TL;DR
- The structural discount Western pharma used to apply to China-originated assets has effectivelydisappeared. Diligence pipelines have matured, data quality is global-standard, and probability-of-success modelling now treats Chinese and Western programs on the same axes.
- A specific cross-border structure — the NewCo model — is now the default playbook for
Chinese assets reaching Western capital markets. - China’s structural advantages remain extraordinary: 12–15 months from PCC to IND at 50–60%
of Western cost, with dozens of companies routinely chasing the same target. - The AI/ML layer in Chinese pharma is the most interesting item to watch in the sector. The raw
inputs are world-class; the data integration layer is not yet built. When it is, the curve will
further steepen. - For anyone evaluating clinical programs, building cross-border partnerships, or trying to read
the next wave of innovation, the frameworks now need to change: China-versus-West is no
longer the relevant axis..
What’s actually changed since 2018?
Back in 2018 and 2019, when I first started paying close attention to China’s biotech ecosystem, my window into it was largely a data window. The clinical pipelines we tracked were starting to show a meaningful uptick in Chinese-originated first-in-human programs, and the question I kept coming back to — with a healthy dose of skepticism — was whether those programs were genuinely novel or whether the volume was masking a lot of recycled biology. The probability-of-success signals were thin. The readouts were inconsistent. Conversations at industry conferences were polite about Chinese innovation but rarely went much further than that.
Seven years on, the data tells a different story, and the Fierce Pharma panel I sat in on in San Diego this April made very clear that the qualitative picture has caught up with the quantitative one. The session pulled together more than sixty years of collective operating experience across China and broader Asia, and the candor with which the panelists walked through the structural changes made it one of the most useful sessions I have attended on the topic. What follows is my attempt to distill what they said, filtered through the lens I actually work in: clinical program data, probability-of-success modelling, and the
AI/ML tooling that increasingly anchors how we evaluate where assets sit on the development risk curve.
Has the Chinese biotech ecosystem really matured — or is this still
narrative?
The biggest single shift since 2018/2019 is that Chinese biotech is no longer a story you have to argue for. Back then, the live debate was whether Chinese CROs, CMOs, and small biotechs could deliver work that met global regulatory standards. Today, the panel made clear that the debate is closed. Chinese service providers and local biotechs are routinely producing global-standard preclinical and clinical packages, and most of the deals that have closed over the last 24 months passed Western pharma due diligence without data quality being the limiting factor.
The change shows up in the data, too. Clinical readouts from Chinese-led trials are increasingly appearing at AACR, ASCO, and ESMO with the same statistical rigor as their Western peers — and in oncology in particular, with cohorts large enough to drive real signal. The Chinese principal investigator network has become deep and sophisticated, and that is now a structural advantage.
A decade ago, the cohort of Chinese biotech executives who had been through a full cycle was small. Today, you can find founders on their second or third company. That kind of bench is what turns a boom into an industry.
A second change, less visible from outside the ecosystem, is leadership depth. A decade ago, the cohort of Chinese biotech executives who had been through a full cycle was small. Today, you can find founders on their second or third company, scientists who have shepherded multiple assets into clinical development, and bankers who have IPO’d companies in Hong Kong, Shanghai, and Nasdaq. That kind of bench is what turns a boom into an industry.
Is the “China discount” still a real thing?
For anyone watching deal flow over the last 18 months, the directional signal has been impossible to miss: the structural discount on China-originated assets is gone. The panel was emphatic on this. Every major global pharma, plus most serious mid-cap biotechs, now has dedicated scouts working Chinese pipelines. ADCs, T-cell engagers, dual and triple GLP-1 targets, SFRs — any asset worth having is being shopped into a competitive auction.
This matters in a way that goes beyond pricing. When you start to look at Chinese assets through the same probability-of-success lens you would apply to a Western program, the comparison is increasingly apples-to-apples. Modality, target precedent, indication, trial design, biomarker strategy — these are the variables that drive the probability curves, and on most of them, Chinese programs are now substantially indistinguishable from their Western counterparts. The discount, in other words, was always partly informational. Once the diligence pipeline matured and the data became legible, the discount evaporated with it.
Why this Matters for Deal-Making?
