02The business

A box of lottery tickets that pays out like an annuity.

A company in India agrees to work out how to make a molecule for a small biotech in Boston. Nine times out of ten, maybe more, the drug dies. Only 7.9% of drugs entering human trials get approved.

When one survives, the factory that made the trial batches is written into the regulatory filing. Moving away from it means repeating the validation work and refiling. So the survivor keeps buying, often for the rest of its patent life.

That is why investors love this business. India’s listed names in it now trade at a median 46.5 times earnings.

What this means

Before paying that, ask the thing nobody seems to print: when these companies put in new money, what did they get back?

Exhibit 1

Only 7.9% of drugs entering human trials get approved

Chance of moving to the next stage, percent

Phase 1 to 2Phase 1 to 2: 52.0%52.0%Phase 2 to 3Phase 2 to 3: 28.9%28.9%Phase 3 to filingPhase 3 to filing: 57.8%57.8%Filing to approvalFiling to approval: 90.6%90.6%Phase 1 to approvalPhase 1 to approval: 7.9%7.9%

Source: BIO, Informa and QLS, Clinical Development Success Rates 2011-2020, 2021.

03New capital

What Rs 100 of growth cost

We took India’s listed contract drug makers and three years of their accounts, FY23 to FY26, leaving out five whose mergers break the numbers and four whose revenue fell or whose history is missing. That left 25.

The question: how much new capital did the typical company add for every Rs 100 of new revenue? The answer is about Rs 150.

Start the clock a year later, to take COVID’s windfall out of the base, and it is about Rs 130. Either way, growth here takes a lot of capital.

On that new capital, the median company earned about 16% before tax. EBIT here includes other income. Take it out and the median is 11.6% from FY23 and 14.2% from FY24. Take treasury income and the cash behind it out together and it is about 20%. A stricter version that also removes plants being built also gives about 20%.

What this means

So the range is about 12% to 20% before tax, about 9% to 15% after it, against a cost of capital we assume at 12%. That is ordinary.

Exhibit 2

About Rs 150 of new capital for every Rs 100 of new revenue

Median of 25 listed companies with comparable accounts

FY23 to FY26FY23 to FY26: Rs 152Rs 152FY24 to FY26FY24 to FY26: Rs 131Rs 131
Return on new capitalMedian
FY23 to FY2616.6%
FY24 to FY2615.5%
FY23 to FY26, without cash and plants under construction20.6%

The median is jumpy. The companies either side of it sit at Rs 143 and Rs 158, so read it as about Rs 150.

Source: stockscans.in, retrieved 30 October 2026; Investace Capital Research. Capital is equity, reserves and borrowings, including leases, at year end.

04The price

What 26x has to assume

Now look at what that return is being asked to support.

Work out what a business growing 20% a year for a decade, earning 15% on new money, should trade at, and you get about 11 times EBITDA on reasonable assumptions. The median stock in this group is at 26.

To justify 26 you need something like 30% growth at a 20% return, for ten years.

What this means

The median company has delivered about 16% EBITDA growth at about 16% returns.

Exhibit 3

What growth and returns can support, against a median of about 26 times

EV/EBITDA. Down: return on new capital. Across: growth a year for ten years

Return15%20%25%30%
15%8.7x11.0x14.3x19.1x
20%10.4x13.9x19.0x26.3x
25%11.4x15.6x21.7x30.6x
30%12.1x16.8x23.6x33.5x

20% growth at a 15% returnClosest to the median of about 26 times

Our inputs, so change them: 12% cost of capital, a ten-year growth stage, terminal growth of 6% at a 15% return, NOPAT at 58% of EBITDA. One counter: trailing EBITDA is depressed where plants are not yet full.

Source: Investace Capital Research, Justified multiple sheet of the Excel; market values stockscans.in, 30 October 2026.

05The build

The building is getting faster.

This is not a warning about debt.

Across 22 companies, capex rose from Rs 7,473 crore in FY23 to Rs 12,172 crore in FY26. Plants under construction nearly doubled to Rs 13,679 crore. Revenue grew 27% in the same years.

