The decision by the Trump administration to allow NVIDIA to export its advanced H200 artificial intelligence chips to approved customers in China represents a material shift in the global AI investment environment.
Nigel Green, CEO of deVere Group, says: “The move changes how capital markets should think about future AI leadership, competitive dynamics, and long-term value creation across sectors.
“This decision alters the speed and scale at which AI capability can spread. It matters for investors far beyond the chipmakers themselves.”
The H200 is one of NVIDIA’s most powerful AI accelerators, designed for large-scale model training and deployment.
“Access to chips of this class directly affects how quickly firms can build and refine advanced AI systems. Until now, restricted access formed a key constraint in the global AI ‘arms race’.”
China has already demonstrated an ability to operate effectively under those constraints. Over the past year, Chinese developers built increasingly capable AI services using limited-performance hardware such as NVIDIA’s H20.
The emergence of models such as DeepSeek underscored how algorithmic optimisation, large datasets and deployment scale can compensate for weaker chips.
“DeepSeek showed that hardware limits did not stop progress,” Green explains. “It simply forced a different approach.”
This context matters for investors assessing what changes when those limits ease. The availability of H200-level computing reduces development timelines, lowers iteration costs and allows more direct competition with leading global AI platforms.
“For investors, this is about acceleration,” notes Green. “When constraints come off, convergence happens faster.”
In the short term, markets are likely to focus on revenue and earnings.
“Expanded access to the Chinese market could generate meaningful upside for semiconductor companies and support parts of the global tech sector.”
But the medium-term investment picture becomes more complex.
Green comments: “Greater availability of advanced computers increases the number of serious AI competitors across industries such as autonomous vehicles, advanced manufacturing, logistics optimisation, healthcare analytics, and defence-linked technologies.
“When AI capability broadens, competitive advantages compress. It changes how investors price leaders versus challengers.”
Crucially, this is not a story of national winners and losers. It is a story about capital allocation, cost curves and scale.
“AI development follows economics,” Green says. “Whoever can combine computers, data and capital most effectively will advance fastest.”
China’s ecosystem brings certain characteristics into that equation. Large domestic datasets, rapid deployment, and a tolerance for lower efficiency in pursuit of capability can produce different competitive outcomes than those seen in US or European markets.
Over longer horizons, investors must consider how faster global diffusion of AI capability affects valuation assumptions. AI underpins productivity gains, automation, cost reduction and service scalability across almost every sector.
“When more players can access similar tools, excess returns shrink,” says Green. “Markets then reward execution rather than exclusivity.”
That shift matters for portfolio construction. Concentration risk rises when markets assume prolonged dominance by a narrow group of companies. Broader AI capability increases dispersion in outcomes.
“Investors need to prepare for a market where AI leadership is contested, not assumed,” Green says.
Supporters of tighter controls argue that restrictions preserve advantage, while opponents argue they distort markets and slow innovation.
From an investor perspective, Green says the focus should remain on outcomes rather than policy intent.
“The approval of H200 exports accelerates the transition already under way: AI becoming a widely deployable industrial technology rather than a tightly held advantage,” he concludes.
