On the most recent episode of Biotech Hangout (#132), Brad Loncar, Eric Schmidt, Tess Cameron, Luba Greenwood, and Tim Opler, along with special guest and veteran biotech columnist Adam Feuerstein, kicked off with a discussion on ‘zombie’ biotech companies. 

The topic was spurred by Adam’s recent STAT News analysis, where he sounded the alarm about “zombie” biotech companies lurking in the industry. These are firms that remain technically alive – still listed or operating – but have little meaningful scientific progress to show. They subsist on dwindling cash, making no real headway while awaiting some miracle or buyout. Feuerstein argues this glut of undead biotechs is dragging down the sector’s vitality.

Adam’s article Why biotech’s future is threatened by zombies, was a clarion call. The next day, Adam’s colleague Jason Mast broke the story - Bluebird Bio sells itself to Carlyle, SK Capital for less than $30 million - all but making every one of Adam’s points.  

Recently, Adam’s latest biotech analysis, The numbers that tell a scary story about the state of biotech, paints a grim picture of the industry’s current downturn. The biotech sector, particularly small- and mid-cap companies, continues to struggle, with relentless sell-offs forcing some healthcare and biotech funds to shut down or reduce their exposure. This has led to widespread declines in stock prices and exacerbated market instability. 

His zombie warning resonates because it captures the post-ZIRP reality of biotech investing. During the late 2010s and 2020, marked by readily available capital, investors financed speculative drug startups with zeal. However, with the current rise in interest rates and scarcity of capital, numerous ventures driven by hype now face a precarious situation. Investors are disillusioned, talented individuals are left in uncertainty, and the future of the entire biotech ecosystem appears grim under the shadow of these numerous "living dead" companies.

Zombies in AI-Driven Drug Discovery

The same zombie phenomenon exists within AI-driven drug discovery. Over the past decade, dozens of AI-first biotech startups raised enormous capital on promises to revolutionize pharma R&D. Today, several of the highest-profile ones look eerily zombie-like – richly funded and still operational, but failing to bring any AI-designed drug to meaningful clinical progress. These “zombie AI biotechs” haven’t died yet, but their pipelines and valuations have deteriorated badly, and optimism has given way to restructuring and rescue plans.

Throughout AI drug discovery, we see classic signs of “zombie” status: sky-high valuations followed by steep declines, pipelines that haven’t yielded any approved drugs (or even solid clinical data), major layoffs and strategy overhauls to conserve cash, and founders exiting under pressure. These companies were among the earliest and most lavishly funded in AI-driven drug discovery – yet years later, they’re largely defined by lack of results. They aren’t outright dead, thanks to their cash war chests and reputations, but by Feuerstein’s standard they are not truly alive either in terms of R&D productivity. For the biotech industry, they serve as cautionary tales that even the trendiest tech cannot escape the grind of drug development. And for investors, they raise red flags: much of that initial AI hype has turned into restructurings and write-downs.

The Newest Fear is Vampires: AI Biotechs Thriving on Funding but Lacking Results

If zombie companies are those slowly wasting away, “vampire” AI biotechs are another creature of concern – seemingly full of life, continuously drinking in new funds and partnerships, yet still unable to show the lifeblood of actual drug success. These are the AI-driven startups that have avoided the immediate death spiral, often by securing one lucrative deal after another, but they, too, have yet to deliver tangible breakthroughs in the clinic. They keep attracting capital (the blood of investors and pharma partners) even as the substantive output (new drugs) remains elusive.

Why do these “vampire” companies keep attracting investment despite lacking clinical results? 

