The AI pharmaceutical industry is extremely hot in the primary market
AI drug development continues to boom.
Since the news of ByteDance spinning off its AI pharmaceutical business line and launching independent financing broke out in June, Anew Labs has closed its first round of financing in less than 3 months.
At USD 290 million, with a post-money valuation of approximately USD 1.5 billion, this deal has refreshed the record for the largest single financing of domestic AI pharmaceutical projects within the year.
Although its fastest-moving pipeline is still in the preclinical stage, its valuation has already reached USD 1.5 billion — the primary market pays a premium for the company's technology platform and ByteDance ecosystem, and there is its own underlying logic behind this.
Looking at the global landscape, this upsurge is even more pronounced. In May, Google's subsidiary Isomorphic Labs secured USD 2.1 billion in Series B financing, with a valuation hitting USD 15 billion, setting the record for the largest single financing in the history of AI drug development.
The overheating in the primary market further highlights the "undervaluation" of AI pharmaceutical players in the secondary market. For example, the latest market capitalizations of Metis Therapeutics and Insilico Medicine are approximately USD 2.6 billion and USD 4.4 billion respectively.
The reasons behind this valuation discrepancy are diverse, including differences in valuation models, risk appetite, and varying levels of conviction in the idea that "AI will transform life sciences".
However, as the global AI pharmaceutical industry accelerates into the performance delivery stage, more and more enterprises will either prove their value with data, or be disproven by clinical trials and the market. This valuation gap between the primary and secondary markets is destined not to exist for a long time.
Overheating in the Primary Market
Anew Labs was born with a "silver spoon": it has abundant capital, a top-tier team, a mature platform, published academic papers, and proprietary molecular assets.
That is why all developments of Anew Labs this year have drawn widespread attention, especially this first round of financing. On the one hand, its investor lineup is extremely star-studded: ByteDance retains a 56% equity stake, the lead investors are HSG, IDG Capital, and GL Ventures, 5Y Capital acts as a co-lead investor, Gaorong Ventures, Chunhua Ventures and Boyu Capital follow on, and strategic investors include China Biopharmaceutical and Shanghai Future Industry Fund.
On the other hand, the USD 290 million financing amount has refreshed the record for the largest single domestic AI pharmaceutical financing of the year. With a post-money valuation of approximately USD 1.5 billion, to a certain extent, this means that Anew Labs has become a unicorn in the primary market right after its inception, with an extremely high starting point.
The core highlight of Anew Labs is AnewOmni, the world's first claimed full-modal drug design large model that covers small molecules, peptides and antibodies, as well as AnewSampling, a generative model that reconstructs the molecular dynamics equilibrium distribution at the all-atom level, which is 1000 times faster than traditional MD, and can accurately capture protein flexibility and ligand binding patterns.
These two platforms also mean that Anew Labs has realized a full-chain layout covering target mechanism analysis, novel molecular creation and druggability optimization.
On top of these platforms, its first AI-generated molecule has been unveiled. On April 25, Anew Labs delivered an oral presentation at AAI 2026, publicly disclosing for the first time its preclinical-stage IL-17 small molecule inhibitor, which was successfully identified using AI-driven virtual screening technology combined with molecular generation algorithms, achieving pan-inhibition of the IL-17 family (AA/AF/FF) at the small molecule level for the first time.
In addition to the IL-17 small molecule inhibitor, the official website of Anew Labs shows that there is another pipeline targeting IL-4R that has completed target hit identification, and two other pipelines with undisclosed targets are in the hit compound discovery stage.
The more remarkable case is Isomorphic Labs. On May 12, this company spun off from DeepMind and led by Nobel laureate Demis Hassabis announced the completion of USD 2.1 billion in Series B financing, with a valuation of approximately USD 15 billion, making it a phenomenal asset in the global AI pharmaceutical industry.
What is more noteworthy than the financing scale is that although Isomorphic Labs relies on Alphabet, is deeply integrated with DeepMind's scientific research system, has top-tier computing power, plus the world's top talent lineup and international capital support, with unique resource advantages, it still lacks mature clinical pipelines at present.
