Three vocabularies
A shared conversation needs three separate translations.
Readiness
What has been proven, and in which system?
Revenue
What creates value before approval?
Probability
What could falsify the programme?
Readiness is not one number
A programme advances only when its evidence travels across three different systems.
Biological
Mechanism · model · replication
Manufacturing
Yield · purity · comparability
Regulatory
Pathway · endpoint · feedback
Replace the revenue curve
A value-inflection path makes the next cheque legible.
Capital
Evidence
Re-rating
Next capital
Two clocks, one financing decision
The fund clock measures return windows. The clinical clock measures proof.
Deployment concentrates in the first three or four years.
Approval can arrive after the fund needs to return capital.
Axis movement
The same “readiness” label can conceal radically different company states.
ImmunoACT
Biology strong
Manufacturing funded alongside R&D
Regulatory pathway built
Shan5
Biology approved
Manufacturing failure point
Regulatory requalification
Probability is a chain of claims
A credible estimate names the base rate, the adjusters, the range and the experiment that can change it.
Base rate
per drug–indication
Adjusters
target + population
Range
40–50%, not false precision
Killer experiment
what would falsify it
Separate the risks
A diligence question becomes useful when each failure mode has an owner and an observable signal.
Biology
mechanism
Manufacturing
batch
Regulatory
pathway
Supply chain
inputs
Economics
margin
Say this, not that
The translation is not simplification; it is evidence with its conditions attached.
Not priceable
TRL 4.
Priceable
TRL 4: efficacy demonstrated in a humanised mouse model, n=24, across two independent cohorts.
The investor’s translation burden
A founder’s evidence must travel through the whole investment committee.
Founder
raw evidence
Technical lead
mechanism
IC
risk + price
Capital partner
next buyer
It maintains separate scales for products and for the development tools like assays, models and reagents used to evaluate them. Biomanufacturing Readiness Levels were subsequently proposed to capture what conventional TRLs and Manufacturing Readiness Levels can miss in biopharmaceuticals: quality, regulatory and operational readiness alongside technical maturity. India now has an additional reason for founders to speak TRL. The ₹2,000 crore BIRAC-RDI fund (BIRAC stands for Biotechnology Industry Research Assistance Council, and RDI stands for Research, Development and Innovation), to be deployed over five years through equity, convertibles and long-term debt, explicitly covers TRL-4 through TRL-9 across areas including biopharma, bio-industrial manufacturing, bioenergy and biocomputation. If you want that capital, you learn the ladder. The problem is carrying that ladder into a room where it means something different.
A founder walks into a venture meeting and says, “We’re at TRL 5.” The investor writes it down. Neither means the same thing. And neither finds that out for another two quarters.
This is an ordinary failure mode in biotech fundraising. It is rarely a failure of intelligence on either side. It is a failure of shared units.
Software venture capital spent two decades converging on a handful of metrics that mean roughly the same thing to everyone in the room: ARR, net revenue retention, CAC payback, burn multiple. A founder says, “$2 million ARR, growing 15% month-on-month,” and an investor immediately has a mental model for stage, risk, round size and the next milestone. The sentence is short because the shared context is long.
Biotech has no equivalent compression. Instead, it operates across three separate vocabularies:
Each means something different to founders and investors. That is where deals quietly die.
The gap is particularly relevant in India. The bioeconomy reached roughly $195 billion in 2025, up from $165.7 billion the previous year, with close to 12,000 biotech startups in the ecosystem. Yet much of the early capital that helped these companies reach their first meaningful data packages came through grants and other non-dilutive or soft-funding mechanisms rather than conventional venture equity.
Founders therefore learn to communicate in the language of grant reviewers. Then they walk into a room full of investors who speak a different language.
The structural mismatch
The clock problem underneath everything
Before getting into the three vocabularies, there is a more fundamental mismatch that nobody says out loud. A venture fund typically has a ten-year life, with one or two extension years, and deploys most of its capital in its first three or four years. Clinical-stage drug development meanwhile can take eight to ten years from first-in-human studies to approval. A company founded in the same month as the fund backing it may still be pre-approval when that fund needs to return capital.
This isn’t a problem either side can solve. It is a structural feature of the asset class. Yet founders routinely produce revenue forecasts that conveniently fit inside the fund’s investment horizon. Investors routinely ask for them.
The more honest framing is this:
A biotech investor is rarely buying the approved product. They are buying a defined step-up in the value of an asset.
