Three MIT Postdocs' AI Drug Discovery Firm XtalPi Files for Hong Kong IPO with $14.1B Valuation

Three MIT Postdocs Build a "Fundraising King" in Global AI Pharmaceuticals: 6 Rounds of Financing in 6 Years, Total Financing Approximately 5 Billion Yuan, Post-Investment Valuation Approximately 14.1 Billion Yuan.

Behind it is a series of well-known institutions, needless to say: Tencent, Sequoia, SoftBank, CICC, Shunwei, Google, ZhenFund, FiveSource, PICC, China Life, and Artisan Partners, a well-known large long-term fund in the United States, among others. It is reported that during its Series D financing, nearly 50 institutions participated in the bidding, with intense competition for quotas.

This AI pharmaceutical star is XtalPi.

In the innovative drug R&D circle, there is a famous "three 10s" rule: developing a new drug requires at least an investment of 1 billion US dollars, a development cycle of 10 years, but even for drugs that have successfully entered clinical trials, their success rate is less than 10%. In recent years, this trend has become increasingly obvious, with pharmaceutical companies' innovation return rate continuously approaching 1%.

AI pharmaceutical companies represented by XtalPi are attempting to address the aforementioned issues: significantly shortening the cycle of each stage in drug development, reducing the cost for enterprises to develop new drugs, and also helping scientists identify novel drug molecules that have not yet been discovered.

In addition to being a veritable financing king, XtalPi is also the second technology enterprise to apply for listing under Rule 18C (the first being Black Sesame Technologies, a developer of autonomous driving computing chips).

The Rule 18C, introduced by the Hong Kong Stock Exchange in March this year, is mainly aimed at specialized technology companies, with the purpose of attracting more innovative and growth-potential technology companies to be listed in Hong Kong.

It has high requirements for the technological attributes of applicant companies, allows no income and no profit, and reduces the listing threshold for special technology companies to a certain extent, making it particularly suitable for companies like XtalPi.

Three MIT Postdocs Start a Business, Post-Investment Valuation at 14.1 Billion Yuan

The high entry barrier for biopharmaceutical startups is undeniable, as they are typically founded by experts in a specific field. XtalPi, for instance, was founded by three MIT postdoctoral physicists, boasting the most impressive academic credentials.

Chairman Wen Shuhao is a physics Ph.D. from the Chinese Academy of Sciences and later a postdoctoral fellow at the University of California and MIT. CEO Ma Jian is a physics Ph.D. from Zhejiang University, followed by postdoctoral research at MIT. Chief Innovation Officer Lai Libo completed dual bachelor’s degrees in physics and mathematics at Peking University, earned his Ph.D. in physics at the University of Chicago, and also became a postdoctoral fellow at MIT.

Three individuals with exceptional academic achievements met at MIT and, inspired by the school's motto "Mind and Hand" (emphasizing both theory and practice), decided to establish a company driven by fundamental scientific theories.

Choosing the pharmaceuticals industry is also because Boston, where the school is located, is the Silicon Valley for Pharmaceuticals, which naturally drew attention to this industry. "We hope to bring the most accurate physical algorithms and the most cutting-edge artificial intelligence algorithms, along with unprecedented computational power, into the pharmaceuticals industry, pushing this somewhat conservative industry into a new phase of digitalization and AI-driven upgrades."

In 2015, they founded XtalPi, targeting the prospects of AI in biopharmaceuticals.

Amid the entrepreneurial boom, it was not difficult for the three postdoctoral researchers to secure funding. In fact, their situation perfectly illustrated what it means to have the support of many and be pursued by capital. At the time of its establishment, the company caught the attention of Ma Huateng, the head of an internet giant. "When the project was submitted, no one had reviewed it because Tencent had never invested in projects like yours before. However, the work you are doing has social benefits," Ma Huateng once remarked. While it seemed that the investment was driven by social impact, in my view, what Ma Huateng really meant was that when investing in such "unfathomable" projects, the key factor is still the people behind them.

In December 2015, XtalPi completed its first round of financing, with Tencent and Renren contributing 24 million. Despite a rapid influx of investment institutions in subsequent rounds, Tencent's presence was evident in both the Series B and Series C financings. Indeed, Tencent is XtalPi’s largest institutional shareholder. The prospectus reveals that, prior to the IPO, Tencent held 13.66% of XtalPi through its subsidiary Image Frame Investment, making it the company’s largest external institutional shareholder.