Operating cash flow before interest, at Rs 14,924 crore, still covered it.

What this means

Nobody is about to go broke. The question is whether those plants fill.

Exhibit 4

In three years, sales grew 27%. Plants under construction grew 89%.

22 listed companies combined, Rs crore, financial years to March

Spent on plants in the year (capex), FY23: Rs 7,473 crore7,473Plants not yet running (CWIP), FY23: Rs 7,244 crore7,244FY23Spent on plants in the year (capex), FY24: Rs 8,229 crorePlants not yet running (CWIP), FY24: Rs 9,762 croreFY24Spent on plants in the year (capex), FY25: Rs 9,384 crorePlants not yet running (CWIP), FY25: Rs 11,097 croreFY25Spent on plants in the year (capex), FY26: Rs 12,172 crore12,172Plants not yet running (CWIP), FY26: Rs 13,679 crore13,679FY26

Spent on plants in the year (capex)Plants not yet running (CWIP)

Source: stockscans.in, retrieved 30 October 2026; Investace Capital Research. Sum of 22 listed companies with comparable accounts; merged companies left out.

06The risk

A plant has to fill on time.

A plant loses money in its first years, and it needs to fill on time.

Run a Rs 1,000 crore plant through a simple model and fill it two years late, and by year eight you are about Rs 900 crore behind. Nothing about that shows up in a capex headline.

Management always says they only spend when they can see the orders. Maybe.

What this means

When you hear it, ask which drug.

Exhibit 5

Same plant, filled two years late: about Rs 900 crore behind by year eight

Cumulative cash after the Rs 1,000 crore cost, Rs crore. An illustration, not any company

-1,00001,0002,000Y1Y2Y3Y4Y5Y6Y7Y8Fills on time, Y1: Rs -1,136 croreFills on time, Y2: Rs -1,135 croreFills on time, Y3: Rs -980 croreFills on time, Y4: Rs -679 croreFills on time, Y5: Rs -290 croreFills on time, Y6: Rs 153 croreFills on time, Y7: Rs 596 croreFills on time, Y8: Rs 1,039 crore1,039Fills two years late, Y1: Rs -1,165 croreFills two years late, Y2: Rs -1,275 croreFills two years late, Y3: Rs -1,287 croreFills two years late, Y4: Rs -1,198 croreFills two years late, Y5: Rs -990 croreFills two years late, Y6: Rs -689 croreFills two years late, Y7: Rs -299 croreFills two years late, Y8: Rs 144 crore144

Fills on timeFills two years late (dashed)

Source: Investace Capital Research, J-curve sheet of the Excel. Rs 1,000 crore plant, revenue at full use 1.5 times cost, 45% variable cost, Rs 180 crore fixed cost, 12-year life, 25% tax.

07The molecule

The drug matters less than you think.

A supplier is paid for kilograms, not for the drug’s price.

Austedo, a billion-dollar movement-disorder drug, needs about a tonne of active ingredient a year. Forum researchers estimate its supplier earned about Rs 100 crore at peak. That is roughly 1% of what the drug sold for. “We supply a blockbuster” can mean almost anything.

Writers on this sector agree on a lesson the companies don’t advertise. Nearly every big gain and every big loss in Indian CDMO traces to a handful of molecules. PI Industries and pyroxasulfone. Syngene and a canine antibody. Laurus and Paxlovid. Blue Jet and bempedoic acid.

What this means

Which molecules pay the bills, and how many kilograms is that?

Exhibit 6

A blockbuster can be worth 1% to the factory that makes it

What a supplier earns against what the drug sells for

DrugActive ingredientSupplier revenueShare of drug sales
AustedoAbout 1 tonneAbout Rs 100 crore at peakAbout 1%
ElinzanetantAbout 2.5 tonnesAbout Rs 75 croreAbout 2%
Anthem’s top productn/aUS$35-40 million2.5-3% of US$1.4 billion

Three cases do not make a law. Change the inputs on the Molecule capture sheet.

