Insilico Medicine Pioneer JingTai Holdings Achieves First-Ever Annual Profitability, Validating Platform-Driven AI Drug Discovery Model

2024-2025: AI Enters the True "Year of Explosion." But this explosion is not romantic:The iteration speed of large models is getting faster and faster, with increasing investment in computing power and talent. The industry is rapidly shifting from "competing on who can build better models" to "competing on who can manage the costs more effectively."

For almost all AI companies represented by large models, the key to competition is becoming increasingly clear——First, whether the cost can be controlled, and second, whether the revenue can be continuously increased.

And when AI enters hard-tech fields such as pharmaceuticals and materials, these two questions will become even more challenging:Data is scarcer, experiments are more expensive, validation takes longer, and failures are more common.As a result, AI pharmaceuticals have long been regarded as one of the fields with "enormous potential but the most challenging commercialization."

Against this backdrop, XtalPi Holdings Limited, a domestic AI pharmaceutical platform company, announced a positive profit forecast for 2025:

Revenue increased to no less than 780 million yuan, with a year-on-year increase of at least 193%; both post-tax profit and profit attributable to the company's equity holders were no less than 100 million yuan, turning from a loss to a profit year-on-year, achieving full-year profitability for the first time, and also becoming the first AI application stock in Hong Kong stocks to achieve profitability.

At the same time, global pharmaceutical companies' attitude towards AI has shifted from "ornamental innovation" to "foundational investment": Sanofi declared "All in AI," Eli Lilly partnered with NVIDIA to invest $1 billion in building an AI lab, and AstraZeneca secured the rights to an innovative long-acting peptide drug, developed with AI involvement, from CSPC Pharmaceutical Group for up to $18.5 billion. Industry consensus is increasingly unified: AI is no longer just a bonus feature but a fundamental capability to avoid obsolescence.

But behind the pharmaceutical companies' full embrace of AI lies the ongoing differentiation in the AI drug development industry. Overseas veteran "SaaS" player Schrodinger has fallen into a dual slump in performance and stock price due to its transition to self-developed research; Recursion, a representative of the self-developed pipeline, has seen its losses widen and was completely divested by NVIDIA.These cases collectively point to one reality: the difficulty of AI drug development has never been about "whether it can be done," but rather "whether revenue can be continuously generated under controllable costs" and "when it will truly start making money."

By contrast, XtalPi's profitability is therefore more worthy of discussion:Why does it have an advantage in the phase of AI arms race that tests cost and revenue models the most?

01 Route Controversy: Building "Industrial AI Infrastructure" or Betting on Pipeline

While other AI pharmaceutical companies are still discussing when their pipelines will pass the ultimate "验收" of Phase III clinical trials, XtalPi has already put the money earned from AI drug discovery into its pocket. The profitability of AI in drug discovery has been validated earlier than the R&D pipelines themselves. Although XtalPi's business model appears complex, its core essence can be summarized in one sentence:It is more like an "enabler" of industrial AI infrastructure, rather than a "gambler" who personally enters the field to make drugs.

XtalPi does not focus on self-developed drug pipelines as its core. Instead, it empowers collaborators’ R&D through an “AI + automated R&D platform” and generates revenue via upfront payments, milestone payments, technical service fees, platform subscription fees, and profit-sharing from certain projects. Therefore, it is difficult to define with a single label. However, it is certainly not a traditional CRO, nor a pure SaaS company, and even less a typical self-developed biotech firm. Traditional CROs are human-centric and do not develop as many algorithms and hardware as XtalPi does.

If we trace back to the company's past performance releases, we can see that the company's long-term goal is to become a super artificial intelligence in vertical industries.The market "can't understand" the company, essentially because its business model lacks a strictly comparable overseas example. However, many great companies, including Google, faced a similar situation in their early days where most people "couldn’t understand." As is well known, the core of artificial intelligence still lies in proprietary data. The purpose of hardware and operations directly points to data, and the purpose of data is to achieve XtalPi's desired "industry large model."

