Claude Science, Anthropic's risky bet ahead of its IPO

Introduction: when an AI lab wants to become a drug company

A launch that changes the very nature of Anthropic

On June 30, 2026, Anthropic unveiled Claude Science, a digital workspace built for researchers, preconfigured with more than 60 scientific databases spanning genomics, proteomics, structural biology, and computational chemistry (Reuters). The tool is not a new artificial intelligence model, but rather a platform that organizes scientific work around Anthropic's existing Claude models, including Claude Opus 4.8 (TechCrunch).

What truly sets this announcement apart is Anthropic's parallel decision to enter drug development itself, a move confirmed by Eric Kauderer-Abrams, the company's head of life sciences, who said the effort would first target neglected diseases, largely abandoned by the traditional pharmaceutical industry (The Verge).

Why this precise moment is no coincidence

This announcement comes as Anthropic confidentially filed for an initial public offering with the SEC on June 1, 2026, after closing a Series H funding round of 65 billion dollars that values the company at roughly 965 billion dollars (SmartAsset). An IPO has been floated as early as October 2026, which could potentially make Anthropic the first artificial intelligence company to cross the symbolic trillion-dollar valuation mark (IG).

Diversifying revenue sources ahead of an IPO under this much scrutiny is far from trivial: it sends a clear signal to investors that the company does not intend to depend solely on chatbot subscriptions or Claude Code, its coding tool that was already generating more than 2.5 billion dollars in annualized revenue as of February 2026 (IG).

What Claude Science actually offers researchers

An architecture designed to eliminate technical friction

The platform runs on a lead coordinating assistant that acts like a project manager, capable of spinning up specialized sub-assistants to divide scientific work, somewhat like a supervisor delegating precise tasks to experts (TechCrunch). A separate reviewer agent then checks citations and calculations before any publication, a quality-control layer rarely automated to this degree.

The tool can run code on a lab's own infrastructure instead of sending data to Anthropic's servers, a choice that directly addresses the privacy concerns of research institutions and partner pharmaceutical companies (TechCrunch).

Recognized databases built in natively

Among the databases listed are UniProt, PDB, Ensembl, Reactome, ClinVar, and ChEMBL, all queryable in natural language without the researcher having to write a custom data pipeline for each one (Pondero). The platform can also generate three-dimensional protein structures, genome browser tracks, and chemical representations directly within the work environment.

The tool is currently available in beta for Pro, Max, Team, and Enterprise subscribers on macOS and Linux, with no Windows support at launch, which for now limits access to a tech-savvy slice of the scientific community (Evermx).

The drug discovery program and its blind spots

Neglected diseases, but few concrete details

Anthropic has chosen to prioritize neglected diseases, the ones the traditional pharmaceutical industry often abandons for lack of sufficient profitability, a direction that could have a genuine humanitarian impact if it materializes (The Verge). Even so, the company has offered very few specifics about which diseases it will target first.

Asked about potential partnerships for lab trials, animal testing, clinical trials, or manufacturing, Anthropic did not respond to reporters' requests for clarification, leaving significant uncertainty hanging over the concrete feasibility of this program (The Verge).

Aggressive hiring in the life sciences

Over the past year, the company has actively recruited biologists and built its own wet labs, an unusual physical infrastructure for an artificial intelligence company accustomed to operating almost exclusively in the digital realm (The Verge). Researcher Namshik Han says Anthropic has successfully poached several candidates from major pharmaceutical companies and prestigious academic institutions.

This massive hiring push shows just how seriously the company is taking this diversification, even though the scientific outcome remains highly uncertain at this early stage of the project.

Pharmacology experts' justified skepticism

AI hasn't made experiments obsolete yet

Researcher Frank von Delft summed up the situation with welcome frankness, saying that artificial intelligence models have not yet managed to make experiments unnecessary, adding that if Anthropic truly wants to develop a drug, the company will have to invest heavily in traditional experimental trials (The Verge). This fundamental scientific reality does not vanish simply because a technology company takes an interest in it.

