AI Drug Discovery: Absci Rally Tests the AWS Squeeze Thesis

Hyperscaler moves into the lab

Amazon Web Services is pushing further into drug research. Its partnership with Novo Nordisk creates a London hub where the drugmaker's R&D staff work alongside AWS engineers, AI specialists, applied scientists, and professional services teams 1. The companies say the hub is meant to cut the handoffs between computational analysis and wet-lab work, and to shorten the path from identifying a drug target to a first human dose 1. Novo's Bøgebjerg framed the goal as pairing advanced technology with deep scientific expertise and a focus on patients 1. Amazon Bio Discovery, AWS's tool for supporting computational drug work, is part of the deal 1.

The launch comes as AI-designed medicines start commanding serious money. Helixon's AI-designed biologics reportedly won a $1.7 billion licensing deal with Sanofi in 2025, and the AI drug-discovery market is projected to reach $4 billion in 2026 5. Analysts describe AWS's approach as an "infrastructure platform" play. In that framing, AWS competes with "intelligence platforms" such as NVIDIA BioNeMo and Isomorphic Labs, and with "application platforms" like Benchling 5. The bet is less about building the smartest model and more about owning the whole workflow from lab to cloud 5.

The squeeze narrative

That framing has produced a clear bear case for independent AI biotechs. Labcritics argues that hyperscalers racing to lock in the lab-to-cloud pipeline puts more pressure on pure-play firms like Absci and Insilico Medicine. Those companies lack AWS's footprint, which reportedly spans 19 of the top 20 pharma companies 5. The logic is simple. If big pharma can get compute, models, and embedded engineers from one vendor it already uses, specialist platforms risk becoming optional.

Absci's market tells a different story

Absci's share price does not look like a company being squeezed out. The stock recently opened at $10.27, against a 52-week low of $2.24. Its 50-day moving average of $9.16 sits well above its 200-day average of $7.26, a sign of sustained upward momentum 2. The company carries a market capitalization of about $1.76 billion, a negative P/E of -12.84 reflecting ongoing losses, and a beta of 2.27, meaning it moves far more sharply than the broader market 2.

Wall Street has been raising its targets. Truist lifted its target from $11 to $17 with a buy rating, and HC Wainwright doubled its target from $8 to $16 2. BTIG set a $15 target, Needham reiterated a buy at $11, and Morgan Stanley kept a positive rating with a $10 target 2. The average analyst target is $13.71, which implies meaningful upside from current levels 2. The spread from $10 to $17 is wide, though. Analysts agree on the direction but not on how far the story goes.

Insilico shows competitors are also customers

Insilico Medicine complicates the idea of a clean rivalry between hyperscalers and pure-plays. AWS itself promotes Insilico as a showcase customer. Insilico moved its model training to Amazon SageMaker, cut the time to deploy model updates by 83%, and sped up model iteration more than 16-fold 4. The company has used generative AI for drug discovery since 2014, long before the approach went mainstream. Its Pharma.AI platform covers target identification, molecule design, and candidate selection 4.

Insilico's 2025 results suggest that independents can sell through the same cloud giants that supposedly threaten them. The company reported $56.24 million in revenue and says it serves 13 of the top 20 global pharma companies. Software revenue grew 23.8% 3. Its model is distributed on both AWS Marketplace and Microsoft's Discovery platform 3. Insilico also listed on the Hong Kong Stock Exchange in late December 2025, with backing from investors including Eli Lilly and Tencent 3.

The reading

The squeeze thesis is not wrong about direction. Hyperscalers clearly want to own the pipeline, and an embedded engineering hub like Novo's London setup is a stickier offering than renting GPUs 15. But the evidence points to a more layered market than a winner-take-all fight.

Infrastructure and drug assets are different businesses. AWS sells the tools for discovery. Companies like Absci and Insilico are ultimately valued on the molecules and programs they produce, plus their ability to license or partner them. Insilico's results show that a specialist can treat the hyperscalers as distribution channels rather than only as threats 34.

For Absci, the rally and the rising analyst targets suggest investors are pricing its own pipeline progress, not its position against Amazon 2. That conviction comes with real volatility, given the stock's high beta and continuing losses. The risk for pure-plays is not that AWS builds better molecules. It is that AWS becomes the default workspace where pharma decides which molecules are worth pursuing. Independents that can show differentiated clinical value should stay relevant. Those selling only software or models face the sharper squeeze.