The pharmaceutical industry has been discussing AI for more than a decade. In the past, skepticism and wait-and-see attitude were the prevailing sentiments. However, in this early spring of 2026, the winds have shifted dramatically.

Ren Feng, Co-CEO and Chief Scientific Officer of Insilico Medicine, pointed out in an interview with Xieyi Jun that the pharmaceutical industry's attention and expectations towards AI have reached an unprecedented level. In the past, discussions often revolved around whether AI could truly empower research and development, with many holding a cautious attitude; now, it has become a common consensus that "AI will profoundly transform drug research and development," and the focus of discussion has completely shifted to "when and on what scale" this transformation will occur.

The consensus shift is backed by the successive implementation of key clinical advancements.

The most typical case comes from Insilico Intelligence, whose self-developed molecule Rentosertib (ISM001-055) has completed Phase IIa clinical trials. This drug, used to treat idiopathic pulmonary fibrosis, is the world's first investigational drug discovered based on artificial intelligence (TNIK) and designed by generative AI. It took only 18 months from target discovery to the nomination of preclinical candidate compounds, with a cost of approximately $2.6 million, significantly demonstrating the potential of AI in speeding up, reducing costs, and innovating.

Dereui Zhiyao has also made significant progress. In December 2025, Dereui Zhiyao announced that its new oral weight loss drug MDR-001, developed with AI-assisted design, had officially initiated Phase III clinical trials, becoming China's first AI-designed drug to enter this stage, and also one of the few AI drug assets globally to reach Phase III clinical trials.

"The core value of AI goes far beyond cost reduction and efficiency improvement. It lies more in systematically enhancing the success rate of research and development, as well as tackling traditional challenges in new drug research and development," Niu Zhangming, founder and CEO of D-Rui Zhiyao, pointed out to Xieyi Jun. Some of the core advantages of D-Rui Zhiyao's AI platform lie in its ability to predict the effectiveness and safety risks (such as toxicity issues) of drug molecules through data models in the early stages of research and development, and to provide certain interpretability functions. This can help avoid or reduce failures that occur after entering the later clinical stages, which require significant investment of funds and time.

Regulatory agencies are also embracing this transformation with unprecedented proactivity. The Center for Drug Evaluation and Research (CDER) of the US Food and Drug Administration (FDA) established a dedicated AI committee in 2024; in 2025, the FDA even released the "Draft Guidance on Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drugs and Biologics", aiming to coordinate and promote the standardized use of AI in drug evaluation.

"Like many cutting-edge technologies, breakthroughs in AI-based drug development are a continuous iterative process," Ren Feng pointed out. "If the achievements of the past few years were more concentrated in certain mature therapeutic areas, then in the past year or two, we have seen significant breakthroughs in two key dimensions."

Firstly, there is an extension in modality. In the past, the application of AI in pharmaceuticals was highly concentrated in the small molecule field. Nowadays, antibodies, peptides, oligonucleotides, and even more complex ADC (antibody-drug conjugates) fields have all begun to see the active presence of AI. "An intuitive feeling is that many newly established AI pharmaceutical companies are focused on using AI to design macromolecular drugs," said Ren Feng. "The scope of AI empowerment is rapidly expanding.". ”

In 2025, Eli Lilly partnered with Creyon Bio, aiming directly at AI-driven targeted RNA oligonucleotide therapy; while Mavbio collaborated with Insilico Intelligence, aiming to empower every aspect of ADC drugs with AI, from novel toxin development to new target discovery.

Tushen Zhihuo is also one of the pioneers in this new field. Its self-developed AI-integrated protein design platform can modify and de novo design biological macromolecules such as enzymes, antibodies, and peptides.

Wang Yuguang, founder of Tushen Zhihuo, told Xieyi Jun that the reason for choosing to focus on antibody design is its stronger universality and easier standardization. Monoclonal antibodies or bispecific antibodies designed for different targets and diseases can be iterated and optimized within the same system, thereby significantly reducing capital and time costs, and also making it easier to demonstrate technological advantages.

"Of course, whether we can successfully launch high-quality candidate molecules ultimately depends on the capabilities of artificial intelligence technology itself. The 'dry-wet closed-loop' iterative system that we have adopted, which combines artificial intelligence with automated laboratories, not only efficiently produces effective results but also generates a large amount of high-quality data. As long as we maintain a certain throughput, these data will continuously improve model performance, forming a data flywheel effect, making the model increasingly stronger in designing effective molecules."

Next is the deepening of the technological core. Ren Feng believes that the development of underlying technologies such as large language models has brought about a key change. "With the evolution and iteration of algorithms and models, AI's 'reasoning' ability is being enhanced and increasingly integrated into drug research and development. Nowadays, AI-based pharmaceuticals are more integrated with data analysis and complex reasoning, giving the system the potential for 'self-evolution'. ”

In early 2026, the large language model training framework Science MMAI Gym, released by Insilico Intelligence, represents an attempt in this direction. It aims to enable general-purpose large models to understand complex scientific language and logic, thereby handling drug discovery tasks more professionally.