The valuation of Anew Labs is 10 times lower than that of Isomorphic Labs, but their valuation logic is identical. Investment institutions are pricing them entirely according to the logic of tech stocks and AI stocks. In particular, the high valuation of Isomorphic Labs essentially stems from the scientific research influence established by AlphaFold, as well as the computing power, model and ecological capabilities provided by Alphabet.
It is not difficult to understand that as tech giants all extend their reach into drug R&D, capital is willing to pay a premium for this "AI + Life Sciences" ticket to obtain technology options for the future.
Conservatism in the Secondary Market
The overheating in the primary market highlights the relative conservatism of the secondary market.
Take Metis Therapeutics as an example. It has just released an impressive semi-annual report, with its revenue in the first half of the year far exceeding the total revenue of last year. More importantly, it adopts a two-wheel-driven model of platform and pipelines, with successive platform cooperation deals. In the first half of the year, it partnered with Hengrui Medicine to deploy its AiTEM platform locally; at the pipeline level, it now has more than 10 R&D pipelines under development, among which MTS-004 has entered the NDA application stage, and MTS-109, an mRNA-encoded trispecific TCE targeting autoimmune diseases such as SLE and lupus nephritis, is advancing IND applications in China and the US. MTS-128 has set a new record for the amount of a single overseas licensing deal for preclinical TCE projects, with a USD 20 million upfront payment and milestone payments and commercial royalties of up to USD 1.6 billion.
At present, relying on NanoForge, Metis Therapeutics has carried out cooperation with a number of enterprises, with the total potential transaction value exceeding RMB 6 billion. At a stage when the AI pharmaceutical industry is generally still burning cash, Metis Therapeutics has moved fast enough.
The same is true for Insilico Medicine. Since 2021, it has nominated 33 drug candidates using its AI platform, 13 of which have entered clinical trials, and its core pipeline Rentosertib has advanced to Phase III clinical trials; multiple pipelines have reached BD cooperation, and the cumulative potential value of in-depth cooperation with MNCs such as Eli Lilly, Sanofi, and Servier has reached the level of tens of billions of US dollars. In the first half of the year, Insilico Medicine also achieved its first profit.
Both Metis Therapeutics and Insilico Medicine are verifying the commercial feasibility of AI technology empowering new drug R&D with tangible cooperation cases, pipeline progress and performance. In addition, both of the two companies are advancing their AI platform strategies, but in different directions: the core of the former is Biological AI, which builds a full set of industrial infrastructure integrating AI, delivery and dry-wet closed loop;
The latter is working on "Pharmaceutical Superintelligence", which forms a complete dry-wet closed loop with three pillars: high-quality vertical data base, life science vertical large model system centered on MMAI Gym, and autonomous experiment system represented by LabClaw.
But if we compare it with the valuation logic of the primary market, we will find that the secondary market is still far too conservative.
The same applies to the US stock market. The two veteran players Schrödinger and Recursion have a market capitalization of around USD 2 billion after a sharp rise, which is at the same valuation level as Anew Labs, which has just completed its first round of financing. In particular, the former, which co-developed Zasocitinib, a new generation of highly selective and potent oral TYK2 inhibitor, with Nimbus Therapeutics (later acquired by Takeda Pharmaceutical), saw Takeda submit its marketing application to the FDA and receive priority review on September 14.
If nothing unexpected happens, the first "AI drug" will be launched in the first quarter of next year. However, Schrödinger's share price is still hovering at a low level.
Of course, the two companies also have their own transformation hurdles to overcome. For example, Recursion completed a corporate merger in 2024, then cut a number of R&D pipelines with poor data in Phase 2 and earlier stages, and cut 4 more pipelines in 2025. The remaining pipelines in its portfolio are relatively early-stage, and its revenue mainly comes from service income as an AI R&D platform.
However, the overall revenue, whether in terms of scale or growth rate, is far less than market expectations. Its latest second-quarter report shows that the revenue is only USD 14.14 million, down nearly 60% year on year.
