This article is written based on publicly available information and is intended solely for informational exchange, not as any investment advice.

A company just 18 months old has secured $400 million, named Chai Discovery.

The lead investor is Index Ventures. Among the co-investors are Sequoia, Kleiner Perkins, Thrive Capital, and OpenAI. Yosemite, founded by Steve Jobs' son Reed Jobs, also participated in this round. The post-money valuation stands at $3.8 billion.

The word 'Chai' means 'life' in Hebrew. They specialize in AI-powered drug discovery.

As this funding round took place, another AI-powered drug discovery company, Isomorphic Labs, was also aggressively raising capital. Born out of Google DeepMind, it holds Nobel Prize-level protein structure prediction technology. Since last March, Isomorphic has raised $2.7 billion. Together, the two companies have raised $3.1 billion. Money is pouring into this track (chǎng jìng, meaning 'field' or 'sector') at an unprecedented pace.

New drug development is a long and expensive gamble. From the laboratory to the pharmacy, it takes an average of over a decade and costs $1 billion, with fewer than one in ten candidate molecules making it to the end.

If any of these three sets of numbers were halved, the economic value released would eclipse all current consumer-grade AI applications.

Index partner Nina Achadjian said in an interview with The New York Times, 'I believe life sciences will be one of the most important and impactful application areas for AI.' In simpler terms: chatting and coding earn attention; making drugs earns lives.

Chai's core product is Chai-1, a foundational biological model. It operates in the same direction as AlphaFold: using deep learning to compress protein structure predictions that once took months or even years in the lab into days or hours. The difference is that AlphaFold focuses on the protein structure itself, while Chai-1 aims to go further: covering the entire drug discovery chain from target identification, molecule generation, to efficacy prediction.

But where there's money, there are high barriers.

The real bottleneck in AI-powered drug discovery isn't the models but the data. Pharmaceutical data isn't as readily available as internet data. Real clinical data, compound libraries, and pharmacokinetic parameters are locked away in the servers of large pharmaceutical companies. Giants like Novartis, Roche, and Pfizer are also investing heavily in building their own AI R&D platforms. They are both potential future clients and competitors for AI-powered drug discovery companies.

To train truly useful drug discovery models, Chai and others must obtain this data. Yet pharmaceutical companies aren't eager to hand over their core assets to Silicon Valley's young talents.

Index clearly sees this bottleneck. Achadjian hinted in the interview that future capital deployments will revolve around 'data assets.' In other words, these $400 million are just the entry ticket. More money will be spent on buying data, pipelines, and clinical resources. The burn rate for AI-powered drug discovery is likely to outpace that of large language models.

Chai's four founders come from meticulously diverse backgrounds. There are researchers from Meta AI, computational biologists from DeepMind, and industry professionals from pharmaceutical companies.

This combination precisely addresses the core challenge of AI-powered drug discovery: pure AI scientists are unfamiliar with the long development cycles and regulatory logic of drug development, while pure pharmaceutical experts struggle to grasp the technical boundaries of deep learning. Those who can speak both languages fluently are likely the ones who will last in this race.

$400 million is a ticket to board. Isomorphic has Google's computational power, while Chai has OpenAI's backing, with the two representing the two major AI camps. AI-powered drug discovery is replicating the duopoly seen in large language models. But drug development differs from software; software can iterate, but drugs must be right on the first try. There are no gray releases, no rollbacks, and no fixes in the next version on this track (chǎng jìng).

Those on board cannot yet see the shore. They just believe it's there.