Anthropic wants to develop its own drugs

Anthropic Announces Claude Science and Its Own Drug Development Ambitions

At the “The Briefing: AI for Science” event earlier this week, Anthropic unveiled Claude Science, a new “AI workbench for scientists” designed to consolidate fragmented tools and datasets into a single environment. The platform also generates figures and visuals, targeting researchers in biology, chemistry, and related fields. But the announcement carried a second, more surprising move: Anthropic said it will develop its own drugs. Head of life sciences Eric Kauderer-Abrams stated the company will focus on discovering treatments for “neglected” diseases.

For developers and builders who follow the AI tooling landscape, this dual announcement matters on two levels. First, Claude Science is a direct competitor to other scientific AI platforms from OpenAI, Amazon, Google, and others. Second, Anthropic’s decision to become a drug developer itself places the company in an unusual position: selling software to drugmakers while potentially competing with them. This raises strategic questions for any team building AI tools for regulated industries like life sciences.

What Claude Science Offers — and What It Doesn’t Yet Reveal

Anthropic framed Claude Science around AI’s potential to “dramatically accelerate the pace of scientific discovery and the development of healthcare interventions.” The company showcased a long list of biotech and pharma customers already using Claude. However, specific technical details about Claude Science remain sparse. Anthropic did not respond to The Verge’s requests for comment on what diseases it plans to target first, nor whether it will partner with other companies for lab work, animal testing, clinical trials, or manufacturing.

This lack of detail is not unusual for a frontier AI company entering a highly complex domain. Experts quoted in the source article emphasize that “AI drug discovery” is a broad term. Namshik Han, a professor at the University of Cambridge and cofounder of AI biotech startup CardiaTec, noted that AI is applied at “every single stage of drug discovery” — from finding new compounds and improving them to supporting research, data analysis, clinical trials, and even manufacturing. Matthew Todd, a professor of drug discovery at University College London, called AI a “catchall phrase” given its wide array of uses.

For developers evaluating Claude Science, the key takeaway is that Anthropic is positioning the platform as an integrated workbench. That means it likely handles data ingestion from various sources, provides model access for prediction and simulation, and generates publication-ready outputs. If Anthropic’s track record with coding tools is any guide — and it’s worth comparing to other offerings in the best AI coding agents of 2026 — Claude Science will emphasize ease of use, a unified interface, and strong model performance on domain-specific tasks.

The Unusual Dual Role: Tool Provider and Competitor

Anthropic’s announcement that it will develop its own drugs is one of the most direct public attempts by a major AI company to enter the drug development space. This puts Anthropic in a category alongside AI-first drug companies like Insilico, Google DeepMind spinout Isomorphic Labs, biotech startups, and Big Pharma companies building or buying their own AI tools. But unlike those players, Anthropic is also selling a platform to the very same organizations it might compete with for drug candidates.

This creates a tension that developers and product managers at other AI tooling companies should watch closely. If Anthropic discovers a promising drug candidate, what happens? The source notes that Kauderer-Abrams did not say what the company would do if it finds any promising candidates. Options could include licensing the candidate, spinning out a separate drug development company, or selling it to a pharma partner. The ambiguity could make some pharma customers wary of sharing proprietary data on the Claude Science platform.

However, the same tension exists in other parts of the AI ecosystem. Cloud providers sell AI services while also building their own models. The key differentiator is trust and data isolation. Developers building similar platforms should consider how to clearly separate their tooling business from any downstream product ambitions — through data privacy guarantees, open-source components, or legal firewalls.

Why This Matters for Developers Building AI Tools

Anthropic’s move signals a broader trend: frontier AI companies are moving beyond developer tools and into vertical-specific applications. For anyone building in the AI infrastructure or tooling space, here are practical implications:

• Scientific workflows are a greenfield opportunity. While general-purpose coding is well-served by tools like Cursor and Claude Code — see our detailed comparison of Cursor vs Claude Code — scientific discovery requires specialized data handling, domain-specific model fine-tuning, and integration with lab equipment and databases. Claude Science is an attempt to standardize that fragmented landscape.
• Integration will be the moat. The fragmented tools and datasets Anthropic mentions are a real pain point. Developers who can build modular, interoperable platforms that connect to existing scientific data management systems (ELNs, LIMS, public databases) will have an advantage over monolithic workbenches.
• Domain expertise matters more than ever. The experts quoted in the source emphasize that “AI drug discovery” is broad. A platform that tries to do everything well may end up doing nothing exceptionally. Developers should consider specializing in one stage — screening, synthesizability prediction, clinical trial optimization — rather than building a general-purpose scientific AI.
• Regulation and ethics are non-negotiable. Drug development is heavily regulated. Any AI tool that touches patient data, animal testing, or clinical trial design must comply with FDA (or equivalent) guidelines and ethical standards. Anthropic’s focus on “neglected” diseases may be partly a strategic choice to avoid direct competition with big pharma on blockbuster drugs, but it also carries ethical weight.

How Claude Science Fits the Wider AI Tooling Landscape

Anthropic already dominates with its coding tools and powerful general-purpose models. Claude Science extends that platform play into a high-value, high-complexity vertical. The company joins a race that includes OpenAI’s ChatGPT for science tools, Amazon’s AWS HealthLake and SageMaker for healthcare, and Google’s Vertex AI for life sciences. The difference is Anthropic’s willingness to become a drug developer itself.

For developers, this is both a threat and an opportunity. Threat: Anthropic may lock up the scientific AI market with a proprietary platform, similar to how Cursor and other coding agents have become indispensable for many engineers. Opportunity: The vertical is still nascent. Few standardized APIs or SDKs exist for building AI-powered drug discovery pipelines. Now is the time to contribute to open-source scientific AI libraries or build specialized tools that integrate with workbenches like Claude Science.

The broader lesson is that AI companies are moving from general-purpose assistants to domain-specific orchestrators. Anthropic’s announcement is the clearest signal yet that life sciences is a key frontier. Developers who understand the biology, chemistry, and regulatory constraints will be in high demand — not just to build tools, but to help define how those tools interface with real-world drug development.

At the same time, skepticism is warranted. The source notes that “the uncertainty surrounding Anthropic’s plans reflects a broader uncertainty around the AI drug boom itself.” AI has been applied to drug discovery for years, yet the number of AI-discovered drugs that have reached patients remains small. Claude Science may accelerate research, but the path from a promising candidate to an approved therapy involves years of lab work, animal studies, and clinical trials — stages where AI’s impact is still unproven.

For builders, the actionable next steps are: watch Claude Science’s API documentation and pricing when it becomes available, evaluate how it handles data privacy and export controls, and start experimenting with its capabilities if your work touches life sciences. And if you’re building a competing tool, consider how to differentiate — perhaps by focusing on a narrower, deeper capability that Anthropic’s broad workbench cannot easily replicate.

Source: The Verge. Details as reported; verify specifics at the source.