Several factors are at play. First, the promise of AI in drug discovery is so transformative that neither VCs nor pharma executives want to miss out – they’d rather risk some money than be left behind if one of these companies does hit the jackpot. Many are long-shot bets with potentially huge payoffs. Like vampires from the movies, many of these companies have charismatic leaders and strong narratives. They speak on stage at Davos or Milken, talking about interdisciplinary teams or cutting-edge technology; they may even captivate attention with intermediate milestones that suggest progress. These accomplishments, short of a drug, are aimed to make you believe the platform is working on some level – enough to persuade investors to continue funding the journey. Third, there’s a bandwagon effect. When big pharma players endorse an AI startup through collaborations, it validates the startup in the eyes of others. The logic is that if sophisticated pharma R&D teams are partnering (and opening their wallets), the technology must have merit – even if it hasn’t yielded a drug yet. Finally, these “vampires” often adjust their business models to sustain momentum. They might take on service-like collaborations (for cash today) rather than advancing only their internal drugs. They also hype metrics like how many compounds or targets their AI has generated, which, while not actual drug results, can be impressive as productivity signals. All of this keeps the funding flowing. However, it’s fair to ask how long this can continue. At some point, these companies need genuine success – or investors may eventually abandon them in search of real opportunities.

Driving the Stake Through the Heart: The AIDD Code

How do we differentiate the real AI-driven drug innovators from the merely undead (zombie or vampire)? Our attempt at separating substance from hype is the AIDD Code, a framework to evaluate whether an AI biotech is truly delivering. The AIDD Code sets out six key criteria that define success in AI-powered drug discovery. 

These six questions provide the industry with a way to identify real AI biotech companies from their zombie or vampire peers.   

The AIDD Code

1. New Biology & Target Discovery – Has the company identified novel biological pathways or targets, as opposed to recycling well-known targets?

2. New Chemical Entities (NCEs) – Is it creating genuinely new molecules not seen before, rather than minor variants of existing drugs?

3. High “Zero-Shot” Hit Rates – Do at least 30% of the AI’s predicted compounds show real activity at sub-20µM potency on the first try? In other words, is the AI good at picking active hits without extensive trial-and-error?

4. Molecular Novelty (Tanimoto < 0.5) – Are the AI-designed molecules structurally diverse? A Tanimoto similarity score below 0.5 indicates the compounds are not just slight tweaks of known drugs.

5. Diverse Pipeline – Is the platform yielding drug candidates across multiple disease areas, proving broad applicability instead of a one-disease focus?

6. Preclinical Proof-of-Concept – Has the company demonstrated that its AI-found compounds produce real efficacy in gold-standard animal models of disease? This is a strong indicator that a drug has merit heading into human trials.

These criteria collectively form a high bar – essentially demanding that an AI drug company both innovates (new targets, new chemistries) and validates (hits that work, in vivo proof). It’s a push for hard evidence over hype. 

As Adam at Stat News points out, the current cycle of zombies and vampires in AI-driven drug discovery is a microcosm of the broader biotech investment climate. It reveals how exuberant investment can sustain companies long after their science has stalled – but also how reality eventually catches up. The reckoning now underway in this sector will likely intensify: many of the AI biotech “undead” will either have to find an exit - asset sales, take-unders, or pivots to service businesses - or face extinction as cash runs out.  Yet, amid the undead, the disciplined firms – the ones that do have solid science and realistic goals – are likely to survive and even thrive in the long run. Those survivors will be the companies that embraced data over hype, set achievable milestones, and maybe even delivered early wins - like a first clinical candidate that shows promise. 

Going forward, investors will demand clear evidence of platform productivity – whether it’s novel IP - new targets/molecules- concrete experimental data, or at least transparently reported metrics that show progress. Venture capitalists and public market investors alike should insist on seeing how a company’s AI outcomes compare to traditional benchmarks. 

Pharma partners, too, will likely recalibrate their approach. Over the last few years, many big pharmas engaged in what one might call scattershot partnering – making multiple bets across various AI startups in the hopes that one pays off. Going forward, pharma business development teams will scrutinize those partners more rigorously. They might structure deals with smaller upfronts and more milestone-based payments. We may also see pharma shift from broad discovery collaborations to more focused, problem-specific projects where the AI partner has to prove itself on a tightly defined task.