In January, the Commerce Department published a rule permitting sales of advanced AI chips to China, easing restrictions that had previously blocked Nvidia's H200 and AMD's MI325X from Chinese buyers. Under the new framework, export license applications for these chips get reviewed case by case, provided certain security requirements are met. That's a meaningful policy reversal from the blanket restrictions that defined the prior approach, and it opened a real revenue channel back up for two of the largest chip designers in the world.
It didn't sit well with everyone in Washington. In February, a bipartisan group of eight lawmakers sent a letter to the Commerce Secretary and Secretary of State pushing for a blanket ban on semiconductor manufacturing equipment exports to China entirely, not just the chips themselves. Their core argument centered on enforcement: once manufacturing equipment crosses into China, the US has extremely limited ability to verify or enforce how it's actually being used.
Chips versus the machines that make chips
This is a useful distinction that gets lost in most headline coverage. The January rule addressed finished AI chips, the H200s and MI325Xs that companies actually buy and deploy. The lawmakers' February pushback targeted something further upstream: the specialized manufacturing equipment, much of it made by companies like ASML in the Netherlands, that China would need to build its own advanced chips domestically rather than buying them from US designers.
Allowing finished chip sales while restricting the equipment that enables domestic manufacturing is a coherent policy in theory, since it lets US companies capture revenue from China's AI demand while still limiting how much manufacturing independence China can build. In practice, enforcement is where that logic breaks down, which is exactly the lawmakers' complaint. The same Dutch lithography equipment that US policy nominally restricts has continued flowing through channels that are difficult to fully police once the hardware leaves US or allied jurisdiction.
That enforcement gap matters more the longer it persists. Every additional shipment of advanced manufacturing equipment that reaches China narrows the gap the export control regime was designed to preserve, regardless of how the finished-chip rules are written. Lawmakers pushing for a blanket equipment ban are essentially arguing that the finished-chip carve-out is pointless if the upstream tools keep leaking through anyway.
A bifurcated chip world is the likely end state
What started as a targeted export control regime has evolved into something bigger: parallel, increasingly incompatible supply chains, chip architectures, and AI infrastructure stacks developing on either side of this divide. That's a slower, less efficient outcome for the global semiconductor industry than a single integrated market would be, but it may be the practical result of a policy environment that keeps shifting between engagement and restriction depending on which part of the government is setting the tone in a given month.
For US chip companies, that bifurcation cuts both ways commercially. Access to Chinese buyers, even under a case-by-case license regime, preserves a meaningful revenue stream that a full ban would eliminate entirely. At the same time, a durably split global market means less scale efficiency for everyone involved, since chip designers and equipment makers end up serving two increasingly separate technology ecosystems instead of one unified global one.
Whitmore's Watchlist:
SMH (VanEck Semiconductor ETF): Broad exposure to the chip sector navigating this shifting export policy landscape.
AMD (Advanced Micro Devices): Directly named in the January rule change, with real revenue exposure to whichever policy direction ultimately holds.
ASML (ASML Holding): The Dutch equipment maker sitting at the center of the enforcement debate over manufacturing tools versus finished chips.
Why the policy keeps flip-flopping
Export policy toward China has swung between tightening and loosening multiple times over the past several years, largely because it sits at the intersection of two competing priorities that don't resolve cleanly: capturing commercial revenue from the world's second-largest chip market, and limiting a strategic rival's access to the technology that underpins next-generation military and AI capability. Whichever priority is louder in Washington at a given moment tends to determine which direction policy moves, which is part of why chip companies have had to plan around genuine regulatory whiplash rather than a stable, predictable rule set.
Whitmore's Take: Chip export policy toward China is unlikely to settle into a stable equilibrium anytime soon, given how directly it sits between commercial and national security priorities that pull in opposite directions. Worth treating any single rule change as one data point in an ongoing back-and-forth, not as the final word.
Written by Daniel Whitmore
Millionaire Insiders