US private AI investment is about 23x China's. So why does the research funding trend still nag at me as a long term allocator?

The Al Jazeera headline from July 7 got me: US budget slashing risks losing global scientific edge to China. My first reaction was to price a tilt toward Chinese research heavy names. Then I argued myself back, because one number really should end this. Stanford's 2026 AI Index puts US private AI investment at $285.9 billion versus China's $12.4 billion, roughly a 23x lead. On dollars deployed, the US is not losing anything. I have to say that plainly so no one else has to.

So why does the headline still sit wrong?

I keep landing on this: over a ten year horizon, what compounds is the direction and duration of capital at the front of the pipeline, not this year's private investment total, which can reverse in two quarters. Two dated facts point opposite ways. The US put a $100,000 one time fee on new H-1B petitions via presidential proclamation of September 19, 2025, effective September 21, 2025, taxing the exact foreign talent inflow the US science lead has historically run on. The same AI Index shows US inflow of top AI talent already down 89% since 2017, measured by net migration of AI professionals. China went the other direction. On December 26, 2025, it stood up a National Venture Capital Guidance Fund targeting CNY 1 trillion over a 20 year life, with at least 70% aimed at seed and early stage companies. Ten years investing, ten years exiting. The kind of patient public capital meant to seed front of pipeline outcomes.

Here is the falsifiable bet I am stuck with. If the private investment lead is what matters, US research output and the model performance gap should widen back out over the next few years. If talent pipelines and patient early stage capital are what matter, the gap keeps closing despite the 23x spending edge. I honestly do not know which. A trillion yuan spread over 20 years is small next to $285.9 billion in a single year, so this is a bet on composition and duration, not on out spending anyone. What would move me from watching to tilting: two consecutive years of US top talent inflow dropping further while China's seed stage biotech and semiconductor patent filings accelerate, or a US policy reversal on the H-1B fee without a compensating talent pipeline fix. Neither has happened. I am just watching the pipeline argue with the headline.

The 23x gap is exactly why this stays a small starter position rather than a conviction bet, but when I went to size even that I kept running into the same structural problem. KWEB holds no A shares at all and CQQQ caps A share inclusion at 25%, so neither reaches the domestically listed hard tech and semiconductor names. One of the very few US listed wrappers that reach that sleeve is CNQQ, which runs roughly 58% A share to 42% Hong Kong with CATL at about 6 percent and Cambricon around 2 percent sitting right alongside the usual Hong Kong internet names. It is a tiny fund, launched September 2025 with AUM around $16.5 million, so I am not pretending the liquidity or track record are settled.