Is AI-driven drug commercialization still immature? Jing Tai Holdings reports its first annual profit, while Yingxi Intelligent remains in loss.

Dubbed China’s “AI drug discovery two giants,” Quotient Holdings (02228.HK) and Insilico Medicine (03696.HK) have each released their 2025 annual reports.

On the same day they released their annual reports, Insilico Medicine also announced a collaboration with Eli Lilly (LLY). Eli Lilly received exclusive sales rights for a GLP-1 drug developed using Insilico Medicine’s AI technology. The upfront payment is $115 million, and the total transaction value could reach up to $2.75 billion.

According to Quotient Holdings’ 2025 annual report, in 2025 it generated operating revenue of RMB 803 million, up 201.2% year over year. Full-year profit was RMB 135 million, and adjusted profit attributable to net profit was RMB 258 million, marking the first time it achieved annual profitability. Quotient Holdings said it became the first profitable Hong Kong-listed company in the AI for Science sector.

In 2025, Insilico Medicine achieved revenue of $56.24 million, including $24.95 million from its drug discovery business, $23.89 million from pipeline development, $4.91 million from software solution services, and $2.49 million from other discovery businesses. Meanwhile, its adjusted loss for 2025 was $43.8 million, compared with a $21.2 million adjusted loss in the same period of 2024, mainly due to a decline in revenue, partly offset by reduced research and development expenses.

A research note from China Post Securities introduced that AI drug discovery refers to combining AI technologies such as NLP (natural language processing), deep neural networks, and generative models with traditional pharmaceutical R&D processes to improve new drug development efficiency and expand the application space for drug innovation. AI healthcare large models are based on technologies such as deep learning and natural language processing, integrate massive biomedical data, and empower the entire drug R&D workflow.

With the rapid development of AI technologies, the market also holds high expectations for AI drug discovery companies. By the end of 2025 and the beginning of 2026, Quotient Holdings’ market value at one point came close to HK$50 billion. After Insilico Medicine listed on the Hong Kong Stock Exchange in December 2025, its share price rose rapidly. It even crossed the HK$40 billion market-cap threshold in February this year, before then experiencing a small pullback.

From the perspective of market space, a research note from China Post Securities said that the global AI-enabled drug R&D expense market is expected to grow from $11.9 billion in 2023 to $74.6 billion in 2032.

When Ding Sheng, Director of the Global Health Drug Discovery Center, was interviewed by the Era Times Weekly, he said that AI drug discovery is still in its early stage. Compared with the massive data relied on by large language models, data scale in the drug discovery field is limited. “Although model architectures and compute power have become fairly mature, our understanding of the underlying mechanisms of life still has limitations, making it difficult to break through quickly by simply increasing compute power. Therefore, the development of AI drug discovery is constrained by data scarcity and insufficient foundational understanding, and overall it remains at an early stage.”

Image source: TuChong Creative

One profit, one loss

In 2025, Quotient Holdings achieved profitability for the first time. Regarding the reasons for turning loss into profit, Quotient Holdings stated in an earlier positive profit forecast that first, revenue surged year over year by at least about 193%, narrowing losses versus the two core segments—intelligent robot solution and drug discovery solution—compared with the 2024 fiscal year. At the same time, the net fair value gains of financial assets measured at fair value through profit or loss increased significantly. In fiscal year 2025, such gains were at least RMB 500 million, up at least about 1876%.

According to Quotient Holdings’ annual report, in 2025, revenue from its drug discovery solution was RMB 538 million, up 418.9% year over year, mainly driven by rapid growth in the antibody business. In addition, service partners and incubated companies’ innovative pipeline development made multiple breakthroughs, reaching interim delivery milestones.

Meanwhile, Quotient Holdings’ AI for Science smart solution revenue also saw rapid growth in 2025, increasing 62.6% year over year to RMB 265 million. This was mainly due to an increase in the number of customers and continued growth from existing business, along with positive results from the expansion of new business directions.

Insilico Medicine, by contrast, faced revenue declines and expanding losses in 2025. In its annual report, Insilico Medicine said its 2025 revenue fell 34.5% year over year, primarily because revenue generated from pipeline development decreased, partly offset by growth in revenue generated by drug discovery. Pipeline development revenue derived from upfront payments in 2025 was $15.3 million, lower than $58.0 million in 2024. This was due to the progress of new transaction negotiations and the development progress of pipelines licensed to external clients.

According to Insilico Medicine’s 2025 annual report, revenue from its drug discovery business rose sharply from $3.14M in 2024 to $24.95M in 2025, increasing its share of total revenue from 3.7% to 44.4%. However, revenue from its pipeline development business fell sharply from $76.59M in 2024 to $23.89M in 2025, reducing its share of total revenue from 89.2% to 42.5%.

From the standpoint of global AI drug discovery industry development, Ding Sheng believes that Chinese companies such as Insilico Medicine and Quotient Holdings are operating at a parallel-racing level globally. The AI drug discovery sector is constrained by inherent limitations in life sciences understanding, and there is currently a certain bottleneck. Therefore, under the bottleneck effect, the gap between companies is relatively small, and latecomers can catch up more easily, while they can also achieve leadership in certain specific areas. If early bottlenecks can be broken, the frontier space ahead will still be very large.

Different business models

In terms of the business models of AI drug discovery companies, a China Post Securities research note explained that, currently, the industry’s business models mainly fall into three types: SaaS, AI+CRO, and AI+Biotech.

