India Makes the World's Medicines. It Buys the Molecules From China.

6-APA is not a household name. It is 6-aminopenicillanic acid, and it is the molecular core of every semi-synthetic penicillin — amoxicillin, ampicillin, the antibiotics that sit in medicine cabinets on every continent. India is the largest supplier of generic medicines on earth.

India imports 95.9% of its 6-APA from China. That was $407.6 million in FY2025.

It is not an isolated line. Of the 90 critical pharmaceutical ingredients the government itself tracks in a parliamentary annexure, 60 are more than 90% China-sourced. Twelve are at 100%.

Read the left column again. It is almost entirely antibiotics — and specifically, antibiotics made by fermentation, the industrial process of growing molecules in vats of living culture rather than assembling them through synthetic chemistry.

That single observation is the whole story. India's pharmaceutical import dependence is not diffuse. It is concentrated in one industrial capability the country largely exited decades ago and has not rebuilt.

The numbers hold up under cross-examination

We were sceptical of a parliamentary annexure as a sole source, so we checked all 90 codes against India's own trade portal — an independently published dataset with a different collection path.

All 90 matched. 83 agreed within 5%. Every apparent discrepancy traced to the annexure rounding to two decimals on sub-$1.5M line items. Two government sources, published separately, describing the same reality almost exactly.

Independent analysis lands in the same place. GTRI puts China at 44% of India's organic chemicals imports. Reporting on India's API position puts China's overall share at 70-74%, with "penicillin-related compounds exceeding 90 to 95 percent" — which is, to the decimal, what the trade data shows.

And it is getting worse, not better. Running India's imports through CEPII's BACI database across eight years: health and life sciences imports from China rose from $7.1 billion in 2017 to $12.5 billion in 2024, a 76% increase, with China's share of the category climbing from 30.7% to 36.3%.

The dependence deepened through the exact period in which reducing it became official policy.

The policy is working, at roughly one percent of the problem

India's Production Linked Incentive scheme for bulk drugs is not a failure, and it would be lazy to call it one. As of December 2025 it had commissioned 38 greenfield projects covering 28 of 41 prioritised products, established 56,800 tonnes of annual capacity, drawn ₹4,814 crore in actual investment — slightly ahead of what was committed — and averted an estimated ₹2,192 crore in imports.

₹2,192 crore is roughly $260 million. Set against a China API import bill running into billions annually, that is a real programme making real capacity, addressing a low single-digit percentage of the gap.

There is also a subtler problem inside the wins. Reporting on the scheme notes that some new facilities have pointed their output toward higher-value organic chemical exports rather than import substitution — precisely in the fermentation-based segments where Chinese suppliers hold structural cost advantages. Given the economics, that is a rational commercial decision by every individual firm and a poor collective outcome for the stated policy goal. Subsidising capacity does not, by itself, determine what that capacity chooses to make.

India's biggest deep-tech sector is pointed somewhere else entirely

Here is where our own data says something the trade figures alone cannot.

Health and life sciences is, by a distance, India's largest deep-tech sector. 6,751 startups. 99,455 patent filings. $2.5 billion raised across 918 funded companies. More startups, more patents and more capital than any other sector we track. Domestic filings outnumber foreign ones roughly three to one — 14,782 against 6,581 in 2024 — which is unusual and genuinely encouraging.

Now look at what all that activity is about.

The top technologies in the sector's patents: machine learning (5,824), deep learning (2,862), IoT (2,213), artificial intelligence (2,055), wearable sensors (1,505), AI-based imaging (1,150).

The best-funded companies: Enveda ($517M, AI-driven drug discovery), Qure.ai ($130M, AI medical imaging), Atonarp ($111M, mass spectrometry), SigTuple ($54.7M, AI microscopy), Pandorum ($52.1M, tissue engineering), Immuneel ($49.8M, CAR-T cell therapy).

This is a genuinely impressive computational-health ecosystem. It is also almost entirely orthogonal to a fermentation capacity gap.

The clearest signal is a classification detail. We searched the platform for fermentation technology. It exists — 312 patents — filed under Agri & Food. Not health. In a taxonomy built from what Indian inventors actually file, industrial fermentation reads as a food-processing technology, because that is predominantly what it is being used for.

Meanwhile "pharmaceutical synthesis" carries 549 patents and "pharmaceutical formulation" 256 — real chemistry activity, but formulation and synthesis are the steps that come after the imported molecule arrives.

What we are not saying

We are not saying Indian deep-tech is investing in the wrong things. AI diagnostics addresses a real and enormous problem, and a country with India's imaging-radiologist ratio has an unusually strong case for building it.

We are saying these are two different problems, and only one of them is getting attention proportionate to its size. Industrial fermentation is capital-intensive, low-margin, competes against entrenched Chinese scale advantages, and produces no demo. It is close to the least venture-fundable thing in life sciences. That is very likely why the gap persists — not oversight, but arithmetic that does not work for private capital at current prices.

Which makes it a policy problem rather than a startup problem, and suggests the useful question is not "why isn't anyone building this" but "what would have to change about the economics before anyone could."

We can't answer that from trade and patent data. We can say precisely where the hole is, how deep, and that the ecosystem currently best positioned to fill it is working on something else.

The three findings, plainly

• India's pharmaceutical import dependence is concentrated in fermentation-derived antibiotics — 6-APA at 95.9%, erythromycin at 97.7%, penicillins at 92.9% — not spread evenly across pharma.
• It deepened between 2017 and 2024, from 30.7% to 36.3% of the category, during the period when reversing it became national policy.
• India's largest, best-capitalised deep-tech sector has almost no activity in the specific capability the gap requires, and the relevant technology is classified outside health entirely.

For an investor, the honest read is that this is not an obvious opportunity — it is a visible, quantified, structurally difficult gap that private capital has so far declined for defensible reasons. That is still worth knowing precisely. Most people discussing "India's API dependence" cannot name the molecule.

Methodology

The 90-ingredient list is from a PIB parliamentary annexure on API China-dependence covering FY2024 and FY2025, cross-validated line by line against HS8 import data from India's Trade Intelligence & Analytics Portal for FY2024-25 (90/90 matched; 83/90 within 5%; all residual gaps attributable to two-decimal rounding on small lines).

Eight-year trend figures are from CEPII BACI V202601, India as importer, calendar years 2017-2024. BACI is mirror-reconciled and uses HS6, so its levels differ from TIA's HS8 fiscal-year figures; for health and life sciences the two agree closely (36.3% versus 36.0% China share), so we treat the trend as reliable. Sector assignment maps HS chapters 29, 30 and 35 to health and life sciences — chapter 29 (organic chemicals) genuinely spans pharmaceutical and industrial chemistry, so the category is broader than pharma alone, and the specific molecule-level figures above are the more precise claim.

Patent, startup and funding figures are Deeptech Navigator's registry as of 2026-07-22. The observation that fermentation is classified under Agri & Food reflects how the platform's technology taxonomy is derived from actual filings; it is a finding about where Indian fermentation patenting occurs, not a statement about the technology itself.

In our data

Sources

• India's API Trap: Incremental PLI Gains Mask a Quantifiable Chinese Dependency Crisis — IndraStra ↗
• China supplies over 30% of industrial goods; overdependence critical: GTRI — Business Standard ↗
• Bulk drugs push must fix India's API dependence — Policy Circle ↗
• CEPII BACI V202601 ↗