NVIDIA's Investment in 12 AI Drug Discovery Startups: Is the Warmth Reaching Chinese Peers?

Text | Ling Xin

Compiled by | Wang Xiao

A Chinese research team has used artificial intelligence (AI) to screen for new biomarkers of Alzheimer's disease (AD), which are superior to the two previously classic biomarkers and have high clinical application value.This research achievement has been published in *Nature Human Behavior* on July 10, 2024.

This is the result achieved by a team led by Professor Jincai Yu from the Department of Neurology at Huashan Hospital, Fudan University, in collaboration with the Feng Jianfeng/Cheng Wei team from the Institute of Science and Technology for Brain-Inspired Intelligence, Fudan University, after analyzing and modeling the largest high-throughput cerebrospinal fluid proteomics data (6361 proteins) to date.The novel biomarker YWHAG achieved accuracies of 96.9% and 85.7% in identifying biologically defined AD and clinically diagnosed AD dementia, respectively.

There are approximately 50 million Alzheimer's patients globally, with China accounting for about one-quarter. The nerve damage suffered by Alzheimer’s patients is irreversible, and existing drugs can only slow the progression of the disease or alleviate symptoms. Therefore, "early detection" is key to treating the disease and underscores the value of the aforementioned research findings.

The research team's AI-screened achievements, once released, significantly boosted confidence in China's AI. A person related to pharmaceutical investments in the secondary market stated, "There is no longer a need to discuss whether AI is promising; everyone knows it is the future."

The former president of Stanford University, who is far away in the United States, has also founded an AI pharmaceutical company, Xaira Therapeutics, which raised 1 billion US dollars three months ago, topping the global second-quarter financing list. This AI pharmaceutical company aims to use AI to reshape drug research and development. Currently, Xaira Therapeutics is building a drug discovery and development platform, which will advance multiple drug projects.

AI, as a tool, is increasingly being used by researchers in clinical settings, and investors' hesitation over the past two years is fading.In the first half of 2024, there were 69 financing deals in the global AI pharmaceuticals sector, basically approaching the same level as in 2022; with an investment amount of $3.336 billion, which, although still a certain gap from $3.851 billion in the same period of 2022, was more than twice that of the same period in 2023.

Among the top ten highest financing deals globally in the first half of the year, a Chinese company appeared: METiS TechBio ranked ninth with 100 million US dollars.In China, there were 22 AI pharmaceutical financings in the first half of 2024, a significant increase from 15 and 16 in the same period of the previous two years; the financing amount was 1.809 billion yuan, also an 11.6% increase compared to the same period in 2023.

NVIDIA, the world's most watched publicly traded company, is a key driver behind the recovery of AI pharmaceutical financing.Since 2023, NVentures, the investment arm, has invested in at least 12 AI pharmaceutical companies. Founder Jensen Huang stated that the era of studying computers has passed, and life sciences represent the future.He even hopes that the use of AI can transform "drug discovery" into an engineering field, rather than just a life science.

"Now everyone expects AI to bring about disruptive changes and create an innovative drug that humans have not achieved. Before this, investment institutions and MNCs (multinational pharmaceutical companies) may invest in some as trials every year," the aforementioned person related to secondary market pharmaceutical investments believes. This is the reason for the "rebound" and "sustained interest" in AI drug development.

Well-known biopharmaceutical venture capital firms such as ARCH Venture Partners and Sequoia Capital, as well as multinational pharmaceutical companies like Bayer and Eli Lilly, have all invested in AI-driven drug discovery companies in 2024.

“AI pharmaceuticals have indeed experienced a certain degree of recovery."Huang Xiao, Managing Director of Yunfeng Fund, analyzed that the premise of this change is the further enhancement of AI models and computing power, as well as the successive entry of AI pharmaceutical products into clinical stages, making the prospect of AI-assisted new drug development more verifiable."

As of May 2024, there are more than 70 AI pharmaceutical R&D pipelines that have entered clinical trials globally.According to Boston Consulting research, AI-generated drug molecules have a success rate of 80%-90% in Phase I clinical trials, compared to the historical average of about 50%. In Phase II clinical trials, the success rate is 40%, which is in line with historical levels.