Source: ValuePickr forum estimates, via a secondary sector report; Priyum Daga, 3 September 2026. Estimates, not company disclosures.

08Contracts

“But it’s contracted.”

The strongest argument for these valuations is predictability. A contract manufacturer knows its molecules, customers and prices going into the year. A generics company doesn’t.

That is true, and it is half the story. A contract fixes who and what. It does not fix how much the customer takes this quarter. Anthem’s revenue fell 23% in the June quarter, even with 60% of its year already in the order book. Blue Jet’s pharma intermediates fell 35% last year, which management attributes to a customer running down its stock, then came back.

What this means

You know the average better. You also get a fatter left tail. A valuation can reflect the first. It still has to survive the second.

Exhibit 7

Known customers, known prices, and revenue still jumped

Two recent cases

CompanyWhat fellByWhat happened
Anthem BiosciencesRevenue, June quarter 2026 (Q1 FY27)22.6%About 60% of the year was already in the order book.
Blue Jet HealthcarePharma intermediates, FY2635%, to Rs 298 croreManagement attributes it to a customer running down its stock. Supply restarted in Q1 FY27.

Source: Anthem Q1 FY27 earnings call; Blue Jet Q4 FY26 and Q1 FY27 earnings calls.

09China

And China?

The popular story is that US law pushes drug companies from China to India. Be careful with it.

The BIOSECURE Act restricts federal money, not private drug companies, and protects existing contracts to about 2033. New molecules may move.

Meanwhile WuXi AppTec, the company supposedly losing all this work, just reported nine-month revenue up 38.6% and a backlog up 36.4%.

What this means

India may still win. The numbers don’t show it yet.

Exhibit 8

The company supposedly losing the work has a growing backlog

What the law says, and WuXi AppTec’s nine months to September 2026

Latest
BIOSECURE Act, scopeFederal money only
Existing contractsProtected to about 2033
WuXi AppTec revenueRMB 45.54 billion, up 38.6%
WuXi AppTec backlogRMB 76.74 billion, up 36.4%

Source: Greenberg Traurig, 3 September 2026; WuXi AppTec, 26 October 2026.

10The sector

So is it a bad sector?

No. It’s a good one with a high price, and the average hides a lot.

Some companies earned over 30% on new capital in these three years. Some earned nothing. Who those are depends a little on which year you start from, and the Excel shows both.

What this means

That spread is where the work is.

Exhibit 9

Three questions for your next annual report

If all of this had to come down to three

Question
1Which molecules pay the bills, and how many kilograms is that?
2What did the last rupee of capital earn?
3What must go right from here for you to earn 15%?

Source: Investace Capital Research.

11Open questions

What we still can’t answer

We would rather tell you than pretend.

We don’t know what share of any company’s revenue comes from molecules it wasn’t making three years ago. It may be the single most useful number in the sector, and nobody discloses it.

We don’t know who really pays the new US tariff. And we don’t know if global peptide capacity is about to be too much, despite how confidently people say so, because no public number exists.

What this means

If you have run one of these plants, write to research@investacecapital.com and tell us where this is wrong.

Exhibit 10

Three numbers nobody publishes

What is known so far

Open questionWhat is known
Revenue from molecules absent three years agoNobody discloses it.
Who pays the US tariffSection 232 sets 100% by default on patented imports. Whether an intermediate is caught, and who absorbs it, is open.
Global peptide capacityNo public global number exists. The only demand figure found is one Goldman estimate of about 99 tonnes by 2030.

Source: WilmerHale, 16 October 2026; Goldman via ANI, 7 June 2026.

12The research

What the research covered

  • 34 listed companies’ accounts, FY23 to FY26
  • 146 quarterly earnings calls across 32 companies
  • 89 claims in one 56-page sector report, checked at the source
  • The Excel model, free to download, every input marked

What this means

Every number here was checked at its source or labelled.

