Institutional money that spent two years piling into Chinese AI-adjacent stocks is rotating into biotech, and the scale of that shift tells you something about where sophisticated allocators think AI valuations are headed.
A single data point captures how fast the ground has moved. In the first quarter of 2025, Chinese companies accounted for 32% of global biotech licensing deal value, up from 21% in both 2023 and 2024, according to a July report from New York investment firm Jefferies. That kind of jump in a single quarter doesn't happen by accident. It happened because Western pharmaceutical companies facing looming patent cliffs turned to Chinese biotechs for fresh pipelines at prices they couldn't find domestically, and because firms like Jiangsu Hengrui and BeiGene had, quietly and over several years, built drug assets worth licensing at scale. Hengrui's $12.5 billion deal with GSK for ex-China rights to HRS-9821 was the single largest outbound licensing transaction from a Chinese biotech ever when it closed in 2025. Hengrui alone has accumulated over $50 billion in total deal value since 2020.
The Bloomberg data on Chinese biotech's licensing surge runs parallel to a broader market rotation that's hard to ignore. Biotech IPOs in the US produced a weighted average return of 55% in 2026, according to Bloomberg's July 21 report, while the ten largest US IPOs of the year declined a weighted average of 6.3%. AI listings, which had been the preferred vehicle for anyone wanting exposure to the technology boom, have underperformed the broader IPO market. The institutions that made fortunes riding Taiwanese semiconductor names and South Korean AI infrastructure plays through the MSCI EM index are now scanning a different column of the spreadsheet.
The AI trade in emerging markets was never really about AI companies in the traditional sense. As Bloomberg's own index analysis noted in July, AI-adjacent EM stocks, primarily chipmakers and hardware suppliers, had become six of the biggest contributors to EM index performance over two years of outperformance. That concentration is the problem. When a handful of names carry an index, the crowding trade eventually becomes its own risk. Funds began fretting visibly about this: a Bloomberg piece from July 12 flagged that asset managers were questioning the $4.4 trillion AI trio's grip on emerging market returns.
Into that unease stepped Chinese biotech, which has a very different risk profile. It isn't a crowded index bet. It's a licensing-driven revenue stream tied to hard drug assets, not sentiment about AI infrastructure spending that has exceeded $400 billion annually but still hasn't produced the enterprise monetization that would justify it. There's no duration mismatch in a biotech licensing deal the way there is between infrastructure capex and software revenue. Either the drug works and the royalties flow, or they don't.
China's biotech firms have also gotten smarter about deal structure. Rather than licensing entire ex-China rights as a single block, companies are now carving out US, EU, Japan, and emerging market territories separately, which lets them run multiple licensing processes per asset and extract more total value. That sophistication signals maturity, and it's the kind of thing institutional allocators notice.
The broader context matters too. Chinese biotechs signed $135.7 billion in global licensing deals across 2025, according to data from Vision Life Sciences. By the first half of 2025 alone, US and European companies had signed 14 licensing agreements worth up to $18.3 billion for Chinese assets, compared with just two such deals in the same period of 2024. The acceleration is real, and it happened while many EM-focused funds were still debating whether to reduce their AI semiconductor exposure.
Whether this is a cycle or a regime change
The honest answer is that it's too early to know. AI hype fatigue is real among sophisticated allocators, but fatigue and reversal aren't the same thing. Goldman Sachs Asset Management, in a mid-2026 note on emerging market equities, flagged AI-driven earnings growth in North Asia as still a central thesis for the asset class, even while acknowledging the concentration risk. BlackRock has made a similar case: AI infrastructure spending hasn't stopped. It's just shifting from a momentum trade to a fundamentals trade, and that transition is painful for anyone who bought on narrative rather than numbers.
What's changed is the opportunity cost. When AI-adjacent EM stocks were cheap and underfollowed two years ago, the risk-reward made sense even on stretched assumptions. They aren't cheap now. Chinese biotech, by contrast, is still in an early phase of global recognition, with assets trading at meaningful discounts to Western peers on a per-deal basis. As PharmaVoice noted earlier this year, the bargain era may be ending, but it hasn't ended yet. That's the window funds are moving through right now.
For AI-adjacent startups competing for the same institutional capital, the implication is direct. The EM allocators who were writing checks into AI infrastructure plays are not cutting exposure because they've lost faith in AI. They're cutting exposure because the easy money has been made and something else looks more interesting. That's a rotation, not a verdict. But it does mean the pool of institutional money chasing AI growth stories in emerging markets has gotten shallower, and the bar for what justifies a premium has gotten higher. In that environment, having actual revenue matters more than it did. So does having a story that isn't already in the price.
Also read: Apple became the world's most valuable company by refusing to join the AI spending race • Nvidia bets $1.5 billion on Amkor to break its chip packaging bottleneck • TSMC beat every earnings record and Wall Street sold the stock anyway