If your valuation model still applies an implicit “China haircut” on PTRS or risk-adjusted NPV, you are mispricing assets in a market where every other bidder has already removed it.
What is the NewCo model, and why does it matter?
If there is one structural innovation worth understanding right now, it is the NewCo model (sometimes called the Spin-off-NewCo or “SON” model). Here is how it works: a Chinese pharma or biotech spins a specific clinical asset, or a focused portfolio, into a new company headquartered in an offshore jurisdiction — typically Delaware, Cayman, or Switzerland — and partners with Western VCs to fund it. The Chinese originator retains equity, often a 20–30% stake, alongside upfront and milestone payments. The new company retains operational access to China’s R&D efficiency while wearing a Western corporate wrapper that resolves geopolitical concerns and opens cleaner paths to Nasdaq or trade sale.
The examples are now numerous and increasingly visible. Hengrui spun three GLP-1 assets into Hercules (subsequently renamed Kailera Therapeutics) in May 2024, with Bain Capital, Atlas Venture, RTW, and Lyra anchoring a $400 million Series A. Hengrui returned to the model later with Braveheart Bio (Forbion and OrbiMed, $1.1 billion). Keymed Biosciences carved two bispecifics into Belenos Biosciences with OrbiMed. Mabwell more recently licensed a dyslipidaemia siRNA into Aditum Bio’s Kalexo Bio for up to $1 billion. The structure has moved from novelty in 2024 to default playbook in 2026.
One of the panelists described their own company as a variant of this pattern: US-incorporated, Swiss-headquartered, with the underlying science originated in Hong Kong and operational execution running through China. From an evaluation standpoint, what makes the NewCo model interesting is that it cleanly separates the asset from the geopolitical wrapper around it. That makes the program easier to benchmark on its scientific and clinical merits, and easier to position into a Western capital market, without losing the cost and speed advantages of the Chinese operating base. It also gives the Chinese originator a way to participate in the equity upside, rather than capping out at licensing milestones — which, in a market where the largest deals are now structured around $1B+ in milestone potential, materially changes the long-term economics.
How fast is fast? The numbers behind China’s structural advantage
One number from the panel deserves to be written in large type: PCC to IND in China now runs 12 to 15 months, at 50–60% of Western cost. That is not a marginal advantage which compounds when paired with the other feature of the Chinese market that Western observers still find hard to internalize — the sheer number of companies working on the same target.
12–15 months — PCC to IND in China
50–60% — of Western cost
Up to 50 — companies routinely chasing the same target
In the US or Europe, once five or six credible competitors are working a target, the sixth entrant typically walks away. In China, that number is routinely multiplied by ten. Fifty companies chasing the same target is not unusual. Part of this is the structure of semi-governmental funding, which ties capital to local R&D and manufacturing. Part of it is cultural — a willingness to make incremental improvements on top of someone else’s concept and see where the iteration leads. And part of it is simply that the cost of a failed program is low enough to make the shot worth taking.
Two implications follow. The first is that China has effectively become a global parallel processor for drug discovery; almost any target worth working on has multiple Chinese programs somewhere in the pipeline. The second is more subtle but increasingly important for anyone trying to evaluate Chinese pipelines: the volume of programs per target makes the underlying pipeline data richer than the equivalent Western dataset. When fifty companies are running variants of the same molecule in parallel, the population-level signal — how often the modality works, on which targets, in which patient populations — sharpens in a way that smaller Western pipelines simply cannot match. That has implications for how we model probability of success, and for how confidently we can read across between programs.
Is “Asia” the right unit of analysis?
It was also a useful reminder that “China” is not shorthand for “Asia.” The panel’s framing of the regional landscape was worth the price of admission on its own.
- Korea is shaping up as a serious secondary market for well-run biotechs with globally competitive assets.
- Japan’s majors — Takeda, Eisai, Daiichi — are best understood as potential partners rather than competitors, and the Japanese IPO market, while open, remains effectively invitation-only and complex.
- Singapore has leaned hard into manufacturing excellence, with real government support behind it and tangible results to show for it.