Based solely on the data, up to Phase II clinical trials, the success rate of AI-driven drug discovery has become comparable to human efforts.

“Especially after the emergence of the new generation of AlphaFold, there is more data and more accurate predictions regarding the correlation between the sequence, structure, and function of biomacromolecules, including nucleic acids and proteins."Huang Xiao said."

In May 2024, Google DeepMind and its sister company Isomorphic Labs jointly launched AlphaFold 3, which made the cover of Nature. DeepMind stated that for interactions between proteins and other types of molecules, AlphaFold 3's accuracy in benchmark tests is 50% higher than the best existing traditional methods.

The interaction of proteins with other molecular types is key to drug discovery and development.When AI successfully breaks through in this field, what Jensen Huang referred to as "life engineering" seems truly possible.

No one knows who will be the lucky one to first use AI to create a "truly innovative drug," just as no one knows who can truly use the spinning jenny to weave the first piece of cloth. But it is this allure that has reignited the investment community's hope in AI drug development and led multinational pharmaceutical companies to continuously engage in project collaborations with AI-driven drug discovery.

According to the statistics from Yaozhi Network, there were over 30 publicly disclosed collaborations between pharmaceutical giants and AI companies in 2023 alone.

In 2023, the large pharmaceutical company Sanofi had the most publicly announced AI collaboration projects, partnering with four collaborators on research such as AI + target discovery and AI + small molecule innovative drugs. One of these was a $1 billion deal with Chinese company BioMap to jointly develop models for biologic drug discovery based on the latter’s large model.

The largest collaboration between multinational companies and AI firms in 2023 came from Novo Nordisk. It reached an agreement with Valo Health to research and develop new therapies for cardiometabolic diseases based on Valo's human datasets and AI computational capabilities. Valo will receive an upfront payment totaling $60 million along with potential near-term milestone payments, and is eligible for up to $2.7 billion in milestone payments, as well as R&D funding and potential royalties.

There are more undisclosed large-scale AI pharmaceutical cooperation projects.Pan Lurong, founder of Circle One Wisdom, revealed that Pfizer from the U.S. and MBL Life Science from Japan have collaborated with them on the de novo design and optimization of antibody reagents, and they have also partnered with Bayer to optimize the activity and selectivity of nanobody protein drugs.

"Most of the demand comes knocking on the door." The clients of CircleOne Intelligence mainly come from North America. Pan Lurong found that the vast majority of large pharmaceutical companies and innovative biotech enterprises in North America are using AI for drug screening, whether through internal development or external collaboration.

Chinese pharmaceutical companies have broader demands, hoping that AI can become a solution for product initiation, target discovery, patent background search, molecular design, and R&D outsourcing.Many Chinese pharmaceutical companies are considering whether to build their own AI departments.

Circle One Wisdom's AI pharmaceutical cooperation services were relatively few in 2023, but there are already dozens ongoing in 2024. Moreover, "all of them involve upfront payments," said Pan Lurong.

When pharmaceutical companies disclose AI drug discovery developments, the wording is often quite vague due to projects being in early stages or a desire to "quietly impress everyone." They only reveal information like “oncology drug research” which gives no real insight. Pan Lurong also confirmed that many collaborations with large pharmaceutical firms require signing confidentiality agreements, meaning “targets and diseases cannot be disclosed.”

Investors in AI pharmaceuticals, on the other hand, are enthusiastically showcasing their passion for AI-driven drug development.NVIDIA, one of the world's most valuable companies, is a prime example.

In March, during NVIDIA's 2024 GPU Technology Conference (NVIDIA GTC 2024), there were 90 sessions related to healthcare and life sciences, ranking first in the specific industry distribution.

After Nvidia's investment arm, NVentures, invested in at least nine AI pharmaceutical companies in 2023, it continued to increase its investments in 2024. In the fifth-largest AI pharmaceutical financing in the first half of 2024, NVentures was present. Evolutionary Scale, which focuses on AI and protein structure prediction, raised $142 million in a seed round, with NVentures participating. In July, CytoReason, an Israeli company that uses artificial intelligence to develop disease research models, announced an $80 million financing round, with Nvidia being one of its investors.

