Enveda: Finding Hidden Medicines in Nature
Scottish Mortgage is Baillie Gifford's flagship investment trust (founded 1909, LSE ticker SMT), known for its maximalist growth style — long-term stakes in Tesla, Amazon and ASML plus bold allocations to private companies like SpaceX and ByteDance. It is the UK retail investor's flagship vehicle for global disruptive growth.
In plain words
This interview features Enveda, a biotech company that takes a different approach: instead of studying single genes, it taps into chemicals shaped by nature over billions of years and uses AI like ChatGPT to decode them, aiming to find new drugs faster. For everyday investors, it shows that drug development is costly and failure-prone, so the biggest risk here is money, not science—but if success rates improve, the upside could be huge. Worth reading because it explains both a promising new method and the honest risks, helping you see what biotech investing is really betting on.
This report is based on a Scottish Mortgage podcast interview with biotech company Enveda, which uses AI and molecular science to discover new drugs from natural plants. Enveda founder Viswa Colluru believes that sourcing drug candidates from chemical compounds that nature has screened over eons can
This Issue Overview
Viswa Colluru is the founder and CEO of Enveda. He comes from an Indian middle-class family; at 13, his mother died of chronic myeloid leukemia (given only about five years to live upon diagnosis), which led him to abandon computer science and turn to biology. He was an early employee at Recursion, also part of the Scottish Mortgage portfolio, and founded Enveda in 2019 with $55,000 of his own funds. The main theme of this issue: Colluru explains how large language model technology is cracking the "sequencing" problem of nature's chemistry, how a "chemistry-first" route replaces single-target reductionism, and argues for Enveda's capital efficiency and its path to a trillion-dollar market cap.
The most weighty judgment in the entire video comes from Colluru: "If you can find better chemistry in nature that has been selected for billions of years then you can find better medicines." — i.e., if you can find better chemical substances in nature that have been screened over billions of years, you can find better medicines. Tom Slater then offers a footnote on the investment scale: healthcare is the world's largest industry, and if the success rate of drug discovery can be improved, the opportunity in fact has no upper limit.
Chemistry's "Sequencing Moment" Arrives Thanks to LLMs
Colluru asserts that the drug discovery industry has only one problem—"what works in the lab doesn't work in the human body"—and the key to solving it is chemistry that nature has screened over eons. Aspirin comes from willow bark, artemisinin from plants, morphine from the opium poppy, and metformin is still in use today; roughly half of FDA-approved small-molecule drugs are derived from nature. This idea is "both intuitive and validated, yet unharnessed."
Why has it remained unharnessed? Colluru's answer: chemistry never had its own "sequencing moment." Today one can extract and sequence the DNA of any sample, but "humanity has never done the equivalent for an organism's chemical code." More than 99% of the world's chemical substances remain unknown; mass spectrometry can detect thousands of compounds in a human blood sample, yet the world's top laboratories can annotate only about 10%. By contrast, scientists have identified roughly 35% of the viral genes at the bottom of polar ice sheets—"we know more about the genes of those viruses than about our own blood chemistry." He even makes a more fundamental biological argument: a dead cell and a living cell possess exactly the same genes; what keeps a cell alive is the "metabolic dance"—on average, about 80 million chemical reactions per adult cell.
The technological breakthrough is the mass spectrometer combined with the Transformer architecture (the underlying technology powering Claude and ChatGPT). The mass spectrometer accelerates each molecule into neutral gas, shattering it to produce a fragment-mass "fingerprint"; traditional machine learning cannot predict molecular structure from these fingerprints, for the same reason computers failed at language translation before LLMs: chemical fingerprints, like language, derive meaning heavily from context. Colluru illustrates this with how the referent of "it" shifts with the word at the end of the sentence—"tired" refers to an animal, "crowded" to a street, "raining" to the weather. LLMs solved the language problem through context-based training; Enveda applies the same approach to translate between "the chemical grammar presented by mass spectrometry data" and "chemical structures usable for drug discovery," drawing on vast numbers of as-yet-ununderstood molecules as training data, no longer limited to the few thousand molecules humans have already resolved.
Colluru attributes industry failures to three categories: unpredictable toxicity, efficacy falling short of predictions, and drugs that never reach their target organs. He believes the genomics-sequencing achievements of the past 25 years have actually misled the industry—"mistaking the map for the terrain": linking individual genes to complex diseases and then hunting chemical molecules for a single target—this reductionist path cannot solve the problem. Enveda's "chemistry-first" approach: select diseases with unmet needs and clear commercial opportunity (such as the need for safe oral anti-inflammatory drugs in inflammatory diseases), build pathways, organoids, and mouse models—but without reducing to a single target—then probe the models with tens of thousands of molecules from thousands of plant species; once effective substances are found, decode the mechanism. Aspirin existed first and then taught us about inflammatory mechanisms; we did not first understand inflammation and then invent aspirin. His advice for applying AI to drug discovery: "not electrocuting the horse, but really inventing the car" (meaning: not shocking the symbolic horse to get a faster carriage, but inventing the automobile).
