The AI Biotech Investment Landscape: It's Not About Who Makes the Drug, But Who Collects the Cash Flow

ProfitVision LAB | AI BioTech Series | Overview | 2026.04

1. AI Biotech Isn't an Industry — It's a Chain Where Profit Is Being Redistributed

The market is used to treating "AI Biotech" as a single "new industry," lumping every company with any connection to AI and biotech into the same asset class, evaluating them with the same logic, and then being confused about why their performance diverges so wildly. That classification itself is the wrong starting point.

AI hasn't created an entirely new industry — it's doing something deeper: redistributing the value of an existing one. The nature of drug development hasn't changed — it's still a high-risk, long-cycle, low-success-rate process, costing over $2 billion on average and taking 10 to 15 years to turn a molecule into an approved drug. But AI is changing the efficiency and decision-making at every step, dramatically raising the value of certain nodes in the chain while loosening the barriers at others.

This change doesn't benefit every company equally — it concentrates value at specific nodes. So the core question in investing in AI Biotech was never "whose AI technology is strongest" — it's: who occupies the most advantageous toll position on this newly redistributed value chain?

The four companies in this research series — Illumina, Tempus, Schrödinger, and Veeva — each sit at a different node on this chain. Their business models, revenue sources, and risk structures barely overlap. Lumping them into the same investment category will only distort your judgment.

2. Four Nodes, Four Ways to Collect a Toll

The most effective way to understand this value chain isn't to look at the technology — it's to look at where each company's cash flow comes from, why customers are willing to pay it, and how hard that payment relationship is to interrupt.

Illumina sits at the very top of the chain — it produces the raw material for every downstream application. Gene sequencing itself won't disappear because of AI; AI will only increase demand for genomic data. Illumina earns money by "making the data exist in the first place," locking in customers for seven to ten years through a closed instrument-plus-reagent ecosystem.

Tempus sits at the data layer. It doesn't produce raw data — it integrates clinical data from multiple sources (genomics, medical imaging, electronic health records) into structured assets that AI can train on, then licenses that to drugmakers. It earns money by "making the data useful," and its moat is the breadth of its data and a first-mover scale advantage.

Schrödinger sits at the molecular-design layer. It doesn't license data — it uses its own physics-simulation platform to help drugmakers design candidate molecules, then participates in the future upside through royalties. It earns potential income by "getting good molecules found," and its moat is thirty years of accumulated physics-computation technology, nearly impossible to replicate in the short term.

Veeva spans the entire process from clinical trials to commercialization. It isn't involved in drug discovery — it ensures drugs can be developed, submitted, managed, and sold in compliance with regulations. It earns an institutional toll for "letting a drug legally exist in the market," and its moat is industry-standard status and regulatory-grade system lock-in — once a customer adopts it, switching is nearly impossible.

3. AI's Role Across These Four Companies: From Indirect to Direct

Another trap in treating "AI Biotech" as one category of company is assuming AI's impact on each is roughly equal. In reality, AI's role across these four companies spans a huge spectrum — from an almost purely indirect beneficiary effect, to being directly the core business model.

This spectrum reveals an important investment logic: reliance on the AI narrative is inversely correlated with revenue certainty. Tempus is the biggest direct beneficiary of AI, but it's also the company whose revenue most depends on the AI narrative holding up; Illumina has the least direct dependence on AI, but its business is the least affected if AI enthusiasm fades. This isn't a question of which is "better" — it's a question of risk appetite.

4. Four Different Payoff Curves: Why You Can't Use One Ruler for All Four

In financial analysis, we're used to evaluating every company with tools like P/E, EV/EBITDA, and DCF. But these tools carry a hidden assumption: that a company's revenue is continuous, predictable, and grows linearly. When that assumption breaks down, these tools fail systematically — and among these four companies, at least two have a revenue structure that simply doesn't fit that assumption.

These four curves represent four fundamentally different investment theses. The mistake would be evaluating Schrödinger with the framework you'd use for Veeva, or evaluating Illumina's consumables-compounding story with the growth logic you'd apply to Tempus. Each curve needs to be paired with a specific analytical tool and holding mindset.