If XtalPi only talks about technology, we would actually be somewhat concerned.The management team also places great emphasis on commercial implementation and financial revenue. From the very beginning, XtalPi was positioned as a "platform," and in its development, it found a unique and advantageous "platform-based" business strategy, eventually becoming the first company in AI drug discovery to generate significant profits.

On the one hand, it deeply engages in innovative R&D through platform capabilities, and by sharing data, it also has the opportunity to share in the long-term upside of the innovation pipeline; on the other hand, it can secure sustainable cash flow with relatively controllable risks.

The barrier comes from the combination of reusable system capabilities – quantum physics computing, AI design, robot experiment-driven data closed-loop, and a multimodal R&D platform covering both small and large molecules – this is not something that can be replicated simply by "piling up manpower and computational power."More crucially, it avoids the most fatal uncertainty for AI pharmaceutical companies: the high clinical risks and cash flow black holes associated with self-developed pipelines.XtalPi does not rely on "betting on Phase III clinical trials" to prove its value, but rather continuously validates its platform capabilities through orders, upfront payments, milestone achievements, and revenue sharing.

At the same time,Platform-based businesses have scale effects: As models and processes continue to iterate and data continues to accumulate, XtalPi can strengthen its leading advantage through extensive cooperation.AI Empowering Scientific Research and Industrial Development is a Definite Trend with a Huge Incremental Market, and XtalPi, a Platform Already Validated and Selected by 17 of the Top 20 Global Pharmaceutical Companies, Possesses Greater First-Mover Advantages and Lower Customer Acquisition Costs.

AI Empowers Pharmaceutical R&D, Enhancing Efficiency, but the Core Lies in Whether It Can Form an Effective Business Model Rather Than a Money-Burning One.

Although all roads lead to Rome, different choices made by different companies can still lead to different outcomes. The AI pharmaceuticals industry faces a structural issue:If taking the "self-developed pipeline" route, the true validation often doesn’t come until the later stages of clinical trials, particularly in larger-scale, statistically more significant Phase III trials.

Thus, a common script often plays out in the market: In the early stages, companies rely on "model narratives + pipeline quantity," targeting high-potential indications, sending valuations soaring; in the mid-term, pipelines begin to enter clinical trials, cash flow pressure mounts, yet verifying the pipelines takes a long time. If financing falls short, tough trade-offs loom; once key clinical trials fail, stock prices and fundamentals collapse simultaneously. In recent years, there have been no shortage of typical examples overseas.

Schrodinger, with its stable software service revenue, as a pioneer in the overseas AI pharmaceutical field, initially focused on drug discovery software services. Supported by steady subscription income, the company developed steadily, and its stock price remained robust over the long term. However, after increasing investment in self-developed drugs and transitioning to an "AI+Biotech" model, its performance fell into continuous fluctuations. Despite significantly higher R&D expenses, clinical results have been slow to materialize, putting sustained pressure on both performance and stock price.

Recursion, which was recently divested by NVIDIA, is also a typical representative of AI biotech, showcasing the drawbacks of the self-developed pipeline model to the fullest extent.

After completing the corporate merger in 2024, Recursion cut a number of R&D pipelines with poor data from Phase 2 and earlier-stage clinical trials. In 2025, the company eliminated four more pipelines, retaining relatively earlier-stage pipelines in its portfolio. Its revenue mainly comes from service income as an AI R&D platform.

From Recursion's pipeline adjustments, it is evident that the model of independently developing drug pipelines will be difficult to validate for its commercial value in the short term.In some therapeutic areas, even if the pipeline is destined to fail to reach the market, it may still successfully pass through Phase I and II clinical trials. Before the results of the larger-scale, more patient-rich Phase III clinical trials are revealed, it essentially remains a high-risk gamble. The fact that even NVIDIA chose to exit reflects the market's risk assessment of "self-developed mega-gambles."

02 Moving the "Proof Point" Forward: Why XtalPi's Model is Easier to Validate Financially

By contrast, the key difference in XtalPi's model is that it has moved the "validation point" forward.