The A+ round saw the entry of Frees Fund and ZhenFund, while the B+ round welcomed Sequoia, Google, China Life, Yaya, and SIG. The C round featured an even longer list of 17 well-known institutions. In addition to the aforementioned shareholders, new faces included FiveSource Capital, SoftBank, PICC, CICC, CMB International, CITIC, Haisong, Shunwei, Mirae Asset Global Investments, IMO Ventures, and Parkway Fund.

In the highly competitive $400 million Series D financing round, led by Five-Star Capital, OrbiMed, and Hopu Capital, with new shareholders Harmony Capital and China Biologic Products Group joining the investment. Early shareholders such as Tencent Investment, Sequoia China, and IMO Ventures continued to increase their investments. Other investors include institutions such as Neumann Capital and Artisan Partners. Notably, Artisan Partners is a well-known long-term U.S.-based fund.

XtalPi has raised a total of $7.32 billion (approximately RMB 50 billion) through six rounds of financing, with a post-money valuation of about RMB 14.1 billion. According to data from Frost & Sullivan, XtalPi ranks first in total financing among global AI-powered drug discovery companies. Following its Series D funding round, reports emerged that XtalPi had secretly initiated plans for an IPO in the United States.

Accumulated Operating Loss of Nearly 1.4 Billion Yuan in Three and a Half Years, XtalPi Assists Pfizer in Developing Oral COVID-19 Drug

The potential of AI in pharmaceuticals is widely recognized, but how can these cutting-edge technologies realize their commercial value?

Looking at the current AI pharmaceutical companies, there are mainly three business models: AI-biotech (using AI to establish their own new drug R&D pipelines to become pharmaceutical companies), AI-CRO (providing drug discovery services), and AI-SaaS (selling the use of AI drug research platforms and software).

XtalPi leans more towards an AI-CRO business model, positioning itself as a "technology platform providing integrated drug discovery solutions for global biopharmaceutical companies," following a platform technology development path. Therefore, in addition to the biopharmaceutical field, XtalPi also provides innovative technologies, services, and products for industries such as chemical engineering, new energy, and new materials.

To date, XtalPi mainly includes two major businesses: drug discovery solutions and intelligent automation solutions.

On one hand, XtalPi mainly provides drug discovery solutions to biotechnology and pharmaceutical companies; on the other hand, XtalPi offers intelligent automation solutions to customers, primarily including solid-state research and development services and automated chemical synthesis services. Its revenue is mainly obtained in the form of service fees.

The prospectus shows that XtalPi generated revenues of RMB36 million, RMB63 million, RMB133 million, and RMB80 million in 2020, 2021, 2022, and the six months ended June 30, 2023, respectively.

Due to the relatively short commercialization period, the losses are reasonable. During the reporting period, XtalPi's operating losses were RMB 126 million, RMB 299 million, RMB 525 million, and RMB 435 million, respectively, with a total accumulated operating loss of RMB 1.385 billion over three and a half years.

The company's R&D investment is relatively high, with R&D expenses during the reporting period amounting to RMB 83.5 million, RMB 212.6 million, RMB 359 million, and RMB 234.4 million, accounting for approximately 234.4%, 338.5%, 269.2%, and 293.1% of the revenue for the respective periods.

Currently, XtalPi has more than 700 scientists and technical experts, holds over 120 authorized patents, approximately 27 ongoing drug discovery projects, 4 research and development centers, and over 10,000 square meters of laboratory space.

XtalPi has established in-depth cooperation with nearly 400 enterprises and institutions, including Pfizer, Johnson & Johnson, and Eli Lilly. Among them, the collaboration with Pfizer has drawn particular attention. Following the outbreak of the global COVID-19 pandemic, XtalPi was commissioned by Pfizer to participate in the development of the oral COVID-19 drug Paxlovid. By combining predictive algorithms with experimental validation, XtalPi helped Pfizer confirm the drug's crystal structure in just six weeks, accelerating critical R&D decisions and subsequent development and market launch.

It is reported that the "ticket" to cooperate with Pfizer came from a crystal form prediction blind test held by Pfizer in 2016. The blind test involved three drugs, and Pfizer invited institutions and teams around the world capable of predicting crystal structures to conduct experimental evaluations separately, which were then compared with the stable crystal forms already synthesized in the laboratory. In the end, XtalPi achieved a prediction accuracy rate of 100%, surpassing many top companies and institutions in Europe and America.