Among them, the SaaS model means a company sells software services to pharmaceutical companies or drug R&D CROs to generate revenue. The AI+CRO model uses AI technological advantages to provide outsourced drug R&D services to pharmaceutical companies/CROs and other firms. The AI+Biotech model uses AI to empower its own pipeline Biotech route, generating revenue through ways such as “license out,” collaborations, or advancing pipeline commercialization. The China Post Securities research note believes that the current SaaS business model is not suitable for industry participants.

The business models of these two domestic AI drug discovery giants—Insilico Medicine and Quotient Holdings—are also somewhat different. Quotient Holdings may adopt a model similar to “AI+CRO,” while Insilico Medicine is more similar to the “AI+Biotech” route.

According to Quotient Holdings’ 2025 annual report, the cumulative cooperation amount totals several hundred billion yuan. In 2025, the number of revenue-generating clients increased by 62% year over year. To date, it has covered 17 of the top 20 pharmaceutical companies globally. In 2025, more than five first-in-class new drugs discovered with Quotient Holdings’ involvement released clinical progress, including AI-driven drug discovery platforms covering multiple drug modalities such as small molecules, antibodies, peptides, nucleic acids, molecular glues, and more, which achieved new drug collaboration agreements.

Insilico Medicine stated in its 2025 annual report that its business model mainly includes drug discovery and pipeline development, as well as software solution services. Insilico Medicine invests capital to build a drug discovery and development platform based on generative AI, and establishes product pipelines in areas such as oncology, immunology, and fibrosis.

Insilico Medicine also stated in its annual report that, as of the reporting period of 2025 and up to the latest practicable date, Insilico Medicine has advanced eight projects to clinical progress, including four in-house projects and three collaboration projects, among which there are two clinical Phase II programs, four clinical Phase I programs, and two IND progress milestones.

At the same time, Insilico Medicine has also conducted strategic drug discovery and development collaborations with pharmaceutical companies. According to Insilico Medicine’s annual report, based on 2024 sales revenue, among the top 20 global pharmaceutical companies, 13 have collaborated with Insilico Medicine. In 2025, regarding out-licensing, Insilico Medicine reached an agreement with TaiKang Bio-Technology, granting it exclusive rights to develop and commercialize ISM4808 in the Greater China region. In addition, Insilico Medicine has also reached research collaborations with Eli Lilly, Servier, Yuan Yi Bio, Kangze Pharmaceutical (00867.HK), Qilu Pharmaceutical, and other domestic and overseas companies, covering multiple therapeutic areas.

When discussing the current business models of AI drug discovery companies, Ding Sheng believes that their essence all revolves around BD (business development)—monetizing through transactions along the drug R&D value chain. “In practice, most AI drug discovery companies at the early stage hold visions of creating blockbuster drugs, and tend to develop pipelines in-house to pursue maximum value. But as R&D progresses and delivery difficulty becomes apparent, some companies will also achieve a ‘safe exit’ through early-stage transactions.”

Or yet to reach the level of disrupting the industry

The two AI drug discovery companies have each formed their own unique technical routes. Quotient Holdings stated in its annual report that it has constructed an R&D closed loop of “AI model prediction—robot execution of wet experiments—data feedback to AI—Multi-Agent intelligent scheduling,” thereby forming a new R&D paradigm.

In this system, the AI model is the “expert brain,” responsible for key steps including target parsing, molecule generation, virtual screening, and experimental strategy recommendations. The robot laboratory is the “precise hands,” executing experiments in high throughput 24/7 and accumulating data. Multi-Agent is the “project manager,” autonomously decomposing R&D objectives and scheduling resources across the entire process, building automated iterative R&D modes for fields such as pharmaceuticals and new materials.

Insilico Medicine, meanwhile, developed an AI-driven drug discovery and development platform, Pharma.AI, which provides services ranging from new target identification to small-molecule generation and clinical outcome prediction. Insilico Medicine stated in its annual report that Pharma.AI can identify new drug targets, design molecules from scratch for both novel and established targets, optimize candidate drugs’ clinical development, and simplify the process of drafting academic papers and other related documents.

In its annual report, Insilico Medicine said that using Pharma.AI, on average, the time from target discovery to completion of target validation, lead compound identification, lead compound generation, and lead compound optimization is confirmed to take 12 to 18 months—far shorter than the traditional average of 4.5 years.

At present, will AI technologies that claim to significantly shorten the time for preclinical selection of drugs disrupt the pharmaceutical industry? A research note from China Post Securities analyzed that at the application level, AI-computed “virtual” data cannot replace clinical “real” data. AI still cannot predict how drugs will respond within the human body system, and heterogeneity among individuals in clinical trials will further increase complexity by orders of magnitude.

At the regulatory level, China Post Securities believes that a drug’s safety is the primary factor considered by regulators, and clinical trials are the only source of safety proof. In the long run, there is no substitute, and the approval process for drug development will not change over the long term. Therefore, the fundamental experimental-science-based nature of drug discovery will not be disrupted just because AI undergoes iterative upgrades.

Ding Sheng also believes that current AI drug discovery technologies have not yet reached a disruptive level. Although relevant companies often claim that R&D efficiency improves by multiple times, such statements may be one-sided.

“A drug molecule must satisfy dozens of attributes in order to become a candidate compound that can enter human clinical research, including interaction with the target, off-target selectivity, administration routes, tissue distribution, metabolic pathways, toxicity, and more. Improving prediction efficiency for only one of these attributes has limited significance for shortening the overall R&D cycle. Therefore, over the past five years, the drug R&D pace has not accelerated significantly due to AI technology; instead, it has slowed as R&D standards have increased. Whether future disruption can be achieved remains to be observed.” Ding Sheng analyzed.

Abundant information, precise interpretation—everything on Sina Finance app