After AI proves its value in the preclinical stage, conquering the "deep waters" of clinical trials becomes the logical next step.

A survey of PhRMA member companies shows that the cost of clinical research accounts for approximately 69% of the total drug development costs for pharmaceutical companies. With the intervention of AI, this figure may gradually decrease.

For example, Unlearn.AI has launched "digital twin" technology, which can create virtual controls for each clinical trial patient, thereby reducing patient recruitment needs by 30%.

A study released at the 2025 ASH Annual Meeting showed that in a trial for the rare disease polycythemia vera, Dyania Health's AI platform, Synapsi, could complete a comprehensive assessment of over 900 patients within a week and identify 22 eligible patients, whereas traditional methods would only allow for the enrollment of one eligible patient every four months on average.

According to research by Define Venture, the attitude of large pharmaceutical companies towards AI has shifted from "pilot" to "strategic focus". About 85% of the surveyed business leaders are increasing their investment in AI. More crucially, despite the industry's traditional emphasis on independent research and development, only 30% of companies now adhere to a strategy of complete in-house research, with the majority preferring a hybrid or external collaboration-first approach.

Andrew Marshall, a partner at Haystack Science, analyzed and pointed out that although large pharmaceutical companies have gradually established their internal AI infrastructure, they still crave external "solutions", especially those that can integrate multimodal datasets.

Under this trend, a new division of research and development is emerging, where pharmaceutical companies outsource early-stage discovery projects to specialized AI biotech companies, while focusing more on clinical development, regulatory affairs, and commercialization. Even some mid-sized biotech companies are beginning to introduce external AI platforms to compensate for the shortcomings of their own technical teams.

In summary, AI has gradually penetrated the entire chain of drug research and development, from target discovery, to molecular design, and then to accelerating clinical research and development. This is an evolutionary process from point to line, and then to area. When capital, technology, and industrial demand resonate in multiple dimensions, AI is redefining the entire path of drugs from the laboratory to the hands of patients.

On the global map of AI pharmaceuticals, China and the United States constitute the main axis of competition.

According to the "IFF Global AI Competitiveness Index Report" released by the International Financial Forum (IFF) and statistics from the World Intellectual Property Organization (WIPO) database, China and the United States have gathered nearly 70% of the world's AI talents and produced nearly 70% of AI technology patents.

So, what exactly is China's differentiated competitive advantage in this top competition?

The most intuitive rigid advantage of supply chain and cost efficiency.

Although many biotechnology processes are complex, the relatively complete supporting facilities in China can support the connection and integration of various links from research and development to production. This easily integrated ecosystem is a significant feature of the Chinese market. At the same time, in terms of cost control, we are also competitive compared to developed overseas markets. The dual advantages of supply chain and cost have prompted more and more large pharmaceutical companies to be willing to pay for it, "said Wang Yuguang.

If supply chain and execution efficiency are hardware advantages, then the accumulation of digital infrastructure and scientific research talents constitutes the driving force behind the rapid iteration of AI pharmaceuticals.

Niu Zhangming, CEO of Deruizhi Pharmaceuticals, believes that the advantage of AI Pharmaceuticals in China is first reflected in the comprehensive cost of data generation and use. The cost of generating new data through the pre clinical CRO industry of Chinese biopharmaceuticals is significantly lower than that of foreign peers. This advantage has been verified in the Internet field. The rise of many leading Internet enterprises in China benefits from this cost efficiency advantage.

In addition, China's advantages are also reflected in two aspects. Firstly, robust digital infrastructure, sufficient energy supply, and computing power layout ensure the high-intensity and low-cost computing power required for AI research and development; The second is a profound talent and research accumulation. A large team of top scientists and engineers can continuously drive the iteration and optimization of underlying algorithms, providing support for the evolution of AI models.

These advantages work together and ultimately internalize into a core competitiveness, which is a fast iteration efficiency far exceeding that of peers.

We have a significant advantage in advancing the 'wet experiment' compared to overseas, "Ren Feng pointed out." Therefore, our overall efficiency in achieving the 'wet dry closed-loop' is likely to be higher. "This means that from algorithm design to molecular verification in the laboratory, China's conversion and feedback links are shorter and faster, which directly translates into time and decision advantages in the highly data-driven research and development model.

Ultimately, all advantages need to be based on the ability to translate technology into practical solutions. Although we may have gaps with overseas counterparts in some cutting-edge basic models, Chinese people have done a very good job in technology application. We are good at flexibly and effectively applying these technologies to specific real-life scenarios, "said Ren Feng.

In the past, the process of discovering a drug target and eventually developing it into a drug was often measured in units of "ten years", which was full of uncertain and long waiting times. Nowadays, candidate molecules designed and screened by AI in the laboratory are moving towards clinical practice at a faster pace and clearer path. This is not only about speed, but also indicates that the research and development process itself is becoming clearer and more controllable than ever before.

The true division also unfolds from this.

As the technological dividend gradually spreads, the core of future competition will no longer be whether to use AI or not, but how to control AI. The winner of the future lies in who can use AI to crack complex disease mechanisms first, and who can truly transform algorithms into one successful new drug after another.