5. The Most Critical Question: Where Does the Revenue Come From, and Is It Sustainable?

The biggest investing trap in AI Biotech isn't a wrong technology call — it's mistaking "narrative" for "revenue." In this space, nearly every company has a story that sounds entirely reasonable: AI is revolutionizing drug discovery, data is the new oil, computation can replace expensive experiments, compliance technology is indispensable infrastructure. These stories are all true — but a true story isn't the same as true revenue.

The investing question is always: has this story already converted into sustainable cash flow? And is the speed and scale of that conversion fully reflected in the current valuation?

Applying this question to the four companies produces very different answers:

Illumina's answer is "yes" — consumables revenue is already a stable cash-flow machine that has run for two decades, the NovaSeq X upgrade cycle is a visible growth driver, GRAIL-related drag is fading, and the overall business is getting back on track. This "already converted" certainty is the core of Illumina's investment case.

Veeva's answer is also "yes" — a gross margin above 75%, a non-GAAP operating margin of 44.9%, zero long-term debt, and 90% subscription revenue all describe a company with a highly mature business model. The Adenza integration creates short-term financial noise, but the core cash-flow machine has never stopped running.

Tempus's answer is "still converting" — Data & Services growth of 37% YoY and the $200 million AstraZeneca contract are real signals, but 75% of revenue still comes from the margin-constrained testing business, and it remains GAAP-unprofitable. Investing in Tempus is fundamentally a bet that "the conversion will complete in 2027-2028."

Schrödinger's answer is "the longest conversion timeline" — the software business is steady but limited in scale, and royalties contributing at scale depends on the clinical success of partnered drugs, a process that takes at least five to ten years. What you're buying today is a portfolio of options with almost no royalty income now, but potentially explosive royalty upside in the future.

6. A Core Framework: Three Questions to Quickly Place Any AI Biotech Company

After researching these four companies, we can distill a general analytical framework to help quickly place any company that claims to be "AI Biotech."

7. The Allocation Logic: Not Picking a Stock, But Building a Structure

Once you understand these four investment theses, the next step isn't "which one to pick" — it's thinking about "how to combine these four different risk-return profiles into a meaningful structure."

The core logic of this allocation structure is: use high-certainty assets to provide a stable base, and use high-uncertainty assets to provide nonlinear upside. VEEV and ILMN together account for 60–80%, providing the portfolio's stability; TEM provides medium-term structural growth potential; SDGR provides potential nonlinear upside in a small position — and its upside is uncorrelated with the other three, which is exactly its value in the portfolio.

8. Conclusion: AI Changes Efficiency; Position Determines the Return

To close this series, we need to return to the most fundamental observation: AI is genuinely changing the biotech industry, but what it's changing is "how things get done," not the underlying logic of "who gets to make money." Technology can change, efficiency can improve, but the business model's toll position — who has the right to collect on this chain, on what basis, and how hard it is for a customer to stop paying — is what actually determines long-term returns.

Illumina sells the ability to make data exist in the first place; as long as genomics keeps advancing, it can keep collecting. Veeva sells the process infrastructure that lets a drug legally exist; as long as the FDA exists, its system can't be swapped out. Tempus sells the ability to make data useful; if the data flywheel successfully spins up, its pricing power will keep improving. Schrödinger sells the intelligence that gets good molecules found; if those molecules eventually succeed, its royalty income requires no further work at all.

Four ways to collect a toll, four levels of certainty, four holding rationales. Understanding this framework is what lets you make judgments based on business reality, rather than market sentiment, in a field as narrative-dense as AI Biotech.

The single most important line in this whole series:
Don't ask who's best at using AI.
Ask who holds the toll right on this chain that's hardest to remove —
that position is the one truly worth holding for the long run.

This article is for research and educational purposes only and does not constitute investment advice. The allocation percentages are a directional reference, not specific trading recommendations. Investing carries risk; please evaluate according to your own circumstances.