In the collaboration model, XtalPi's value validation does not need to wait until Phase III — signing the contract and receiving the upfront payment means that the client is willing to pay for the platform's capabilities (rather than a "try-and-see" small Proof of Concept — PoC).The advancement of its participating pipeline into clinical trials to achieve milestones indicates that the platform's delivery has driven real research and development progress. If there are subsequent sales or rights sharing, it will provide longer-term upside potential.

More importantly, the cost structure of milestone payments is more favorable.When the project enters the clinical stage, the major risks and investments in subsequent clinical trials are mainly borne by pharmaceutical companies; the milestone revenue obtained by XtalPi often does not require a proportional increase in costs, thus being closer to high-margin incremental profits.

This gives XtalPi an edge in both "cost control" and "revenue growth.": On the cost side, automation and platformization bring economies of scale, reducing marginal costs; on the revenue side, upfront payments/milestones allow for faster and more sustainable income realization.

Currently, the company has officially announced dozens of collaborative pipelines. Its clients include global leading pharmaceutical companies such as Eli Lilly, as well as innovative biotech firms, covering a wide range of pipelines from cancer, metabolic diseases to neurological disorders and rare diseases.The larger the pipeline scale, the deeper the data and algorithm沉淀, the stronger the platform, forming a positive cycle of "becoming cheaper to operate, more accurate, and increasingly capable of securing large contracts."

Moreover, XtalPi is highly favored by innovative biotech companies that are "small and beautiful." If collaborators choose to out-license pipelines at an earlier stage, the timing of XtalPi’s revenue-sharing may be further advanced, thereby increasing its revenue potential.

If "AI empowering scientific research and industrial implementation" is the prevailing trend, then XtalPi's positioning is more like the infrastructure for industrial AI applications:Able to tap into a large market while isolating risks effectively and retaining predictable revenue on their own balance sheet.

As a result, over the past year, national-level inspections and investigations have frequently taken place at XtalPi, indirectly indicating two facts:The track is highly valuable, and leading companies are forming a competitive edge.

03 Where Does the Moat Come From: Not the Model, but "Data Closed-loop + Automated Production Capacity"

In AI applications, what is truly scarce has never been "having a model," but rather "the ability to continuously produce high-quality data and transform that data into reusable assets." For XtalPi to become an industry enabler, the key lies in whether its capabilities are strong enough and its barriers deep enough.

XtalPi's moat concept is to transform algorithms, automation, and data into a new generation of infrastructure, known as the full-stack technology loop of "quantum physics + AI + robotics."

This is a heavier, more difficult, but more effective combination that addresses two key issues: an automated experimental platform that solves the pain points of data accuracy and data silos in the pharmaceutical field; and data feedback driving AI algorithm iteration, forming a closed loop of "delivery—data—model—re-delivery" to continuously strengthen the model.

As everyone knows, the core three elements of AI are computing power, algorithms, and data. The computing power aspect is basically provided by upstream sources, making it difficult to create a gap. The key factors that determine the moat, or create a gap, are "data + algorithms."

This is also where the "implementation challenges" of AI drug discovery lie: many companies have seemingly powerful models, but lack a sustainable supply of high-quality experimental data and a closed-loop iteration mechanism, ultimately tending to remain at the PoC stage.

A major challenge AI faces in scientific research is how to obtain sufficient high-quality data to train effective models.Even the popular AlphaFold is limited by insufficient complex structure data in drug discovery scenarios, impacting its practical application effectiveness. The March 2025 report in *Nature* also mentioned that AlphaFold faces a shortage of drug-related data, which directly affects model performance and hinders the advancement of this tool in relevant applications.

Therefore, an excellent AI pharmaceutical enterprise must possess two major characteristics:The ability to efficiently acquire or independently generate high-quality data, as well as the capability to complete validation through a mature experimental platform and continuously achieve algorithm iteration in a closed loop.