That value step-up may be realised through a later financing, a licensing deal or a strategic sale. Once you say that out loud, the question changes.
Not: When do you generate revenue?
But: What does the next ₹40 crore prove and who will pay more for the company once it is proven?
Everything else follows from that reframe.
Vocabulary one
Readiness: why TRL breaks in biology
Technology Readiness Levels originated at NASA in the 1970s as a way to assess how mature hardware was before it flew. The framework subsequently spread to the US Department of Defense and Department of Energy and through Horizon 2020 and Horizon Europe, to the European Commission. It works well for things you can build, test in a controlled environment and then test in the field.
Biology is different. TRL has consequently been criticised for reducing research and innovation to a linear pipeline and for failing to capture the messy transition between scientific discovery and an actual product. The field has responded with adaptations.
BARDA (Biomedical Advanced Research and Development Authority), for example, maintains a separate readiness framework for medical countermeasures, covering everything from basic research on disease mechanisms through animal models and clinical studies. It maintains separate scales for products and for the systems in which they must operate.
Where the translation breaks
A founder says:
TRL 5.
They may mean:
The assay works reproducibly in our hands.
An investor may hear:
The technology has been demonstrated in a relevant environment and most of what remains is execution.
Those are different companies. And they warrant different valuations. The divergence often surfaces during diligence, precisely the worst possible moment.
The fix: never state a readiness level without the evidence
A readiness claim needs three things: the level, the evidence, and the system in which the evidence was generated.
Priceable:
TRL 4: efficacy demonstrated in a humanised mouse model, n=24, across two independent cohorts, with a 2.1× effect versus vehicle.
Not priceable:
TRL 4.
Better still, stop reducing readiness to a single number. Readiness actually moves along three independent axes:
| Axis | What it measures | Evidence that pins it |
|---|---|---|
| Biological | Does the mechanism work in a system that meaningfully predicts human outcomes? | Model system, n, effect size, replication, blinding, independent lab |
| Manufacturing | Can you produce it at the scale, cost and quality the next stage requires? | Batch size, yield, purity, comparability, GMP status, COGS trajectory |
| Regulatory | Does the regulator agree on the development path? | Pre-IND/pre-submission meeting, minutes, endpoint agreement, written CDSCO/FDA feedback |
Many companies that look mature on the biological axis are years behind on manufacturing and regulatory readiness. That asymmetry is one of the most common sources of post-term-sheet surprise in biotech.
A founder who says:
We’re technically ahead of ourselves. Manufacturing is the real constraint, and here is the plan to close that gap.
has converted a future diligence problem into present-tense credibility.
There is one more discipline worth adopting: state the inference gap. A result in a mouse model is not simply a weaker version of a result in humans. It is a different claim. It requires a bridging argument. Founders who can explain that bridge and acknowledge the known failure rate of translating their particular model or modality sound like people who understand their own uncertainty, and that matters.
Case studies
Two Indian companies, one lesson
The three-axis framework becomes more useful when you watch those axes come apart in the real world. India offers two instructive examples.
ImmunoACT: when the axes move together
ImmunoACT spun out of IIT Bombay’s Biosciences and Bioengineering department in 2018, building on research that began in 2013. In October 2023, it received CDSCO marketing authorisation for NexCAR19, a humanised second-generation anti-CD19 CAR-T therapy with a 4-1BB costimulatory domain, since renamed talicabtagene autoleucel, for relapsed or refractory B-cell lymphoma and leukaemia.
It was India’s first approved CAR-T therapy. Its price was roughly ₹30–40 lakh, compared with approximately ₹3–4 crore for imported alternatives. The pivotal data was subsequently published in The Lancet Haematology: an open-label Phase 1/2 study across six tertiary cancer centres in India involving 64 patients, with an overall response rate of 73% at the efficacy dose.
But the more interesting story for investors is what happened around the science. Laurus Labs, an early backer, invested more than $18 million, explicitly directing capital toward a GMP manufacturing facility alongside the R&D facility at Navi Mumbai and toward the Phase 2 study. Manufacturing wasn’t treated as a problem to solve after the biology worked. It was funded on the same clock.
The regulatory axis was similarly difficult. As the first domestic cell-and-gene therapy to pass through CDSCO, ImmunoACT wasn’t simply navigating an established regulatory pathway. It was helping create one. That is an enormous risk and an easy one to hide.
