Version 3

Anthropic launched Claude Science on June 30, 2026 as a purpose-built research workbench for scientists, entering public beta on macOS and Linux with 60+ pre-configured domain tools, NVIDIA BioNeMo GPU integration, and automated reproducibility tracking.[1][2] FirstWordPharma reported that Anthropic simultaneously unveiled its own drug discovery plans alongside Claude Science, directly corroborating a conflict-of-interest concern flagged by commentators: Anthropic is both selling tools to pharmaceutical companies and running competing drug programs.[7][8] Media coverage spans Reuters, TechCrunch, MIT Technology Review, CNBC, GenomeWeb, and pharma-specific outlets, with Google and OpenAI reported to be working on competing scientific AI products but without product announcements yet.[11][12]

The pharma trade press picking up Anthropic's drug discovery plans alongside Claude Science means the conflict-of-interest question has reached exactly the audience — pharma executives and R&D decision-makers — whose trust Anthropic needs to build a customer base.[7] Whether Anthropic can credibly position itself as a neutral tool vendor while running parallel drug programs is now a visible commercial question, not just a commentator concern.[8]

Anthropic launched Claude Science on June 30, 2026, entering public beta as a desktop application for macOS and Linux. The product is not a new model but an application layer: a single coordinating agent that can invoke more than 60 pre-configured skills and connectors covering genomics, single-cell analysis, proteomics, structural biology, and cheminformatics, allowing researchers to direct multi-step workflows through natural language without manually configuring software environments or stitching together separate tools.[1][2] Claude Science was part of an Anthropic life sciences and healthcare initiative in development since October 2025, and its launch coincided with the release of Anthropic's Sonnet 5 model — though social media commentators noted Claude Science as the more significant announcement of the day.[3][4] Anthropic positions it as a first-class product alongside Claude Code and Claude Cowork.[5]

Two features anchor Anthropic's scientific positioning. First, reproducibility: every output carries the exact code, execution environment, and full message history used to produce it, making results independently verifiable without relying on researcher recall.[1][3] Second, an integrated reviewer agent checks citations, traces numbers to source data, and verifies figures match underlying code, with self-correction during pipeline runs.[1] Anthropic's early beta evidence includes a UCSF epidemiologist completing germline analyses in one-tenth the usual time and the Allen Institute compressing multi-year literature reviews to weeks, with some users producing review drafts exceeding 100 pages.[1][3] The platform can submit compute jobs to HPC systems or Modal and scale from one GPU to hundreds while keeping datasets local.[3]

The NVIDIA partnership is integral to the launch. NVIDIA announced its BioNeMo Agent Toolkit at BIO 2026 a week before launch, packaging GPU-accelerated bioinformatics — Evo 2 for DNA sequence modeling, Boltz-2 for biomolecular interaction prediction, OpenFold3 for protein structure prediction, and RAPIDS-singlecell compressing a 1.3-million-cell workflow from 52 minutes to 25 seconds — as callable skills within Claude Science.[6][2] NVIDIA's stated adoption figure, 18 of the top 20 pharmaceutical companies using BioNeMo technologies, anchors the partnership in existing enterprise relationships.

A structural concern has now reached the pharma trade press: FirstWordPharma reported that Anthropic unveiled its own drug discovery plans alongside Claude Science, a dual role that commentators and pharma-industry outlets have flagged as a potential conflict of interest.[7][8] Pharmaceutical companies considering Claude Science as a tool vendor may be cautious about partnering with an entity running parallel internal drug programs. Financial analysts viewed the launch as net positive for genomics and life sciences instrument companies including 10x Genomics and Illumina, while concern circulated in the health AI startup community about competitive displacement.[9][10] TechFundingNews reported that Google and OpenAI are working to develop competing scientific AI products, though no product announcements from either have emerged.[11]