Researcher Matthew Todd, for his part, notes that the field remains very far from seeing an artificial-intelligence-designed drug win regulatory approval for human use, stressing that there is always a lengthy timeline required to experimentally demonstrate the safety of a treatment (The Verge).

Human oversight that remains indispensable

Todd also insists that the drug discovery process cannot run autonomously, requiring constant human supervision throughout development, a crucial point that technology companies' marketing communications sometimes tend to downplay (The Verge). To date, no artificial-intelligence-designed drug has yet completed the full path of clinical trials through to FDA approval.

Some candidates developed with the help of artificial intelligence have indeed entered clinical trials, but it remains difficult to precisely assess AI's real contribution to these partial successes, or whether these candidates truly outperform drugs developed through conventional methods (The Verge).

Fierce competition in AI applied to science

Anthropic joins an already crowded pack

Anthropic is not the first Western technology company to move into this space: OpenAI, Amazon, and Google already have their own tools and platforms dedicated to the life sciences (The Verge). The company also joins a broader race that includes pharmaceutical companies built natively around artificial intelligence, such as Insilico, as well as Isomorphic Labs, the Google DeepMind subsidiary specializing in drug discovery.

What sets Anthropic apart, however, is its peculiar position: the company now sells software to other pharmaceutical companies while potentially becoming their direct competitor in the discovery of new treatments, an unusual business situation that could create friction with its own customers (The Verge).

Prestigious clients already on board

The company has cited Novo Nordisk and the Allen Institute as case studies among its early clients, suggesting that leading pharmaceutical organizations are already collaborating with multiple artificial intelligence providers simultaneously (TechCrunch). This early adoption by recognized institutions lends the project a degree of initial credibility.

Anthropic is also funding up to 50 external research projects in artificial intelligence applied to science, with grants of up to 30,000 dollars each, with applications open until July 15, 2026 (TechCrunch).

The Western strategic stake behind this announcement

Why the West must dominate this scientific race

This announcement fits into a broader context of global technological competition in which the West must imperatively keep its lead in scientific research applied to artificial intelligence, facing strategic rivals such as China, which is investing massively in its own biotechnology capabilities. The ability to accelerate scientificdiscovery is not merely a commercial matter, it is a long-term question of technological sovereignty.

Companies like Anthropic, which apply rigorous safety and biosecurity evaluations to their models before any scientific deployment, offer a reassuring contrast to less scrupulous players who might neglect these essential safeguards in their own race to innovate (Reuters).

The risk of a divide between rich and poor labs

Claude Science's current business model, restricted to paying subscribers with no free tier, nonetheless risks widening a gap between well-funded Western labs and research institutions with more modest resources, particularly in developing countries where access to this kind of tool could transform public health research.

This question of accessibility deserves particular attention from Western policymakers, because a scientific innovation that benefits only the wealthiest ends up widening global inequalities rather than reducing them.

What the tight timing of these multiple announcements reveals

A communication cadence that raises questions

The pace of Anthropic's announcements in recent weeks, from the confidential IPO filing on June 1, to the launch of Claude Science on June 30, to the parallel revelations about its drug development program, traces out a communications strategy clearly orchestrated to maximize the attention of potential investors.

This synchronization is not inherently suspicious, but it calls for a cautious reading: each announcement must be judged on its actual scientific merits, not simply accepted because it comes from a company in the midst of preparing one of the most significant technology IPOs in recent history (VNCMac).

The question of long-term credibility

If Claude Science and the drug discovery program produce tangible scientificresults in the years to come, this current period will be seen in retrospect as the pivotal moment when Anthropic stopped being merely a chatbot company and became a full-fledged scientific player. If results are slow to arrive, the announcement risks being seen as a mere public-relations maneuver ahead of an IPO.

Time, not press releases, will settle this essential question for the company's long-term credibility.

Historical precedents of AI in pharmacology

Repeated promises, still modest results

The recent history of artificial intelligence applied to drug discovery is dotted with ambitious promises that often took longer than expected to materialize. Companies like Insilico Medicine have indeed managed to advance drug candidates into clinical trials, but none has yet crossed the full finish line to commercial regulatory approval.