Our regulatory readiness is genuinely low because there is no precedent. Here is how we’re building the pathway with the regulator.
is communicating something valuable and true. A founder who quietly scores regulatory readiness as high because the underlying science is strong is storing up a future problem.
The diligence lesson
There is also a useful lesson in how data gets communicated. An October 2023 announcement described a 60-patient trial with approximately 70% ORR and highlighted low toxicity and minimal cytokine release syndrome. The peer-reviewed publication reported 64 patients and a 73% ORR, but also showed that grade 3 or worse haematological events were common, including neutropenia, thrombocytopenia, anaemia and febrile neutropenia.
Both descriptions can be true. But the second is the version a diligence team will find. The gap is not necessarily damaging.
The version of your data that survives scrutiny is the version worth leading with.
Shan5: when manufacturing becomes the risk
The counter-example is Shan5 at Shantha Biotechnics. It is painful precisely because science was never the problem. Shantha Biotechnics, founded in Hyderabad in 1993, was one of India’s genuine biotech success stories. It built its own cGMP facility early and that manufacturing capability became central to its credibility, helping it secure WHO prequalification and attract large-pharma interest.
In 2008, the company won a UNICEF contract worth approximately $340 million to supply its pentavalent vaccine Shan5 between 2010 and 2012. Sanofi paid more than $775 million for control of the company in 2009, substantially on the strength of the proposition that Indian manufacturing could serve emerging markets at low cost. Then, in 2010, reports emerged of white sediment adhering to glass vials that would not re-suspend even under vigorous shaking.
What followed was severe:
- WHO conducted site inspections, commissioned expert review and undertook independent testing.
- Prequalification was withdrawn.
- The vaccine was recalled and approximately 24 million doses were destroyed, affecting immunisation programmes across seven countries.
- Shantha was given a window to identify the root cause and implement corrective action. It could not do so within the deadline and was disqualified as a supplier.
- Revenue fell as low as ₹50 crore.
- Prequalification was regained only in 2014, following a fresh WHO inspection.
The founder later noted, accurately, that there had been no adverse reactions to the vaccine. But notice what actually failed, it was not the biology, not the clinical evidence, not the regulatory pathway. The product had been approved, prequalified and shipped at scale.
Manufacturing and quality failed. And they failed at a company whose investment thesis rested heavily on manufacturing.
A company can sit at the top of the biological and regulatory ladders and still be one manufacturing event away from major value destruction if its manufacturing risk is poorly understood and its economics cannot absorb a stoppage.
For founders, this is an argument for volunteering manufacturing constraints rather than burying them. For investors, it is an argument for interrogating root-cause investigation capability, batch-failure history, supply-chain concentration and margin structure with the same seriousness applied to efficacy data.
Vocabulary two
Revenue timelines: what if there is no revenue?
Revenue starts in FY29.
For a preclinical biotech company, that sentence is usually an exercise in fiction. Everyone in the room knows it. The founder says it because the investor asked. The investor asks because the model requires a number.
The scale of the mismatch
The cost of bringing a drug to market is enormous, and published estimates vary substantially depending on methodology.
| Estimate | Source | What it measures |
|---|---|---|
| ~$2.56B | Tufts CSDD | Capitalised cost per approved drug, including failures and cost of capital |
| $985M median / $1.34B mean | JAMA, 2020 | 63 drugs approved 2009–2018, based on public SEC filings |
| ~$2.77B median | Same JAMA study | Oncology drugs specifically |
| $314M–$2.8B | Literature range | Varies by definition, failures included and data source |
The spread matters. Different studies count different things as a “new drug.” Some include failures and the cost of capital. Others don’t. Some use company-supplied data; others rely on public filings. No seed-stage revenue model survives contact with all of these assumptions.
Replace the revenue curve with a value-inflection curve
The investor is not really asking:
When does money arrive?
They are asking:
When does this asset become worth materially more than I am paying today, what will it cost to get there, and who will re-rate it?
Answer that directly.
| Inflection | Evidence produced | Capital required | Who re-rates you |
|---|---|---|---|
| IND-enabling package complete | Tox, CMC, regulatory alignment | ₹X crore / 18 months | Crossover investors, strategic corp dev |
| Phase 1 safety read-out | Human safety, PK, biomarker movement | ₹Y crore / 14 months | Series B investors, licensing partners |
| Proof-of-mechanism in patients | First efficacy signal in target indication | ₹Z crore / 20 months | Pharma BD, regional licensing |
This is more useful than a revenue forecast. It tells the investor what the capital buys; what evidence is created; what the next financing can price; and who might value the company differently. Most importantly, it reflects the reality of biotech investing: value can be created and realised before commercial approval.