Wen Shuhao also stated that AI drug discovery will become the infrastructure for drug research and development in the future. XtalPi's collaboration with several highly influential pharmaceutical companies worldwide demonstrates that China's AI drug discovery technology and enterprises have gained global competitiveness.

AI Drug Discovery: A Long and Snowy Slope

Amid the global AI boom ignited by ChatGPT, we have reason to believe that the next blockbuster drug could be invented by AI.

Currently, there are mainly the following three types of players in AI pharmaceuticals:

Firstly, leading pharmaceutical companies such as Pfizer, Johnson & Johnson, Novartis, and Bayer from abroad, and in China, companies like WuXi AppTec, Chia Tai Fenghai, Hansoh Pharma, and Yunnan Baiyao are also involved in AI-driven drug research and development.

Secondly, leading Internet companies in China, such as Tencent, are also building their own AI drug discovery platforms by leveraging AI models and platform advantages. For example, Tencent's "Yunshen Zhikang" is an AI preclinical drug research platform constructed based on its own algorithms and databases.

Thirdly, there is a large number of AI drug discovery companies. Data disclosed by Guanzhi Hainai Consulting shows that globally, there are as many as 392 companies solely engaged in the early drug development field of AI pharmaceuticals, while hundreds more operate in other areas of AI drug discovery.

2021 can be said to be the best year for AI pharmaceuticals. That year, hundreds of millions of financing cases for AI pharmaceutical companies were everywhere, and there was frequent good news of cooperation with pharmaceutical enterprises.

In 2020, after Schrödinger went public on the US stock market, its share price continued to soar, with gains reaching nearly sixfold at one point, igniting the AI pharmaceuticals sector. The boom quickly spread to China, where domestic AI pharmaceutical companies such as XtalPi, Insilico Medicine, Accutar Biotechnology, BioMap, and StoneWise gained significant attention in the primary market.

However, this boom began to decline sharply in 2022. In terms of financing, the amount in 2022 was only half of that in 2021. Many companies quietly shelved some of their pipelines. In 2023, with an even more "chilling" investment environment, this figure became worse.

On the one hand, there is a significant change in the investment environment, and on the other hand, the role that AI can play in pharmaceuticals is still relatively limited.

A few days ago, at the "2023 Guangdong-Hong Kong-Macao Greater Bay Area Hard Technology Industry Conference" held in Guangzhou, we had a roundtable discussion on AI + pharmaceuticals. Qian Zhang, Chief Operating Officer of Magpie Pharmaceuticals, described the current state of the industry: although AI can continuously improve R&D efficiency, it still cannot help us solve core technologies and relies more on experts' research and accumulation. At present, AI can only quickly learn based on existing human knowledge and develop new things through algorithms, but it is difficult to explore or lead in areas unknown to humans.

More specifically, AI drug discovery is generally constrained by data. Challenges such as significant differences in data collection, inconsistent quality, and the difficulty in obtaining data on failed outcomes limit the development of AI in pharmaceuticals.

However, as a veteran with more than 20 years of experience in AI, Deng Yafeng, founder and CEO of Carbon Silicon Intelligence, paints a more realistic picture: the current goal is not to use AI to replace experts, but rather, by making good use of AI, experts can perform better. "The future will definitely be based on a model that combines AI, automation, and experts, replacing the traditional approach that relies more heavily on experts and experimentation."

Although no drug designed by artificial intelligence has successfully reached the market yet, in the long term, this is definitely the trend of industry development. This is not a fantasy, and it even has practical significance.

As Ma Jian said, "Everyone should not only focus on the ultimate success. This emerging technology has already achieved phased successes and validations in more specific R&D stages, such as enabling faster hit compound discovery, accelerating lead compound optimization, and advancing more quickly to the preclinical candidate compound stage, all of which carry the significance of 'accumulating small victories into a major triumph'."

More specifically, although no drugs designed by AI have been approved for marketing yet, in the preclinical research stage, possibly over 30% to 40% of drug molecule pipelines involve the participation of AI technology. In time, as the iceberg gradually emerges, we will suddenly realize that many of the reserve pipelines have already become AI-driven drug development pipelines.