Start with the number, because it is the whole story in one line: China’s cross-border deals for innovative drugs totaled a record US$110 billion in the opening half of 2026. That is money moving one way — toward Chinese-discovered molecules — and it is the strongest signal yet of where the industry believes the science is coming from.
The official read backs the market’s. Lan Gongtao, a deputy director general for drug registration with the National Medical Products Administration, says China accounts for roughly 30 per cent of the new drugs now in development around the world, enough for second place globally. On his telling, the country has climbed into the top tier internationally in innovation efficiency — the measure of how much usable drug discovery each unit of research effort produces.
The mechanics here are worth spelling out. A cross-border innovative-drug deal is, at bottom, a foreign pharma company paying a Chinese developer for access to a molecule: the right to develop it, trial it and sell it in markets the originator may never reach. When those payments reach record levels, buyers are effectively voting on the quality of China’s pipeline with their own capital. That is a harder test than any ministerial ranking, and on the first-half numbers the verdict was emphatic.
The catch is that a record half is backward-looking. Deals signed today were built on candidates discovered years ago, so the US$110 billion says nothing about the pipeline that has to sustain the next wave. That is precisely the gap the industry is now pointing at: after the record run, the sector says it is pivoting toward artificial intelligence-powered candidates as the engine of future transactions.
On closer reading, the pivot is a statement about production, not fashion. If AI-designed candidates compress the distance between idea and testable drug, the country that already produces about 30 per cent of the world’s development pipeline gains twice: once from the volume of candidates it can feed into dealmaking, and once from the speed at which it can replace what it sells. The regulator’s framing — innovation efficiency in the top tier — is exactly the metric that matters for that arithmetic.
Follow the money, as ever, and the incentives line up. Global buyers want candidates; Chinese developers want monetization and global reach; the licensing structure lets each side take the risk it is best placed to carry. What the first half of 2026 established is that the flow has a record behind it. What the stated AI turn promises is a deeper bench feeding the same channel. The next test is whether the machine-discovered candidates can command the same prices as the record-setters — and that is a claim only the deal log, not the forum stage, can settle.