Then explain your revenue architecture
Pre-approval biotech companies can still generate revenue. But founders need to be explicit about where it comes from.
- Platform-as-a-service. Screening, sequencing, CRO-style contracts and discovery partnerships. The advantage is real revenue. The problem is that these activities often compete for the same scientists and infrastructure required for the lead programme. If this is your model, show the staffing split.
- Milestone and licensing revenue. Upfront payments, development milestones and regional licensing. This can create substantial value, but it is lumpy, non-recurring and dependent on partner behaviour that you do not control. Model it as optionality, not as your base case.
- Adjacent product revenue. A diagnostic, reagent or tool derived from the core platform that can reach the market years before the therapeutic. This is entirely legitimate. But the founder must demonstrate that the adjacent business does not fragment focus. The org chart should make that obvious.
One caution: grants are not revenue
Non-dilutive funding from BIRAC and similar programmes is valuable. It provides third-party validation and extends the runway. But it is not revenue. Give it its own line.
Risk-adjusted capital that reduces the equity required to reach the next value inflection.
That distinction may seem semantic. In a diligence process, it isn’t.
Vocabulary three
Probability: stop pretending certainty exists
Probability is the hardest vocabulary of all. Founders generally make one of two mistakes. They refuse to give a number: “Science doesn’t work that way.” Or they give one that bears little relationship to base rates: “We have an 80% probability of success.”
Investors make the mirror-image mistake. They apply blunt industry base rates to a specific programme and treat the result as objective. Neither approach works.
First, understand the base rate
Published clinical development success rates vary from roughly 7% to 20%. Much of that spread is methodological. Three questions matter.
What is the unit?
A 2025 Nature Communications analysis covering 20,398 clinical development programmes and 9,682 molecular entities found an overall success rate of around 5% in recent windows when measured by drug-indication programmes. Measure success by molecule, meaning success in any indication, and the figure rises to roughly 14.5%. Same underlying data. Same period. Nearly three times apart. So before discussing probability, establish the unit: are we talking about the drug, or the drug in this indication?
What counts as failure?
Studies differ on how long a programme must remain inactive before being considered dead. Some use one year. Others use 1.5, two or three years. Changing that threshold can move the measured success rate by several percentage points.
Whose programmes?
The same analysis found overall success rates of approximately 9.2–9.8% for top-20 pharmaceutical companies versus 8.0–9.0% for biotech firms. A separate benchmarking study found a wider gap. The exact numbers differ. The direction is consistent.
A venture-backed biotech should not automatically model itself on aggregate industry rates dominated by large-pharma execution.
The numbers worth carrying
| Metric | Figure | Source |
|---|---|---|
| Overall probability of approval from Phase 1 | 7.9% all programmes; ~5.7% for NMEs | BIO / Informa / QLS, 2011–2020 |
| Phase 2 transition | 28.9% | Same |
| Phase 3 transition | ~57% | Same |
| Oncology, Phase 1 to approval | 5.3%; 3.4% in a larger independent sample | BIO; MIT |
| Phase 1 transition trend | Fell from ~70% to ~50% over two decades | Nature Communications, 2025 |
Two adjusters are particularly worth understanding. First, Phase 1 is becoming harder. It now carries PK, PD and increasingly early efficacy questions that it once did not, alongside deliberate “quick-kill” strategies. Second, biomarker selection matters. Trials that use biomarkers to select patients generally show higher success probabilities than trials that do not.
The fix: decompose the probability
A credible answer to “What are your odds?” has four parts.
- Name the base rate and its unit.
Programmes in our modality and indication clear Phase 2 at roughly 29% in the BIO dataset. That’s per drug-indication programme, not per molecule. - Name your adjusters.
Explain why your programme should perform better or worse than the base rate. We’re up because the target is genetically validated and the population is biomarker-selected. We’re down because the endpoint hasn’t previously been accepted by the regulator in this indication. - Give the number and its width.
We think we’re at 40–50%. The range is driven primarily by uncertainty around the enrichment strategy. - Name the killer experiment.
The next ₹6 crore and nine months resolves enrichment. If the biomarker doesn’t stratify response, the thesis is wrong and we should stop.
