Parth
on
Parth Shah
Register Valuer | CA | CPA | 15+ Years of Experiance
Parth Shah is the Founder and Team Leader of the company, bringing extensive expertise in business valuation and financial advisory.
Deep tech startups in India cannot be reliably valued using discounted cash flow, because they are typically pre-revenue for five to ten years with no comparable transactions. Practitioners instead use risk-adjusted NPV, real options analysis, cost-to-duplicate for IP, and stage-based methods such as Scorecard or Berkus – usually triangulating two or three to produce a defensible range rather than a single figure.
A founder building a quantum computing company came to us last year with a ten-year DCF model. It was beautiful. Every assumption was documented, the WACC was defended, the terminal growth rate was conservative.
It was also, in any meaningful sense, fiction – because year six revenue depended entirely on whether a technology that did not yet exist would work, and no discount rate can price that.
This is the central problem of deep tech valuation, and it is a problem India is going to face at scale. Deep tech has moved from a niche to roughly 15% of all venture and private equity activity in the country, and the regulatory framework was rewritten around it in early 2026.
How much have Indian deep tech startups actually raised?
The headline numbers are strong and accelerating.
| Year | Deep tech funding (approx.) | Notes |
| 2023 | Rs. 850M–900M (USD) | 480+ new deep tech ventures launched, nearly double the prior year |
| 2024 | ~USD 1.6B | A 78% increase over 2023 (Tracxn) |
| 2025 | USD 2.1B–2.3B | Up ~37%; AI accounted for 91% of capital deployed |
| H1 2026 | USD 1.1B by June | Already near 80% of the 2025 full-year total |
The structural shift matters more than any single year. Over the past decade, deep tech investment in India totalled roughly USD 27.9 billion across 2,178 deals involving 1,217 companies. Deep tech’s share of total VC–PE activity has risen to about 15%, up from 4% in 2016 – a move from experimental allocation to core allocation.
India now hosts more than 4,200 deep tech startups, with over 550 added during 2025 alone. In 2025, AI accounted for 188 deals worth USD 1.22 billion, a 58% year-on-year increase, while non-AI deep tech sectors – semiconductors, space, robotics, biotech, climate – logged 147 deals worth USD 1.19 billion.
Why the published numbers disagree
If you compare sources, you will find 2025 deep tech funding reported as USD 2.3 billion, USD 2.1 billion, and in one case above USD 600 million. All three are defensible. The gap is definitional, not arithmetic.
| Source | 2025 figure | What drives the difference |
| Nasscom–Zinnov | ~USD 2.3B | Broad definition; includes AI-application companies |
| India Deep Tech Alliance | ~USD 2.1B | Deal-count driven; splits AI from non-AI deep tech |
| Narrower trackers | ~USD 600M+ | Excludes AI-application layer; counts core R&D only |
This matters commercially. If you are benchmarking your startup’s valuation against “the market,” the multiple you cite depends entirely on which universe you are comparing against. An AI-application company benchmarked against core semiconductor deals will look expensive; the reverse will look cheap. Always state your comparable set before you state your multiple.
Why standard valuation methods break down here
Deep tech violates the assumptions underlying almost every conventional technique.
No revenue, for a long time. A semiconductor or quantum venture may be pre-revenue for five to ten years. DCF requires cash flows; there are none, and projected ones rest on technical milestones rather than commercial ones.
Binary technical risk. A SaaS company that misses its target grows slower. A deep tech company whose core technology fails is worth its IP and its team, and little else. Conventional discount rates model gradual underperformance, not binary outcomes.
No comparables. Comparable-company analysis needs comparable companies. For genuinely novel technology, there are none – and forcing a comparison to an adjacent listed company imports the wrong risk profile entirely.
Value sits in intangibles. Patents, research pipelines, exclusive licences and specialist teams. None of this appears meaningfully on the balance sheet, which is why net asset value produces absurdly low answers for exactly the companies raising the largest rounds.
Capital intensity distorts the picture. Deep tech burns heavily before revenue, so early-stage book equity is often negative even as enterprise value rises.
