Why did the rejected chip export bill become Trump's "three birds with one stone"?

Question

The chip export bill was just rejected, and Trump immediately announced: NVIDIA is allowed to sell the H200 to China, but there will be a 25% cut on each chip. At the same time, the most advanced Blackwell and Rubin are still banned from sale. The US Department of Commerce is finalizing the details, and companies like AMD, Intel, and Broadcom will follow the same rules.

Many people only see the “loosening of controls,” but in reality, this move is: A carefully crafted “three birds with one stone” by Washington.

1⃣ Banning top-tier products while allowing the previous generation → Helps US companies secure $50–60 billion in revenue

Although the H200 is not the most cutting-edge, it is still the “golden mid-tier” that leading global AI companies are willing to pay for.

Allowing exports of previous generation products like H200 and MI300 to China means:

NVIDIA, AMD, Intel, Broadcom
Can steadily gain an extra $50–60 billion in annual sales This is an epic boon for the US tech sector, which has recently been weighed down by high inventory and slowing orders.

The US doesn’t spend a dime, yet gives domestic tech giants a channel to recoup profits.

2⃣ 25% cut → US government earns an extra $15–20 billion annually

The Trump administration’s logic is simple: You can sell, but I must take a cut.

The H200 is a high-price, high-margin product, with a huge export base to begin with. The US government takes 1/4 of the sales, netting the treasury $15–20 billion annually without needing to invoke tax law procedures.

This is a win-win-win for political correctness, corporate satisfaction, and fiscal profit.

3⃣ Most critically: Using the “H200 temptation” to delay China’s breakthrough in high-end domestic chips

This move is even more ruthless than the first two.

The most advanced domestic computing chips are already approaching the performance of the H200. If the US continued the ban, Chinese tech companies would be forced to go all-in on domestic chips, accelerating domestic replacement at an exponential rate.

But now that the US suddenly loosens restrictions on the H200, what’s the effect?

👉 Leading AI companies will quickly turn back to buy the H200 Because:

Mature CUDA ecosystem
H200 memory bandwidth of 4.8TB/s
NVLink 4.0 interconnect is still world-leading
Minimal engineering migration costs
Performance still slightly better than domestic chips

With efficiency and cost at stake, large model companies will make a clear choice: Prioritize the H200, and domestic chip orders will be “diverted” in the short term.

In other words: By “giving a little, but not everything,” the US is artificially delaying the progress of Chinese domestic chips in the high-end market.

This isn’t a conspiracy—it’s an open strategy.

Summary: Why is this Washington’s “three birds with one stone”?

✔ US company revenues are protected (+$50–60 billion/year) ✔ US government earns a cut (+$15–20 billion/year) ✔ China’s domestic substitution is delayed (high-end market occupied again by the H200)

On the surface, it’s “allowing exports,” But in essence: Binding the Chinese market with previous generation products, binding companies with revenue cuts, and binding the technology gap with bans on top-tier products.

The situation may appear to be loosening, But what’s really loosening is America’s carefully designed strategic rhythm— It’s not that the tech blockade is relaxing, It’s that the US believes “what’s being relaxed isn’t enough to threaten its lead.”

In short: The most advanced products won’t be sold, keeping you forever half a step behind.