- Hong Kong, through the Chapter 18A pre-revenue pathway, has become the default IPO market for Chinese biotechs that cannot or will not list in Shanghai — which, despite a nominal pre- revenue segment, operates by invitation and imposes multi-year founder lockups that most founders find unworkable.
- Nasdaq, meanwhile, is tentatively reopening to Chinese biotech issuers after a long hiatus. Whether that window widens will depend on geopolitics as much as fundamentals, but the early signals are constructive.
Why is China’s AI-in-pharma boom still under-realized?
This is the section I expected to find most familiar, and the one where the panel’s framing actually surprised me, in a useful way.
China has the raw ingredients for an AI-in-pharma boom: enormous patient cohorts, vast hospital tissue repositories, a deep bench of computer scientists, and abundant funding. By every input metric, you would expect Chinese AI/ML in drug discovery to be at least as advanced as the West’s. And yet, on the panel’s reading, the explosion has not fully materialized. The bottleneck is not computational capacity. It is data integration, regulatory acceptance, and the commercial models needed to turn fragmented hospital data into usable training sets at scale.
The frontier is rarely the algorithm. The frontier is the data plumbing, the ontologies, the linkages, and the cleaning.
That maps closely to what we see from the AI/ML side of clinical program evaluation. The most useful machine-learning work in this domain is fundamentally a data integration problem: assembling clinical trial outcomes, regulatory filings, patient-level signals, and program-level metadata into a substrate that models can actually learn from. The frontier is rarely the algorithm. The frontier is the data plumbing, the ontologies, the linkages, and the cleaning. China’s particular bottleneck, extraordinary raw data assets but underdeveloped integration and access pathways, is a more acute version of a problem the whole industry is wrestling with.
The interesting question is what happens when those bottlenecks resolve. Given the volume and granularity of the underlying Chinese clinical data, the China leg of the AI/ML curve is potentially steeper than the West’s once the integration layer matures. The same is true for clinical-program evaluation more broadly: the more comprehensive and standardized the data substrate, the better the probability-of-success models trained on top of it perform. That is a three-to-five-year story rather than a six-month one, but it is the part of this market I would tell someone to keep closest tabs on.
Five takeaways for anyone evaluating Chinese assets in 2026
- The China-versus-West framing is increasingly obsolete. The relevant question is no longer “is the Chinese asset good enough?” — it is which assets, anywhere, best fit a given strategy. Pipelines now need to be evaluated on the same axes regardless of country of origin.
- The data is legible enough to model. Chinese clinical programs have matured to the point where probability-of-success and risk-adjusted valuation work the same way they do for Western assets. The discount that used to cover for informational uncertainty has largely closed.
- The NewCo model decouples the asset from the geopolitics. It is reshaping how Chinese science reaches Western capital markets, and understanding the structure — Hengrui/Kailera, Hengrui/Braveheart, Keymed/Belenos, Mabwell/Kalexo — is now table stakes for anyone trying to read cross-border deal flow.
- Asia is plural. Korea, Japan, Singapore, and Hong Kong each play distinct, increasingly differentiated roles. A flat “Asia” thesis will miss most of the picture, and most of the value.
- AI in Chinese pharma is the most interesting watch-item in the sector. The raw inputs are extraordinary; the integration layer is not yet built. When it is, the curve is likely to steepen quickly, and the implications for how we evaluate clinical programs globally will follow.
Closing thought
What stayed with me after the session was the sense that this is no longer the breakthrough phase. The talent, the capital, the deal volume, the regulatory maturation, and the cross-border structures are all now reinforcing each other. The acceleration that started around 2015, and was visible in the pipeline data by 2018/2019, has not run out of runway. If anything, the curve is steepening, and the AI/ML layer that the panel was rightly cautious about is the most likely accelerant of the next phase.
For anyone evaluating clinical programs, building cross-border partnerships, or simply trying to keep up with where the next wave of innovation is coming from, the takeaway is the same: the next seven years will look less like an extension of the last seven and more like a phase change. The frameworks we use to read this market, and the data and tooling underneath them, need to evolve accordingly.
FAQs
What is the NewCo model in biotech cross-border deal-making?