The methods that actually work
No single method is mandated. Selection depends on stage, sector and the purpose of the report.
| Method | Best suited to | How it works | Main limitation |
| Risk-adjusted NPV (rNPV) | Biotech, pharma, medtech | Cash flows weighted by probability of technical and regulatory success at each stage | Probability estimates are judgment-heavy |
| Real options | Staged R&D, platform technologies | Treats each funding round as an option to continue, priced on volatility | Mathematically complex; inputs hard to defend |
| Cost-to-duplicate | Pre-revenue, IP-heavy | Estimates what a competitor would spend to replicate the technology and team | Ignores future upside; produces a floor |
| Scorecard / Berkus | Seed and pre-seed | Qualitative weighting of team, technology, market, traction against benchmarks | Depends on quality of the benchmark set |
| VC method | Series A onward with an exit view | Works backwards from a projected exit value and required investor return | Sensitive to exit multiple assumptions |
| Comparable transactions | Sectors with visible deal flow | Benchmarks against recent rounds in the same sub-sector | Requires an honestly chosen comparable set |
Risk-adjusted NPV is the workhorse for anything with a regulatory pathway. Rather than assuming revenue arrives, you assign a probability of success to each stage – preclinical to Phase I, Phase I to Phase II, and so on – and weight the cash flows accordingly. It forces the honesty a plain DCF allows you to avoid.
Cost-to-duplicate gives you a defensible floor. What would a well-funded competitor need to spend, and how long would it take, to reach where you are? For a company with granted patents and a specialist team, this often produces a surprisingly robust number – and it is the method most likely to survive scrutiny in a dispute.
In practice, a credible deep tech report triangulates two or three methods and reports a range. Any valuer offering a single precise figure for a pre-revenue deep tech company is offering false confidence.
What the DPIIT 20-year window changes about valuation
This is the part most commentary has missed, and it has direct technical consequences.
In early 2026, DPIIT revised the Startup India framework to formally carve out deep-tech startups as a distinct category. Eligible deep-tech ventures can now retain recognised startup status for up to 20 years, double the previous limit, with a turnover threshold of Rs. 300 crore against Rs. 200 crore for other startups.
The policy rationale is sound – science-led commercialisation cycles genuinely run longer than consumer internet cycles. But the valuation implications follow directly:
- The explicit forecast period extends. A ten-year model no longer captures the commercialisation curve. Terminal value moves further out, which means it carries more uncertainty and should carry less weight.
- Discount rates need reconsideration. A longer horizon with binary technical risk cannot use a rate borrowed from a growth-stage SaaS comparable.
- Sensitivity analysis stops being optional. Over a 20-year horizon, small changes in probability-of-success assumptions swing the answer enormously. A report without a sensitivity table is not a complete report.
- Stage-gated valuation becomes the sensible norm. Rather than one number, value the company at each technical milestone, with probabilities attached.
If your existing valuation model was built against a ten-year startup lifecycle, it is now mis-specified for a deep tech company that qualifies under the revised framework.
When a deep tech startup legally needs a valuation
Separate from what investors want, several events trigger a statutory requirement.
| Trigger | What is required | Framework |
| Issuing shares at a premium to investors | Registered valuer report | Companies Act 2013, Sec 62(1)(c) + Rule 13 |
| Preferential allotment | Registered valuer report | Companies Act 2013 |
| Foreign investment into the startup | Pricing certificate at or above fair value | FEMA NDI Rules, 2019 |
| Granting or exercising ESOPs | FMV determination | Income-tax framework |
| Transfer of shares below fair value | FMV determination | Income-tax framework |
| Merger, acquisition or swap ratio | Registered valuer report | Companies Act 2013 |
| AIF portfolio holdings | Independent valuation, at least semi-annually | SEBI AIF Regulations |
That last row is newer and often missed. AIFs must now obtain independent valuation of portfolio holdings at least semi-annually – previously annual for Category I and II funds. If a deep tech AIF holds your shares, you will be valued twice a year whether you plan for it or not.
The recurring trap: valuation reports are purpose-specific. A FEMA pricing certificate cannot be reused for income tax purposes, and neither substitutes for an Ind AS 113 financial reporting valuation – same company, same date, same numbers, different standards and different authorised professionals.
What changed in 2026
Angel tax is gone. Section 56(2)(viib) no longer applies from FY 2025-26. This removed a real obstacle for deep tech companies raising at premiums that were hard to justify on conventional metrics. It did not remove valuation obligations – fair market value still governs ESOP perquisites, transfers below value, FEMA pricing and every Companies Act trigger above.