NVIDIA Vice President and General Manager of Healthcare Kimberly Powell stated, while announcing the relevant financing information, "Over the past year, CytoReason has strengthened its platform using NVIDIA's latest accelerated computing and AI platforms, achieving more than a 10x speedup in inference workloads. Our continued collaboration with CytoReason will help more life sciences companies benefit from CytoReason’s predictive clinical insights."

This is NVIDIA's "open strategy": on the one hand, investing in and supporting more AI projects; on the other hand, making money by selling its own products. Who can blame it, being the world’s most advanced AI chip provider?

No.2

At the Crossroads of Making Money and the "AI Revolution"

Global AI Pharma Recovery: Has the Warmth Reached Chinese Peers? Some say yes, others feel that the harsh environment makes survival difficult.

In China, Two Different Development Paths of AI Pharmaceutical Industry Are Experiencing Different Levels of Investment Enthusiasm.

Some companies with cash flow that provide AI-assisted solutions to large pharmaceutical enterprises, such as METiS TechBio, successfully listed on the Hong Kong Stock Exchange in June, and as of the time of writing, their stock price had slightly increased.

Others who want to use AI to make innovative drugs are relatively more difficult.Like Insilico Medicine, which was once the first AI pharmaceutical company in China to submit an IPO application but missed being “the first” due to the expiration of the application documents, and resubmitted its IPO materials in March 2024.

In terms of revenue alone, Insilico Medicine is no slouch. In 2023, it reached $51.18 million, approximately 370 million yuan, surpassing XtalPi's 170 million yuan. However, its expenditures are also high, with annual spending at $210 million, nearly twice that of XtalPi.

Whether they have the ability to continuously generate value is the reason why Chinese investors show different attitudes towards the two development paths of AI pharmaceuticals.

Insilico Medicine's main focus is on research and development, which means "spending a lot of money." The latest prospectus describes it as an "AI-driven biotech company" with 15 drug candidate pipelines. It dedicates a significant portion to introducing a candidate drug for treating pulmonary fibrosis, calling it "our core product." Clinical trials are the most expensive part of the pharmaceutical industry.

METiS TechBio's main businesses are drug discovery and intelligent automation solutions. From 2021 to 2023, these two areas grew by 49.3% and 92.3%, respectively. A significant part of its success is due to the support of major clients. According to a research report by Zhongtai International, Pfizer, Johnson & Johnson, and Merck use the company’s services, indicating recognition of its service quality. In 2022, among the top 20 global biotech companies with the highest revenue, 16 were its clients.

"In the final analysis, it still comes down to whether the business model is viable, which is the same question that all AI + healthcare must answer."The aforementioned secondary market pharmaceutical investment professionals introduced that China's medical investment is now more rational and pragmatic.

And multinational pharmaceutical companies are willing to pay, not only betting on the future but also using AI to improve R&D efficiency and save costs.

Boston Consulting Group predicts that, based on the success rates of AI in Phase I and Phase II clinical trials, the overall efficiency of pharmaceutical R&D will double.

METiS TechBio Co-Founder and CEO Daicai Lai introduced that a top global pharmaceutical company had a small molecule drug in the drug development stage, which had been underway for about two and a half years without yielding satisfactory experimental results. METiS TechBio took over the project and "reached the drug development stage in six weeks." Securing $100 million in financing in 2024 was undoubtedly influenced by the willingness of large pharmaceutical companies to invest.

Without stable orders from large pharmaceutical companies, the only way to prove our strength is to successfully develop truly innovative drugs.

The aforementioned secondary market pharmaceutical investment professionals have observed that it is hard to say that some AI pharmaceutical products currently announced to have entered the clinical stage are superior to similar drugs from before. "Therefore, there are still questions about whether the optimization capability of AI models can match the experience of human professionals in the industry."