The NewCo model (also called the Spin-off-NewCo or “SON” model) is a structure in which a Chinese pharma or biotech spins a specific clinical asset, or a focused portfolio, into a new company headquartered in an offshore jurisdiction — typically Delaware, Cayman, or Switzerland — and partners with Western VCs to fund it. The Chinese originator retains equity, often a 20–30% stake, alongside upfront and milestone payments. The model cleanly separates the asset from the geopolitical wrapper around it, makes the program easier to benchmark on its scientific and clinical merits, and allows the Chinese originator to participate in equity upside rather than capping out at licensing milestones.
How does the NewCo model affect long-term economics for the Chinese originator versus traditional out-licensing?
The main difference is equity participation. In a traditional out-licensing deal, the Chinese originator receives upfront cash, milestone payments tied to development or commercial events, and tiered royalties on net sales. In a NewCo deal, the originator typically receives a smaller upfront payment plus a meaningful retained equity stake in the new company, often in the 20–30% range. Hengrui retained roughly 19.9% of Hercules (now Kailera) in May 2024; iBridge HK, Keymed’s subsidiary, took approximately 30% of Belenos Biosciences in July 2024. That equity stake matters because the NewCo’s value is then driven by subsequent rounds of dilutive VC financing, clinical milestones, and ultimately an IPO or trade-sale exit — outcomes that historically delivered substantially more value than royalty streams. The trade-off is that the originator surrenders direct control of ex-China development and accepts that any return on the equity stake is contingent on the NewCo’s clinical and commercial trajectory. For assets where the originator believes in the molecule but lacks the capital or operational reach to develop it ex-China themselves, the model is now widely seen as offering better long-term economics than traditional licensing alone.
Why is the “China discount” on biotech assets disappearing?
The discount was always partly informational — it covered for uncertainty about whether Chinese clinical data, manufacturing standards, and program quality met global benchmarks. As Chinese service providers and biotechs have consistently produced global-standard packages, and as Western diligence processes have matured, the informational uncertainty has closed. Probability-of-success and risk-adjusted valuation models now treat Chinese assets on the same axes as Western ones, removing the basis for a structural discount.
If the China discount has closed, are Chinese assets now at risk of being overpriced?
The directional signal in 2026 is that pricing has firmed considerably. According to Evaluate data reported by Fierce Biotech in February 2026, the average upfront value for cross-border licensing deals involving Chinese innovators rose from $52 million in 2022 to $172 million in early 2026 — a 230% increase. The 2026 average is already running 22% above 2025’s full-year average. Senior dealmakers have begun noting that some Chinese biotechs are now pricing aggressively, and AstraZeneca’s $18.5 billion deal with CSPC for weight-loss assets ($1.2 billion upfront) and GSK’s $12 billion partnership with Hengrui (announced in 2025) signal that the largest transactions are now priced competitively against any global benchmark. Whether that constitutes overpricing depends on what the assets deliver in pivotal data — and the universe of Chinese assets in late-stage trials is large enough that some will inevitably miss. The more durable answer is that the asymmetric information advantage Western buyers used to enjoy has closed, but Chinese assets still offer favourable economics in three structural respects: lower development cost from PCC to IND, faster trial enrolment via the Chinese clinical infrastructure, and access to first-in-class candidates in modalities (ADCs, bispecifics, GLP-1 combinations) where the Chinese pipeline density is unmatched. A “value” thesis is still defensible. A “bargain” thesis is not.
Why has AI in Chinese pharma not exploded the way the inputs suggest it should?
The bottleneck is not computational capacity, talent, or funding — China has all three at scale. It is data integration: the regulatory frameworks, commercial models, and technical infrastructure needed to turn fragmented hospital data into usable training sets. The frontier in AI for drug discovery is rarely the algorithm; it is the data plumbing. Once China’s integration layer matures, the underlying volume and granularity of Chinese clinical data could make its AI/ML curve steeper than the West’s.
Edoardo Madussi is Senior Director, Business Development & Strategic Initiatives at Intelligencia, where he focuses on cross-border partnering strategy, asset evaluation, and AI-enabled clinical program assessment. This piece reflects his observations from a Fierce Pharma panel on China’s biotech rise, San Diego, April 2026.
For more tailored, data-rich insights, let’s talk.