This is the core highlight of XtalPi's "quantum physics + AI + robotics" model. Relying on its self-built robotic laboratory R&D platform, it addresses data issues from the source and continuously transforms its operations into data assets and algorithmic assets, feeding back into model optimization.A positive cycle of "business development - data accumulation - algorithm iteration - service upgrade" has been formed, building up a differentiated advantage that is difficult to replicate.

04 Can Growth Be Sustained

Of course, after turning in a profitable report, what the market cares most about is whether XtalPi can continuously run through the "technical service + joint R&D" light asset model to bring about consistently high growth performance?

It still takes time to verify, but from the current industry practice and cooperation progress, the company has given positive answers in at least five core dimensions:

First, the continuous extension capability of New Modality.

From the perspective of industry development patterns, the R&D bottlenecks of single-drug modalities are becoming increasingly prominent, and multi-modal drug development has become a crucial direction for overcoming drug development challenges and expanding therapeutic boundaries. For AI-driven pharmaceutical companies, tackling highly challenging targets that traditional R&D struggles to address through more diverse drug development pathways is key to validating their technology and algorithmic capabilities.

Currently,XtalPi has established an integrated R&D capability covering multiple types of drugs, including small molecules and antibodies, and is rapidly expanding.It has already driven the implementation of platforms such as molecular glue, mRNA, peptides, and siRNA, fully demonstrating its deep technical accumulation and innovative strength, which also corresponds to new market opportunities and blue ocean spaces.

Among them, siRNA, as the core细分领域 of small nucleic acid drugs, is experiencing a historic leap from "niche rare disease treatment" to "mainstream chronic disease therapeutic solution," standing at the C位 of the global innovative drug wave with tremendous market growth potential.

Frost & Sullivan data shows that the global small nucleic acid drug market size is expected to surge from USD 5.2 billion in 2024 to USD 20.6 billion in 2029.XtalPi's AI Small Nucleic Acid Design Platform, XtalSilence, can design innovative siRNA molecules that significantly reduce off-target risks through AI iteration. These molecules precisely meet industry technical demands and are expected to seize the initiative in this rapidly growing field, leveraging technological advantages to share market dividends.

The potential of molecular glues is equally impressive. As a new generation of proximity-induced therapy, they, together with PROTAC, form the core direction of protein degradation technology. This area represents the most imaginative field of new molecules in the medium to long term and is also the key to unlocking the treasure trove of "undruggable targets." It is a blue ocean market with clinically and commercially validated value.XtalPi, on the other hand, tackles the core technical challenges in molecular glue research and development by leveraging physical computational simulations, high-throughput screening, and proprietary predictive tools, combined with an AI-driven robotic laboratory, to build a differentiated competitive advantage.

While tackling more challenging targets through additional drug development pathways, XtalPi can even conduct internal competitions targeting different mechanisms and drug modalities for the same disease, then advance the best-performing molecule into clinical trials.This is also one of its core advantages that distinguish it from other AI pharmaceutical companies, which helps continuously broaden the technical boundaries and cooperation scenarios.

Second, the ability to continuously secure large orders.

From XtalPi's nearly $6 billion scaled package order with DoveTree to the deepened cooperation totaling $600 million in two phases with global leading pharmaceutical companies such as Eli Lilly, continuously securing major projects demonstrates that the platform’s value has gained high international recognition. Its capability to secure large orders is sustainable and no longer relies on a single project for growth.

Third, the cooperative pipeline is proceeding efficiently with a high level of fulfillment.

After receiving technical empowerment from XtalPi, the partner has made smooth progress in pipeline R&D, with milestone nodes being achieved intensively.

For example, over the past year, its collaborative achievements have been remarkable: the project with Intelligene Biotech has successfully entered the clinical stage; RTX-117, the AI+RNA new drug co-developed with Relay Therapeutics, has successfully administered the first patient dose, marking the project's entry into a critical clinical validation phase; the gastric cancer-targeted drug in collaboration with Signet Therapeutics not only entered clinical trials smoothly but also received a nomination for the Galien Award, highlighting the clinical value and industry recognition of the pipeline; additionally, with technical support from XtalPi, Layman Bio has achieved 100% complete remission in blood tumors and lupus erythematosus using CAR-T at one-thousandth of the standard dose, and recently completed nearly 200 million yuan in financing.Multiple collaborative pipelines simultaneously advancing to the clinical stage have, to a certain extent, validated the genuine efficiency and core drug development value empowered by XtalPi's AI technology.

