by DeepSeek’s rise have not been limited to equities. Bitcoin, often seen as a hedge against economic uncertainty, dipped below the $100,000 mark amid the broader market turmoil. However, Nigel Green, CEO of financial solution provider deVere Group, views this as an inflection point rather than a cause for alarm.
“This is not the time for panic; it’s a time for perspective,” Green said. He argued that Bitcoin’s decentralised, borderless nature uniquely positioned it to thrive in an era of geopolitical tensions and technological disruption. Unlike equities, Bitcoin is not tied to the performance of any single company or nation. “It operates outside national borders and policies, making it a safe harbour in turbulent times,” Green added.
Green pointed to Bitcoin’s growing institutional adoption as a sign of its long-term resilience. “Short-term dips don’t deter long-term institutional players. They see Bitcoin as a strategic asset in portfolios, one that hedges against inflation, diversifies risk, and offers growth potential,” he explained. For Green, the current volatility is a reminder of Bitcoin’s core strengths—scarcity, decentralisation, and increasing adoption—which remain unchanged despite market noise.
Technology & operations leaders to gather at FundsTech Forum 2025
DeepSeek’s innovations have broader implications for how investors approach the AI sector. Its V3 model, which matches GPT-4 performance with just 5% of the graphics processing unit compute, and its R-1 model, delivered at 1/13th the cost of earlier iterations, underscore the shift toward efficiency-driven innovation. “AI’s future isn’t just about throwing more GPUs at the problem,” said Rahul Bhushan, MD of thematic ETF issuer ARK Invest Europe. He highlighted how DeepSeek’s resource-efficient models challenge the traditional focus on hardware-heavy solutions, urging investors to diversify their AI strategies.
Bhushan also emphasised the significance of DeepSeek’s open-source approach. “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,” he said. He warned that focusing too heavily on GPUs risks missing transformative opportunities emerging in software, platforms, and open-source ecosystems. “The paradigm is shifting—AI portfolios need to shift with it,” Bhushan added.
DeepSeek’s rise has also reignited concerns about the risks of concentrated market leadership in the tech sector. Jamie Mills O’Brien, investment director at asset manager Abrdn, pointed to Nvidia and other members of the so-called “MAG7 ( Apple, Microsoft, Alphabet, Amazon, NVIDIA, Tesla and Meta stocks) ” as examples of how market concentration can heighten vulnerabilities.
“DeepSeek represents another version of the risks tied to such narrow market leadership,” O’Brien explained. Nvidia, for instance, accounted for more than 20% of the S&P 500’s returns last year, but its dependence on specific pieces of news flow has made it susceptible to volatility.
O’Brien noted a shift in the market’s focus from “multiple expansion to earnings growth” as a driver of equity performance. “Earnings, not multiples, are driving relative performance in 2024,” she said. This trend, combined with a narrowing growth premium between large-cap tech stocks and the broader market, suggests a potential rotation towards a less concentrated market structure.
As the US tech sector recovers, emerging markets, particularly China, are poised to deliver stronger earnings growth. O’Brien highlighted the valuation discrepancies between US and non-US equities, noting the potential for regional and sectoral rotation in 2025 and beyond.
“With a longer-term perspective, the next 10 years are likely to see lower global growth, driven by factors like reduced fiscal headroom, demographic shifts, and deglobalisation,” Jamie said. This environment will place a premium on companies capable of growing independently of economic cycles, a trend that could benefit markets outside the US.
For thematic investors, DeepSeek’s rise serves as a cautionary tale. Kenneth Lamont, principal at Morningstar, called the start-up’s innovations a game-changer that could rewrite the investment case for AI. “Until now, the best AI models relied on massive datasets and immense computational power, favouring hardware giants like Nvidia,” he said. “DeepSeek’s models demonstrate how efficiency gains in AI development can reduce reliance on brute-force computing power.”
Lamont suggested that disruption in the AI sector could ripple through portfolios, particularly those heavily exposed to dominant tech players. To mitigate risks, Kenneth recommended diversifying away from the Mag7 and exploring equal-weighted strategies. “Mega-trends rarely unfold as expected, and today’s dominant players might not be tomorrow’s winners,” Lamont warned.
In agreement, César Gimeno, fund manager at independent non-banking manager in Spain Mapfre AM, points out that unlike its US counterparts, DeepSeek—founded by Chinese hedge fund manager Liang Wenfeng—operates on a significantly lower budget of just $6 million and relies on only 2,000 Nvidia H800 GPUs, drastically reducing both costs and energy consumption. Following its announcement, Nvidia’s stock plummeted by 17%, wiping out over $600 billion in market value, while semiconductor stocks suffered a steep 12.7% drop. “These results call into question the rationality of the multi-billion dollar investments we have seen announced recently,” Gimeno said, citing projects like OpenAI and SoftBank’s $500 billion Stargate initiative. If DeepSeek’s model proves sustainable, it could force major AI players to rethink their investment strategies, signalling the end of US dominance in the AI race and ushering in a new era of global competition, added Gimeno.
Source: Fund Europe