This reality does not invalidate the technology's potential, but it is a reminder that between an impressive technical demonstration and a drug truly approved for patients lies a considerable gap in time and money that few technology companies fully account for when they make their initial announcements.

The lack of quality experimental data still holds the sector back

ResearchersTodd and Han both stress that the lack of high-quality public experimental data could considerably slow efforts to develop drugs through artificial intelligence, a structural obstacle that even Anthropic's considerable financial resources will not overcome overnight (The Verge).

This fundamental technical limitation deserves repeating every time a new technology company announces its intent to revolutionize pharmacology through artificial intelligence.

Precedents among rival technology giants

Google DeepMind and Isomorphic Labs lead the way

Google DeepMind's subsidiary, Isomorphic Labs, has been working for several years on applying artificial intelligence to drug discovery, drawing notably on advances from the AlphaFold system for protein structure prediction. This accumulated experience gives Google a methodological head start that Anthropic will have to catch up on quickly.

OpenAI has also multiplied partnerships with biotechnology companies, betting on a different approach that combines its own generative models with specialized partner labs rather than building internal wet labs as Anthropic is now doing.

A diversity of strategies that ultimately benefits research

This plurality of approaches among the major Western artificial intelligence labs is, paradoxically, good news for overall scientific progress, with each company exploring different angles that could complement rather than needlessly duplicate one another.

Competition among these technology giants also accelerates the pace of scientificpublications and methodological innovations, a side benefit rarely highlighted in media coverage of this commercial rivalry.

The role of external funding agreements

An attempt to rally the independent scientific community

The grant program launched by Anthropic, offering up to 30,000 dollars to 50 selected external projects, explicitly targets doctoral students and postdoctoral researchers exploring the frontiers of science across multiple disciplines, with a particular focus on biomedical research. Applications will be evaluated and results announced before the end of July 2026.

This kind of external funding, while modest compared to the company's overall resources, could have a significant multiplier effect by giving talented young researchers access to tools normally reserved for large, well-funded institutions.

A computing partnership that lowers technical barriers

Infrastructure partner Modal is providing additional computing credits worth up to 2,000 dollars for certain selected grant recipients, complementing Anthropic's offering to reduce technical barriers to entry for independent researchers.

These funded projects are expected to run between September and December 2026, with the first public results expected in early 2027, which will offer a first concrete glimpse of this platform's real value for independent scientific research.

The implications for biosecurity and governance

Internal evaluations that offer partial reassurance

Anthropic states that the models used by Claude Science underwent the company's standard safety and biosecurity evaluations before deployment in a sensitive scientific context, a particularly important precaution when dealing with tools capable of manipulating potentially dangerous genomic and chemical data.

This transparency about internal procedures sets Anthropic apart from other technology players less inclined to publish their evaluation methodologies, a positive point in a sector where the race for speed could easily win out over caution.

The ever-present risk of malicious dual use

A tool capable of accelerating research in chemistry and biology necessarily carries a dual-use risk, where the same capabilities that enable the discovery of treatments could theoretically be diverted for malicious purposes, an issue the Westernscientific community must take seriously starting now.

Anthropic has not publicly detailed the precise monitoring mechanisms put in place to prevent such misuse, leaving an important governance question unresolved for the months ahead.

The scientific community's mixed reception

Cautious enthusiasm among early users

Organizations that tested the platform during its beta phase reported significant efficiency gains, according to Anthropic's own claims, though these testimonials come primarily from sources selected by the company itself rather than from independent evaluations conducted by neutral third parties.

This methodological nuance does not invalidate the tool's potential benefits, but it calls for additional caution before accepting without reservation the performance figures put forward by the company in its own promotional communications.

The questions the academic community keeps asking

Several university researchers question the long-term sustainability of access to these tools, fearing that growing dependence on private commercial platforms could weaken the independence of public scientific research in the face of large technology companies' commercial interests.

This tension between private innovation and public scientific independence is not new, but it takes on a particular dimension when the company in question is simultaneously preparing to become a competing pharmaceutical player itself.