Interestingly, Schrödinger, whose core business was drug R&D software services in the early years, is now accelerating the layout of its self-developed pipelines.
Whether in the Hong Kong stock market or the US stock market, compared with the overheating of the primary market, it is not difficult to draw the same conclusion: there is a clear valuation dislocation between the primary and secondary markets in the AI pharmaceutical industry.
The primary market is willing to pay a tech premium for "AI transforming life sciences", while the secondary market still mostly calculates the R&D risk, failure rate and cash consumption of pipelines in the way of traditional biotech.
Delivery and Reversal
Of course, this valuation dislocation essentially stems from the structural differences between the primary and secondary markets. The primary market has a long investment cycle and high risk tolerance, so it can price AI pharmaceutical companies entirely according to the standard of tech stocks — the competition focuses on computing power, models and ecosystems, and the core is to bet on the long-term trend of AI technology revolutionizing pharmaceutical R&D.
The secondary market has a different set of hard constraints. Its buyers are mainly long-term public funds, insurance capital and hedge funds, which are affected by factors such as quarterly financial reports, cash burn rate, and drug R&D pace. More realistically, the Biotech sector has seen successive break of IPO prices and net asset break in the past few years, and the market has long been accustomed to discounting assets that "have no income and rely on stories". Short selling, index weight and liquidity further amplify this caution. In other words, the secondary market does not reject AI, but applies a discount according to the biotech risk framework.
In short, the investment entities, risk appetite, and the level of conviction in the technology of the primary and secondary markets are different. However, price will eventually return to value, and when the delivery stage is approaching, the pricing logic will change accordingly.
Signs have already emerged. After the marketing application of Zasocitinib was submitted, Schrödinger finally received positive feedback from the capital market: from September 14 to 17, its share price rose by more than 45% in four trading days, driving other AI pharmaceutical assets to rise collectively. This is just the beginning.
After all, the AI pharmaceutical industry has passed the early stage where players only tell stories, and the market is starting to demand tangible results: the advancement of R&D pipelines, the verification of clinical success rate, and whether the business model can really work.
That is why the core competition in the AI pharmaceutical industry is shifting from algorithm capabilities to data assets and closed-loop capabilities. Public data (such as the PDB protein structure library) is equally accessible to all, and mainstream algorithm architectures are also spreading rapidly through academic papers and open source code. What truly determines the upper limit of platform performance is high-quality, standardized, and sustainably iterable private data.
This is also why AI pharmaceutical companies are all emphasizing the construction of automated wet laboratories, combining dry experiments and wet experiments: continuously running wet experiments can generate a large amount of high-quality, standardized proprietary data sets for training AI models; and AI models in turn guide high-throughput wet experiments to generate new data to feed back model iteration. Only the dry-wet closed loop can form a self-reinforcing data flywheel.
But it is not easy for the flywheel to spin. High-quality, dynamic training data is still scarce, models in complex systems still highly rely on initial templates, and standardized experimental data requires a lot of time and capital expenditure to accumulate, which cannot be replicated through shortcuts. In other words, closed-loop capability is the right direction, but it is also a threshold that the vast majority of AI pharmaceutical companies have not yet truly crossed.
From this perspective, the main theme of the next stage may no longer be who has the most dazzling model, but who will be the first to deliver tangible results.
Conclusion
It remains unknown when the continuously rising valuation anchor in the primary market will reverse and affect the secondary market. But one thing is certain: as the global AI pharmaceutical industry accelerates its performance delivery, AI pharmaceutical companies in the secondary market are strengthening their platform, model and ecological capabilities, and the valuation dislocation caused by capital mismatch will not exist for a long time.
Delivery and reversal will become the next key theme of the AI pharmaceutical industry. Of course, the flip side of reversal may also be falsification.
In either case, when the tide recedes, the companies that stay standing will definitely be those that have successfully pushed their molecules to clinical trials and the market.
This article is from the WeChat official account "Amino Watch", Author: Amino Jun, published with authorization from 36Kr.