Exhibit 11

Of 89 claims in one sector report, 55 were confirmed

Result of checking each claim at the source, number of claims

ConfirmedConfirmed: 5555DifferentDifferent: 1919Out of dateOut of date: 77Could not be verifiedCould not be verified: 88

Source: Investace Capital Research, 30 October 2026.

The model behind this study

Inside the Excel

The study’s capital, valuation and plant numbers can be followed here, and the formulas are live. Blue numbers are inputs. Change them and the answer moves. Company data: stockscans.in, retrieved 30 October 2026.

Download the model (.xlsx)9 sheets · 135 KB · opens in Excel, Numbers or Google Sheets

Preview every sheet9 sheets, read-only
  1. Read meWhat each sheet does, the conventions and the limits.
  2. UniverseAll 34 listed names by business model: revenue, margin, material cost and valuation. The medians are shown here; the company rows are in the download.
  3. CapitalWhat each extra Rs 100 of revenue cost in capital, the return on it, and whether operating cash flow before interest covered capex, from both start years. The medians are shown here; the company rows are in the download.
  4. What price assumesThe EBITDA growth each price needs to earn your required return, against what the company delivered. Change the three inputs. The inputs and the medians are shown here; the company rows are in the download.
  5. Shortlist filterTwo tests on every company: return on new capital in both windows, and growth delivered against growth needed. A filter for where to read first, not a buy or sell recommendation. The company rows are in the download.
  6. Justified multipleWhat EV/EBITDA a given growth rate and return on new capital can support.
  7. Molecule captureDose, patients and days, then price per kg: how much of a drug’s sales its supplier can earn.
  8. Funnel valueWhat a molecule at each clinical phase is worth to a contract manufacturer.
  9. J-curveOne plant from commissioning to full. Change the ramp and watch the return and the payback.

What each sheet does, the conventions and the limits.

1India's contract drug makers: sector model
2Kavish Raj Singh · Investace Capital Research · investacecapital.com/research/cdmo
3Rs crore unless a column says otherwise.
4Company data: stockscans.in. Prices at the 30 Oct 2026 close.
5The question: Where does a contract drug manufacturer earn more than its cost of capital, for how long, and what does the price already assume?
7Sheets
8Universe: every listed name, sorted by business model. Revenue, margin, cost, cash cycle, valuation on one EV basis.
9Capital: what each extra Rs 100 of revenue cost in capital, the return on that capital, and cash flow against capex before and after interest. The sheet to argue from. It shows both start years, FY23 and FY24. Capital employed is a formula: change a blue equity, reserves or borrowings cell at the far right and it moves.
10Leverage: net debt / EBITDA, interest cover, short-term debt and any guided capex, per company.
11What price assumes: the EBITDA growth each price needs to earn your required return, over 3 and 5 years, against what the company delivered from two base years, name by name, with sector medians at the foot. Change the inputs.
12Shortlist filter: two tests on every company, return on new capital in both windows and growth delivered against growth needed. A filter for where to read first.
13Justified multiple: what EV/EBITDA a given growth rate and return on new capital can support, at a cost of capital of 11%, 12% and 13%.
14Molecule capture: dose x patients x days, then price per kg: how much of a drug's sales its supplier can earn. Rows are backed out to fit public estimates, so they show a method, not a measurement.
15Funnel value: what a molecule at each clinical phase is worth to a CDMO, as the present value of pre-tax EBITDA, with retention flat and by stage.
16J-curve: one plant from commissioning to full. Change the ramp and watch ROCE and payback. An example on assumed inputs, with fixed cost tested against the accounts.
17Red and green sit inside the numbers. Green means the number passes the test on that sheet, red means it fails, and numbers in between stay uncoloured.
18Green is not a buy signal and red is not a sell signal. A colour reports one test on one number, nothing more.
19Conventions
20Blue on pale blue: an input. Change it. Black: a formula. Every input carries its source on the sheet.
21Fiscal years end in March. TTM is the latest twelve months the data holds, and it ends on different dates across names.
22Inputs marked Our assumption are ours, to be replaced. Inputs marked Published estimate come from secondary sources and are leads, not checked facts.
23Before tax and after tax: EBIT returns on Capital are before tax. The 12% cost of capital and 15% required return are after-tax numbers. At a 25% tax rate, 16% before tax is about 12% after tax and 20% is about 15%, and the colours use 16% and 20%.
25Limits
26Capital employed = equity + reserves + borrowings, which include leases, at year end. Cash is not netted, so cash-rich names look less efficient than they are. Non-controlling interest and share-warrant money are left out of capital and EV, while EBIT and EBITDA are 100% of subsidiaries.
27Net debt is borrowings incl. leases less cash and bank, at 31 March. It is not complete: liquid funds sit outside the cash field. Read it from the balance sheet for any name discussed.
28EV is market cap + our net debt, one basis for every name. Market cap is at the pull date and net debt at 31 March, so a raise in between is in one and not the other.