A research and development staff member working in an AI pharmaceutical company admitted that his company had already laid off employees, "mainly because no drugs have been successfully developed, and without money, it can't sustain itself."

Fudan University Uses AI to Screen Novel AD Biomarkers with High Clinical Application Value, Which is Particularly Precious at This Moment. "In fact, there are already AI pharmaceutical companies in China beginning to target the world's first-in-class or best-in-class tracks for layout, and have started to make breakthroughs in this direction, which also remains to be further observed," Huang Xiao introduced.

Similarly planning to deeply cultivate AI pharmaceuticals, CircleOne Intelligence is "grasping with both hands," providing services to multinational pharmaceutical companies on one side, while opportunistically advancing its self-developed pipeline on the other."Our technology is different from mainstream technologies like Google's Alphafold. It does not require the computation of 3D structures and can perform calculations for membrane proteins and conformationally unstable targets, enabling computation across the entire proteome." Pan Lurong still hopes that her company’s technology will bring about a "revolutionary" breakthrough, which requires hundreds of millions of yuan in funding. She is currently in contact with investment institutions both domestically and internationally to secure opportunities.

Indeed,Investors in China's primary and secondary markets have also started to pay attention to AI pharmaceuticals again.

In the first half of 2024, China saw 22 AI pharmaceutical financing events, with five occurring in June alone. Companies such as BioMap and Relay Therapeutics received strategic investments. Among the funded enterprises, some are targeting "AI-driven innovative drugs." For instance, LaiMang Bio secured an angel++ round of 50 million yuan in financing. The raised funds will be used for the IND application of a "very low dose" metabolically enhanced CD19 CAR-T cell therapy drug and to accelerate the clinical development of metabolically enhanced cell therapy drugs for solid tumors.

For AI pharmaceutical companies that may receive investment, Huang Xiao's view is that application scenarios are more important than data, and data is more important than algorithms. In terms of data, in addition to public data, it is necessary to have a rich, large, and structured database through cooperation or independent accumulation.In terms of algorithms, it is his expectation that they should not only align with the inherent characteristics of biomacromolecules but also meet the pharmaceutical industry's requirements for the physicochemical properties, human absorption, and biodistribution of medicinal molecules.

In the most critical scenario, Huang Xiao believes that the key lies in whether molecules with high market attractiveness can be found in specific therapeutic areas, mechanisms of action, and targets themselves. At the same time, with the assistance of AI, there is "potential to achieve better improvement compared to traditional drug discovery."

Insilico Medicine is fully advancing the development of key products, with six drugs having obtained clinical trial approval. The aforementioned core product has simultaneously entered Phase II clinical trials in both China and the United States.

In the eyes of those involved in pharmaceutical investments in the secondary market, the most noteworthy progress of Insilico Medicine includes the out-licensing of two products under development.In September and December 2023, the company licensed a small molecule inhibitor to Nasdaq-listed companies Exelixis and Stemline, respectively, receiving upfront payments of $80 million and $12 million, as well as potential milestone payments of up to $775 million and $394 million, respectively.

Just like other Biotech companies successfully licensing their R&D products to multinational pharmaceutical enterprises, the fact that someone is willing to buy the R&D products of AI pharmaceutical companies serves as an important basis for the investment community to assess their value. It adds a bit more persuasiveness to the idea that AI can create "truly innovative drugs."

The long-term collaboration between XtalPi and Pfizer has also accelerated the enhancement of its own technology. XtalPi stated that a series of collaborations began in 2016, and in 2021, they successfully confirmed the crystal structure within six weeks, expediting the development of the anti-coronavirus drug Paxlovid.

However,Even METiS TechBio, with a market value of nearly 20 billion yuan, has yet to achieve profitability. Investors' expectations for it go far beyond an annual revenue of over 100 million yuan and will not be limited to double-digit growth rates."Otherwise, it's quite good to build a small and beautiful company with an annual revenue of tens of millions of dollars," said a person related to pharmaceutical investments in the secondary market.

After all,What supports the entire AI pharmaceutical industry's financing recovery is technological progress and successful commercial applications.