Anthropic Enters Drug Discovery
Anthropic launches drug development programs for neglected diseases and a new AI workbench called Claude Science, as tech giants race to transform pharmaceutical research.
AnthropicAI in Medicine
Anthropic5 mins read
Anthropic Launches Drug Discovery Programs for Neglected Diseases and Unveils Claude Science
Anthropic is entering drug development for neglected diseases that traditional pharma considers unprofitable, while also launching Claude Science — an AI research workbench connecting scientists to 60+ databases. Novartis CEO Vas Narasimhan says AI could cut drug development timelines from 12 years to 7–8 and double success rates from 8% to 16%.
The Decoder
Anthropic launches its own drug discovery programs to tackle diseases Big Pharma considers unprofitable
STAT
AI company Anthropic announces it will begin developing drugs of its own
TechCrunch
Anthropic's Claude Science bets on workflow, not a new model, to win over scientists
MedCity News
Anthropic Launches Claude Science to Court Pharma Ahead of IPO
Anthropic Steps Into Drug Development for Neglected Diseases
Anthropic announced it will launch its own preclinical drug discovery programs focused on diseases that the traditional pharmaceutical and biotech industry considers unprofitable. The company says this aligns with its nonprofit mission and will give it firsthand experience using its own tools — including its new Claude Science workbench — in real scientific workflows. The announcement was made at Anthropic's AI for Science event, where executives emphasized that going hands-on with life sciences research is essential to building better AI models for the broader industry.
Anthropic's head of life sciences Eric Kauderer-Abrams noted the company had been asking itself what it should be doing beyond training models and building products — and drug development for underserved disease areas was one answer. It's a significant strategic move, signaling that Anthropic sees healthcare not just as a customer vertical, but as a domain where it can directly create impact.
Claude Science: A Research Workbench, Not a New Model
Claude Science is Anthropic's dedicated AI workbench for computational scientific research, built on top of existing Claude models (including Claude Opus 4.8) with no special biological fine-tuning. Rather than launching a specialized model, Anthropic's bet is on workflow: Claude Science connects to over 60 scientific databases and includes prebuilt toolkits for genomics, protein structure, and chemistry. A main AI assistant acts as a project manager, delegating tasks to sub-assistants or user-built expert agents, while a separate fact-checker AI validates citations and calculations before publication.
Early real-world results cited by Anthropic include a UCSF researcher using Claude Science to detect a viral contamination in minutes that his team had missed for a year, and the tool analyzing 100 rare genetic diseases in under an hour, flagging 32 for computational screening. Claude Science is available in beta to Pro, Max, Team, and Enterprise subscribers, and Anthropic is funding up to 50 academic projects with up to $30,000 in credits each.
The Numbers Behind AI's Promise for Pharma
Novartis CEO Vas Narasimhan — who also sits on Anthropic's board — broke down why even modest AI gains could be transformative for the industry. Drug development currently takes about 12 years. He categorized delays into information latency, operational latency, and biological latency. AI tools like Claude Science can compress the first two categories, which account for roughly 40% of development time, potentially bringing total timelines down to seven or eight years.
Narasimhan also projected that better safety predictions and molecular optimization could double the drug success rate from 8% to 16%. He cautioned, however, that identifying the right biological target for a disease remains AI's biggest unsolved challenge. Bristol Myers Squibb CEO Chris Boerner added that BMS already runs all its small molecule candidates through AI screening before they reach a wet lab, and has set an internal goal of cutting development cycle times by 30% using AI — a target it is reportedly on track to surpass.
How Anthropic, OpenAI, and Google DeepMind Are Competing in Science AI
Anthropic's Claude Science launch sharpens a three-way competition for the scientific research market. OpenAI released GPT-Rosalind in April 2026 — a specialized, biology-focused model gated behind an enterprise qualification process, with early access limited to partners like Amgen, Moderna, and Novo Nordisk. Google DeepMind takes a third approach: it owns foundational science models like AlphaFold and AlphaGenome that neither competitor can replicate, and bundles them with 30+ life science databases in its Gemini for Science platform. Notably, AlphaFold co-developer John Jumper recently left DeepMind to join Anthropic.
Anthropic is going broad — wide subscription access, open beta, and academic grants — while OpenAI goes narrow and enterprise-gated, and Google leans on proprietary models. Independent experts, including Oxford's Catherine Pope, still urge caution about deploying AI in clinical settings, warning that benchmark results can be disconnected from the messy realities of everyday healthcare. The race to own the AI operating layer for life sciences will be an early test case for how AI companies compete across all specialized professional verticals.
Discover More