America needs to stop getting shocked by Chinese AI

The market is likely overreacting to a model-level convergence story and underweighting who actually captures value if frontier performance becomes easier to replicate. In the near term, that hurts premium multiple AI software names most: if customers believe capability is commoditizing, willingness to pay for standalone model/API layers fades faster than spending on compute. By contrast, the infrastructure complex can still hold up because the competitive response to parity is usually more experimentation, lower batching efficiency, and more inference load — not less.

The second-order effect is a valuation reset, not an immediate capex collapse. The first place to watch is whether hyperscalers revise data-center spend guidance over the next 1-2 earnings cycles; absent that, the selloff in semis is probably a positioning event rather than a fundamental regime break. If the models are genuinely closer to parity, the biggest losers are application vendors with thin workflow lock-in and weak distribution, while enterprises gain leverage in pricing negotiations.

Contrarian view: the consensus is treating this as bearish for AI spend, but the more durable mechanism is likely accelerated competition on both sides of the Atlantic, which often increases aggregate compute demand. Getty Images is only a very indirect beneficiary; any upside from rights-cleared content demand is too slow and too speculative to underwrite a trade today. The cleanest falsifier is a downward revision to hyperscaler capex or a sustained compression in AI-related cloud growth over the next 1-2 quarters.