The record

Sources

Company data comes from stockscans.in, retrieved 30 October 2026. A secondary source is marked as one: it is a lead until checked at the primary.

Every source10 sources, with what each one supports
  1. 30 October 2026

    stockscans.in

    Company data: annual statements FY23 to FY26, market values and earnings call transcripts.

  2. 26 October 2026

    WuXi AppTec, Q1 to Q3 2026 results

    Nine-month revenue RMB 45.54 billion, up 38.6%. Backlog RMB 76.74 billion, up 36.4%.

  3. 16 October 2026

    WilmerHale, Section 232 pharmaceutical tariffs

    The 100% default rate on patented imports.

  4. 3 September 2026

    Greenberg Traurig, BIOSECURE Act update

    Federal scope of the law. Existing contracts protected to about June 2033.

  5. 3 September 2026

    Priyum Daga, Anthem Biosciences

    Anthem’s top product: US$35-40 million, 2.5-3% of US$1.4 billion. A secondary source.

  6. July 2026

    Anthem Biosciences, Q1 FY27 results and earnings call

    Revenue down 22.6%. About 60% of the year in the order book.

  7. June 2026

    Goldman estimate, via ANI

    About 99 tonnes of peptide demand by 2030. One bank’s model, reported second hand.

  8. FY26 and Q1 FY27

    Blue Jet Healthcare, earnings calls

    Pharma intermediates down 35% to Rs 298 crore in FY26 on customer destocking, restarting in Q1 FY27.

  9. March 2025

    ValuePickr forum, Neuland thread, post 593

    Austedo: about a tonne, about Rs 100 crore at peak, roughly 1% of drug sales. A forum estimate, a lead and not a fact.

  10. 2021

    BIO, Informa and QLS, Clinical Development Success Rates 2011-2020

    7.9% from Phase 1 to approval: 52.0%, 28.9%, 57.8% and 90.6% by stage.

Method and limitsThe sample, the returns, the checks and the limits

Correction, 2 November 2026. The first version counted leases twice, because the data provider’s borrowings already include them. The median is now Rs 151 per Rs 100 of new revenue (was Rs 152), and the return on it 16.6% from FY23 and 15.5% from FY24 (were 16.1% and 15.9%). “The same 25” added up to Rs 176 and 15.6%; the Rs 187 and 14.6% printed earlier were 28 companies, also with leases twice. The return that removes cash and plants under construction is 20.6%, and it does not remove liquid investments, as the earlier label said. The Excel is updated. The conclusion is unchanged.

Capital employed is equity, reserves and borrowings, including leases, at year end. Cash is not netted, so companies that have just listed look less efficient than they are. Medians use the 25 companies whose accounts are comparable across the window; with merged and acquiring companies included they are Rs 155 and 16.3%. Added up instead of taking a median, the same 25 needed Rs 176 per Rs 100 and earned 15.6%, because the larger companies were less efficient. The capex and cash flow figures add up 22 companies with comparable history. Market values are as of 30 October 2026. A company with no figure is left blank, never zero. All of it looks backward: plants still filling understate what they may earn.

Join free on WhatsApp