The Income-tax Act, 2025 took effect on 1 April 2026, with the Income-tax Rules, 2026 replacing the 1962 rules. The old Rule 11UA framework now sits at Rule 57. If your ESOP scheme, grant letters or board resolutions cite the old sections, they need review.
Government capital is arriving. The Rs. 1 lakh crore Research, Development and Innovation Scheme was approved by Cabinet on 1 July 2025 and launched on 3 November 2025, running over six years with provisions for Deep-Tech Funds of Funds. The Technology Development Board and BIRAC were appointed Second Level Fund Managers, TDB issued its first call for proposals on 4 February 2026, and by mid-2026 had selected its first cohort of startups. Budget 2026-27 allocated Rs. 20,000 crore for private-sector R&D and announced a dedicated Deep Tech Fund of Funds.
Practically, this means more Indian deep tech companies will be raising from institutional and government-linked capital over the next three years – and institutional capital comes with valuation documentation requirements that founder-friendly angel rounds did not.
Five mistakes we see repeatedly
- Presenting a ten-year DCF for a pre-revenue technology. Sophisticated deep tech investors discount it on sight. It signals the founder does not understand their own risk profile.
- Benchmarking against the wrong comparable set. An AI-application company citing core semiconductor multiples, or the reverse. State your universe before your multiple.
- No probability weighting. If your model does not assign a probability of technical success, it is a plan, not a valuation.
- Ignoring IP in the valuation entirely. Patents, exclusive licences and research pipelines are frequently the majority of enterprise value and are frequently unvalued.
- Commissioning the wrong report. Getting a FEMA certificate when you needed a Companies Act registered valuer report means paying twice, under deadline pressure, mid-round.
Expert insight
“With deep tech, the honest answer is a range, and the useful work is in the assumptions rather than the number. When we assign a 40% probability of technical success at a given stage, that figure has to be defensible to an investor’s diligence team and to a regulator years later. Founders sometimes want a single confident figure. The ones who get funded are the ones who can walk an investor through why the range is what it is.”
CA Parth Shah – Founder, My Valuation. IBBI Registered Valuer for Securities or Financial Assets under Section 247 of the Companies Act, 2013 (Reg. No. IBBI/RV/06/2020/13086). FCA, licensed CPA (US), DISA.
Frequently asked questions
1. How are deep tech startups valued in India?
Through a combination of risk-adjusted NPV, real options analysis, cost-to-duplicate for intellectual property, and stage-based methods such as Scorecard or Berkus. Standard DCF is generally unsuitable because deep tech startups are pre-revenue for extended periods with binary technical risk.
2. Can a pre-revenue deep tech startup be valued at all?
Yes. Pre-revenue valuation relies on qualitative and replacement-cost approaches – the strength of the team, granted and pending patents, technical milestones achieved, market size and comparable early-stage transactions.
3. How much did Indian deep tech startups raise in 2025?
Between roughly USD 2.1 billion and USD 2.3 billion depending on the source and how deep tech is defined, representing growth of around 37% year on year. AI accounted for roughly 91% of that capital.
4. What is the DPIIT deep tech startup category?
A distinct category introduced under the revised Startup India framework in early 2026, allowing eligible deep-tech ventures to retain recognised startup status for up to 20 years with a turnover threshold of Rs. 300 crore, reflecting longer science-led commercialisation cycles.
5. Do deep tech startups still need a valuation after angel tax was abolished?
Yes. Section 56(2)(viib) no longer applies, but valuation remains mandatory for share issues under the Companies Act, foreign investment under FEMA, ESOP perquisite calculation, and share transfers below fair value.
6. Which valuation method do investors prefer for deep tech?
Most institutional deep tech investors expect to see risk-adjusted NPV or a stage-gated model with explicit probability assumptions, cross-checked against comparable transactions. A single-method valuation is usually treated as incomplete.
7. How long does a deep tech valuation take?
Typically five to seven business days from receipt of complete financials, cap table, IP documentation and technical milestone data. IP-heavy engagements requiring patent review may take longer.
Raising a deep tech round?
The valuation model you take into a term sheet discussion says as much about you as the technology does. Investors read it as a proxy for how you think about risk.
My Valuation is an IBBI Registered Valuer-led firm working with startups, AIFs and financial institutions across India. Our clients have raised over Rs. 1,500 crore, with more than 95% of reports accepted by VCs and PEs without major rework, delivered in five to seven business days.