Cross-Border Division of Labor Brings 3-10x Output Efficiency
Colluru argues that Enveda's cross-border division of labor delivers a 3-4x time and cost advantage over the world's largest pharmaceutical companies, and produces 3-10x more drugs per dollar invested than the industry average. The Boulder, U.S. site handles work that can be miniaturized, automated, and predicted (mass spectrometry, discovery, prediction); Hyderabad, India handles wet-lab experiments and drug development that are unpredictable and cannot be automated. This division began out of financial "necessity" — no large venture incubation round, only self-funded capital — and has now become a relay advantage: "the sun never sets on Enveda" (meaning the sun never sets on Enveda).
| Dimension | Enveda | Industry Comparison |
|---|---|---|
| Time and cost | 3-4x advantage | Versus the most favorable outsourcing contract at the world's largest pharma |
| Drug output per dollar (pure hybrid model) | 3x | Industry average |
| Drug output per dollar (incorporating natural chemistry thesis) | 10x | Industry average |
| Proof of concept | ~$20 million / ~2 years | $1-2 billion required for eventual success |
Pipeline scale: approximately 12 potential therapies covering dermatology, inflammation, obesity, and more; entered clinical trials 4 years after seed financing; candidate drugs to increase to 3 by year-end; the company has grown from 3 to 330 people. Colluru's three-stage roadmap: ① Within the next six quarters, four programs will generate roughly 12 clinical catalysts involving multi-billion-dollar sales potential, and the company will graduate from "a venture-scale startup perpetually on the brink of death" to "a multi-billion-dollar market capitalization with significantly lower cost of capital"; ② Find partners for one or two assets to share the burden of late-stage development and commercialization while sharing profits; ③ Move toward global multi-disease expansion in about 10 years. The ultimate goal is "a trillion-dollar company with multiple blockbuster drugs," and "becoming a trillion-dollar company requires only four drugs."
The Biggest Risk Is Funding, Not Science
Tom Slater argues that the greatest irony in the biotech industry is that "the biggest risks have nothing to do with science"—i.e., the biggest risks have nothing to do with science—but with funding. If you only develop drugs and never sell them, you have only costs; the inherent failure rate of drug discovery means you must keep raising capital to survive the inevitable failures. Slater says Scottish Mortgage's role is that of a long-term supportive partner.
Managing scientific risk: rather than pinning all hopes on a single molecule, you build a portfolio and rank it according to the best science. Regarding the trillion-dollar goal, Slater points out that large pharmaceutical companies exhibit a "power-law distribution"—a few blockbuster drugs far exceed the average drug, confirming the analysis that "four drugs would suffice"; no drug company has ever reached trillion-dollar scale, and the crux is that "the success of one drug does not mean the next drug will also succeed." If Enveda's platform approach can genuinely increase the probability of success for each drug, combined with the willingness to chase blockbusters, "the combination of these two could bring truly successful outcomes."
One should note the narrative element: Colluru's account carries a founder's salesmanship—"the first pharma company to be loved globally," the trillion-dollar goal, driven by personal tragedy—this is the perspective of a position holder; but he also admits that in the early days of the startup he experienced "several months of severe anxiety and panic attacks," and TS also clearly states that "some things simply will not succeed, and we knew that from the start."
Companies Mentioned
| Company | Guest Stance | Key Data |
|---|---|---|
| Enveda | Bullish — Scottish Mortgage built a small position over a year ago (private company) | ~12 potential therapies; 4 programs; ~12 clinical catalysts over the next 6 quarters; $20 million / ~2 years to proof of concept; 3–10x drugs per dollar |
| Recursion | Not stated — also in Scottish Mortgage's portfolio; Colluru's former employer, credited with "creating the AI-first biotech space" | Colluru worked there for several years before 2019 (self-described "a little over two years"), spanning science, product, and commercial strategy |
| Sanofi | Neutral — strategic investor; TS views it as validation of the approach, while also raising the question of whether "acquisitions kill entrepreneurship" | Invested in Enveda in early 2025 |
| Microsoft | Not stated — balance-sheet investor | A "rare" biotech company to receive Microsoft investment |
Judgments Worth Remembering
1. Colluru: The drug discovery industry has only one problem — what works in the lab doesn't work in the human body; the solution is to return to the chemistry that nature has screened over eons, "finding the next aspirin in weeks, not decades or centuries."
2. Slater: Healthcare is the world's largest industry, with enormous unmet need; if the success rate of drug discovery can be improved, "the opportunity is effectively without limits" — but the precondition is solving the funding problem — "the biggest risk is funding."