That fourth move separates founders investors fund from founders they merely admire. Stating what would falsify your own programme isn’t weakness. It is one of the strongest signals that the science is being run by someone who wants to find the truth rather than defend a position. It also gives investors something they rarely get: a bounded question that their specific cheque can answer.
Make the risks legible
Separate the five risks
Founders often collapse everything into: “Science is hard.” Investors often collapse everything into: “It’s early.” Neither description is useful. Separate the risks.
| Risk | Question |
|---|---|
| Technical | Does biology work? |
| Execution | Can this team run the programme? |
| Manufacturing | Can you make it reproducibly and at the required cost? |
| Regulatory | Will the development pathway hold? |
| Market | Will anyone pay for it? |
Once these are separated, the conversation can focus on the risk that actually dominates. In many Indian biotech companies, that risk may be manufacturing or regulatory rather than biology. Shan5 is that whole argument in one word.
Say this, not that
The easiest way to improve the conversation is to change the language.
| Instead of | Say |
|---|---|
| We’re at TRL 5. | We’re TRL 5 on biology, pinned by [experiment]. Manufacturing is at TRL 3. Regulation is unprecedented, and we’re building the pathway with CDSCO. |
| Revenue begins in FY29. | Our next value inflection is IND-enabling completion. It requires ₹X crore over 18 months and should re-rate us with crossover investors and pharma BD. |
| We have ₹4 crore in revenue. | We have ₹4 crore of inflows, of which ₹3.2 crore is BIRAC grant support, validation and runway, not a business. The remaining ₹80 lakh is service revenue at 20% margin. |
| Our success probability is 80%. | The base rate is 29% per drug-indication programme. We adjust up for these two factors and down for this one, giving us a 40–50% range. Here’s the experiment that resolves the biggest uncertainty. |
| Science doesn’t work on probabilities. | Here’s our estimate, here’s the range, and here’s what would make us abandon the programme. |
| Manufacturing is straightforward. | Manufacturing is our binding constraint. Here’s our current batch performance, the gap to clinical scale and what it costs to close it. |
For the other side of the table
What investors need to fix
The translation burden isn’t the founder’s alone. Investors who want better biotech deal flow need to become better translators too.
Ask for evidence, not levels
Instead of: What TRL are you? Ask: What experiment moves you from 4 to 5?
Price the inflection, not the exit
Demanding a ten-year revenue model from a preclinical company signals a misunderstanding of the asset class.
Diligence manufacturing as hard as the science
Look at batch-failure history, root-cause investigation capability, supply-chain concentration and single points of failure. And, as Shan5 demonstrates, ask whether the company’s margin structure can survive a stoppage.
State your own base rates
If your investment committee is underwriting a 15% probability based on a house view of the modality, say so. And say whether that number is per molecule or per indication. The founder may have evidence that changes it.
Separate “too early” from “not good”
These are completely different messages. Yet founders routinely receive them as the same rejection.
Build scientific reading capability
If the investment team cannot independently interrogate the data package, say so. Bring in the scientific advisor. Route the conversation appropriately. Good biotech investing does not require every investor to become a scientist. It does require investors to know what they don’t know.
A practical checklist
The four artifacts every biotech founder should carry
If a founder takes only four things into a venture meeting, they should be these:
- A readiness statement.
Split across biological, manufacturing and regulatory readiness, with each claim pinned to named evidence. - A value-inflection table.
Show: Milestone → evidence produced → capital required → who re-rates you instead of relying on a conventional revenue forecast. - A decomposed probability.
Include: Base rate → unit → positive and negative adjusters → range → dominant source of uncertainty. - A falsification statement.
State: What experiment would make you stop? How much will it cost? How long will it take?
These four artifacts do not make hard science easier. They make hard science legible. And legibility is what gets funded.
Conclusion
The shared grammar
Capital available to Indian biotech is deepening. There is public risk capital through frameworks such as RDI, a growing base of deep-tech-literate domestic funds and increasing strategic interest from global pharma in India as both a discovery and manufacturing base.
What remains scarce is a shared grammar. A grammar that allows a founder and an investor to establish, inside a single meeting, that they are discussing the same company. That grammar is learnable.
It reduces to three habits:
- Never state a claim without the evidence that pins it.
- Never state a timeline without the capital that buys it.
- Never state a probability without the unit it is measured in and the experiment that would prove you wrong.
Science may remain difficult. The conversation doesn’t have to be.