Fourth, relying on the vast application space of AI, continue to expand the second and third curves.

XtalPi is not merely an AI pharmaceutical company; it has also expanded into incremental markets such as consumer goods, new materials, and agriculture.Especially successful in entering the consumer health field, XtalPi has successfully designed two innovative topical active ingredients for hair growth based on its self-developed AI molecular development platform ID4 — the small molecule Remeanagen™ and the peptide AquaKine™.

The test data showed that subjects using the two AI-designed hair growth molecules could observe a preliminary increase in anagen follicle density as early as 14 days after use, with a rapid onset of action. After 45 days of use, over 90% of subjects observed a visibly noticeable increase in hair density; the average number of hairs shed by subjects decreased by 33% to 45%, and safety performance was excellent.

Currently, both of these AI-designed hair growth molecules have successfully passed INCI (International Nomenclature of Cosmetic Ingredients) registration, and their combined formulation has been completed by the externally incubated consumer brand Groland for FDA cosmetic filing, meaning that the product can now be sold through e-commerce channels.

The market size for hair growth products goes without saying. And"The strategy of 'developing consumer products with pharmaceutical standards' has long been officially endorsed by Eli Lilly's weight-loss wonder drug Tirzepatide, becoming the biggest driver propelling pharmaceutical companies toward a trillion-dollar market value."XtalPi's successful transition from a pharmaceutical-grade R&D platform to consumer-level functional ingredients has also brought new room for imagination and growth expectations. If this sector develops appropriately,Will Bring Explosive Growth to XtalPi Faster Than Drugs, which is worth paying close attention to.

Fifth, rely on cash reserves to accelerate business and technology expansion.

In the AI industrialization phase, "cash flow and balance sheet" has become one of the core competencies, representing the company's confidence to continuously invest in R&D, expand the market, and maintain a competitive edge.Since its listing, XtalPi has bolstered its cash reserves through multiple rounds of placements and zero-coupon convertible bonds, with the issuance of zero-coupon convertible bonds maturing in 2027 reaching HKD 2.264 billion, indirectly demonstrating its appeal among institutional investors.

This also means that XtalPi has nearly 10 billion in cash reserves.，It can quickly fill in technical modules with "buy-buy-buy," expand the list of overseas customers, build new business teams or JVs, including investing in promising upstream and downstream companies, converting its super healthy cash reserves into higher profits and competitive barriers. How to seize these opportunities, XtalPi, with sufficient cash in hand, has more initiative compared to its peers.

05 Conclusion

Back to the original question: What does XtalPi's profitability tell us?

It shows that in the AI arms race, what is truly scarce is not "the ability to build models," but whether one can simultaneously address two issues in real-world industries:Reduce costs and increase revenue.

XtalPi, as a "new species," combines the imaginative potential of AI-driven drug discovery with the vast market of large-scale models, while also avoiding the risks associated with self-developed pipelines through a platform-empowerment model.Shift value verification to the upfront payment and milestone stages, and continuously accumulate data and algorithms through the closed loop of "quantum physics + AI + robotics" to create a self-reinforcing scale effect.

It remains to be seen over time how large XtalPi can grow and how far this model can go. However, it is undeniable that AI-powered research and its industrial implementation are the inevitable trends. As the first AI application stock in Hong Kong's stock market to achieve profitability, XtalPi’s value continues to rise.

The core lies in,XtalPi has successfully broken out of the "burning money without profit" dilemma in the AI pharmaceuticals industry, validating the commercial feasibility of AI technology in empowering new drug research and development with solid orders and pipeline progress., which is the source of its long-term competitiveness, and also paves a replicable and referential implementation path for the entire industry.

This article comes from the WeChat Official Account"Amino Observation", Author: Anji Jun, published with authorization from 36Kr.