What this means for patients with neglected diseases

Measured hope for often-forgotten populations

Neglected diseases mainly affect populations in low- and middle-income countries, often abandoned by the traditional pharmaceutical industry due to a perceived lack of sufficient profit potential. If Anthropic's program delivers on its promises, it could partially fill this glaring therapeutic void.

It's worth remembering, however, that a stated intention guarantees no concrete outcome whatsoever, and that many similar initiatives announced by other technology companies in the past have not always delivered the promised treatments within the timelines originally suggested.

The importance of measuring progress over several years

Developing a drug, even with the most advanced artificial intelligence assistance, generally takes several years between the initial discovery of a promising candidate and its eventual regulatory approval, a timeline that far exceeds the current media cycle surrounding Anthropic's IPO.

Serious observers of the sector will therefore need to follow this story over several years rather than judging its success or failure in the months immediately following the initial announcement.

The broader context of the financialization of scientific research

When science becomes a stock market argument

This announcement fits into a broader trend in which major technology companies use their scientific capabilities as a sales pitch to investors, gradually turning fundamental research into a tool of financial communication rather than a purely academic end in itself.

This growing financialization of science is not necessarily negative if it generates additional resources for research, but it deserves close monitoring to prevent scientific priorities from being dictated solely by short-term stock valuation considerations.

A delicate balance to preserve for the future

The challenge for Anthropic, as for the entire Western technology industry, is to maintain a delicate balance between the legitimate commercial imperatives of a private company and the scientific and ethical responsibilities that come with developing tools as powerful as those offered by Claude Science.

This balance, if maintained with rigor, could serve as a positive model for the entire Western technology sector in the years ahead.

Conclusion: between scientific promise and financial maneuvering

A bet worth watching closely

The launch of Claude Science and the announcement of an internal drug development program represent a bold step for Anthropic, at a strategically crucial moment for the company ahead of its long-awaited IPO. The ambition to target neglected diseases deserves praise, but it must now be followed by concrete, verifiable results rather than mere statements of intent.

For now, the company has delivered an impressive technical tool from a software engineering standpoint, but it has not yet demonstrated that it can turn this infrastructure into genuine therapeutic discoveries that will actually benefit patients.

What the West should take away from this episode

This announcement confirms that the race for scientific artificial intelligence is now a major strategic stake for Western technology companies, no longer just a marginal academic research topic. The West has every interest in encouraging this competition, provided it maintains rigorous standards of safety and scientific transparency.

It remains to be seen whether Anthropic will manage to honor its humanitarian promises toward neglected diseases, or whether this initiative will remain, above all, a skillfully timed sales pitch ahead of one of the most anticipated IPOs of the current technology decade.

By Maxime Marquette, columnist

Columnist's transparency note

Who I am and my acknowledged biases

I am a columnist, not a biologist or a financial analyst, and I rely on verified technology and scientific reporting for this commentary. I believe the West must stay at the cutting edge of scientific and technological innovation against its strategic rivals, a conviction that shapes my generally favorable outlook on this kind of initiative, without stopping me from flagging real areas of uncertainty.

What I don't know and my method

I cannot personally assess the scientific validity of the results produced by Claude Science, nor predict whether Anthropic's drug development program will lead to treatments that are actually approved. My method consists of cross-referencing the company's official communications with critical analyses from independent pharmacology researchers, clearly flagging the limits of what I can state with certainty about such a recent project.

Sources

Primary sources

The Verge — Anthropic wants to develop its own drugs, July 3, 2026

Reuters — Anthropic unveils Claude Science for scientific research, June 30, 2026

TechCrunch — Claude Science bets on workflow rather than a new model, June 30, 2026

Secondary sources

Yahoo Finance — Anthropic launches Claude Science to diversify revenue ahead of its IPO

India Today — What to know about Anthropic's drug discovery program, July 1, 2026

SmartAsset — Guide to Anthropic's IPO, valuation, and timeline

Pondero — Anthropic launches Claude Science, a workspace connected to more than 60 scientific databases