Nvidia, Microsoft, TSMC in U.S.-China AI race

For investors, the implication is clear: the winners are increasingly the picks-and-shovels of the AI buildout, not the rhetoric around it. Nvidia, Microsoft and Taiwan Semiconductor Manufacturing Co sit at the center of that capital cycle. Nvidia shares have retreated from a peak near $235 to about $211, with the stock still above its 200-day moving average but trading below the 50-day line after a recent pullback. Microsoft has held far better, rebounding to about $505 after a violent midyear selloff, while TSMC remains in a powerful uptrend above both major moving averages despite a weaker tone in sentiment around earnings.

The market is pricing in a massive AI spend cycle, but it is still not fully reflecting the second-order effects of a U.S.-China technology arms race. When governments treat AI as a national-security asset, the result is more data-center buildout, more advanced-node wafer demand, more networking gear, more power and more restrictions on who gets access to the best hardware. That is bullish for the infrastructure layer and bullish for suppliers with scale, but it also raises the odds of volatility in names exposed to policy shifts and export controls.

Nvidia remains the purest expression of that trade. Its price action shows the stock consolidating after a strong run, with conventional momentum indicators cooling from earlier highs. That is not a reason to give up on the name; it is a reminder that the long-term thesis depends on the pace of AI adoption, not on any single quarter. If China’s ecosystem keeps narrowing the gap, the U.S. response is unlikely to be restraint. It is more spending, more competition and more demand for the chips that make AI run.

Microsoft is the cleaner way to play that reality. It has transformed from a software incumbent into one of the biggest AI infrastructure investors on the planet, and its rebound suggests institutions still want exposure to the cloud and model-distribution layer. TSMC, meanwhile, is the indispensable toll road in global semiconductors. The company’s shares have recovered sharply from summer weakness and remain well above the 200-day moving average, but sentiment around earnings has slumped to extreme fear even as awareness is high — a classic setup when the underlying demand story is stronger than the mood around it.

Our thesis is that the market is still too focused on AI enthusiasm and not focused enough on AI sovereignty. That distinction matters. A U.S.-China technology race tends to extend the investment cycle, not end it. It pulls in more winners downstream: chip equipment makers, power providers, advanced packaging, networking, memory and industrial buildout. The companies selling the tools and the infrastructure usually capture more durable upside than the end users chasing headlines.

If you want exposure before the next leg higher, stay with the core enablers of the AI stack rather than the most crowded application names. Nvidia, Microsoft and TSMC remain the best barometers of the race — and, in our view, the most direct ways to profit from it. The next catalyst will not be a speech about AI safety. It will be another round of capex, export policy and supply-chain spending as both superpowers race to lock down compute.