India Shut Chinese Vendors Out of Its Telecom Networks. The Components Inside Didn't Change.

Split India's telecom equipment imports into two piles and something strange appears.

Finished equipment — base stations, transmission apparatus, handsets: 13.4% came from China in FY2024-25 ($570M of $4.26B).

Parts — antennas, populated circuit boards, sub-assemblies: 52.3% came from China ($8.04B of $15.37B).

Same chapter of the customs code. Same year. A four-fold difference.

The first number is roughly what a decade of security policy was designed to produce. The second suggests that policy stopped at the enclosure and never reached inside it.

The gap is the policy, drawn in trade data

India progressively restricted Chinese vendors from its telecom networks after 2020, culminating in trusted-source rules that in practice excluded Huawei and ZTE from network buildouts. The finished-equipment line shows it worked: base stations specifically are down to $0.5M from China against $3.9M worldwide — 12.4%. For a category China dominates globally, that is a policy fingerprint, not a market outcome.

Now look at what India buys instead. The single largest telecom import line in the entire dataset is HS 85177990 — "other parts" of telephone and network apparatus — at $7.15 billion, of which 51.7% is Chinese. Populated printed circuit boards run 55.4% Chinese. Antennas and aerial reflectors, 69.3%.

The equipment India installs is no longer Chinese. The components inside that equipment substantially still are. Whether that satisfies the original security objective is a question for people who set it — but it is not the same achievement as the headline suggests, and the trade data does not let you claim both.

The customs data probably understates it

Here is a wrinkle worth being precise about.

We ran the same sector split through two independent trade sources: India's own TIA portal (customs, as-reported, fiscal year) and CEPII's BACI database (calendar year, and critically, mirror-reconciled — it cross-checks what importers report against what exporters report).

For most sectors the two agree almost exactly. Agri-food differs by 0.1 percentage points. Energy, 0.1. Health and life sciences, 0.4. Materials, 1.0. Robotics, 1.3.

Then three sectors break the pattern, all in the same direction:

The close agreement everywhere else argues against a general data-quality problem — the two sources track each other well when nothing complicated is happening. The divergence is concentrated precisely in electronics, the categories where transshipment through Hong Kong and Singapore is heavily documented, and where China's export records would count a shipment as India-bound that Indian customs logs under a different last origin.

We can't prove that mechanism from this data alone, and we're not going to pretend otherwise. But the shape of the discrepancy is consistent with it, and the direction matters: if it holds, India's official statistics understate Chinese dependence in exactly the three sectors where the strategic stakes are highest. The honest version of the headline is that 52.3% is a floor, not a ceiling.

The patents belong to someone else too

If import dependence were the whole problem, domestic IP would be the escape route. In communications it isn't one yet.

Deeptech Navigator's registry counts 54,384 communications patent filings in India — the second-largest patent pool of any sector we track. But 61.9% of filers are foreign, and the annual split is starker than the cumulative one: in 2024, 7,929 foreign filings against 1,695 domestic — 82% foreign.

The top filers explain why. Qualcomm alone accounts for 9,786 filings. Nokia, 2,611. Samsung, 2,588. Huawei, 2,332 — the same company excluded from selling equipment into Indian networks is the second-largest foreign patent filer in India's communications sector. Xiaomi, 1,873. Ericsson, 1,126.

One Indian company appears anywhere near that scale: Jio Platforms, at 1,083 filings (310 granted). After that, the domestic list drops to academic institutions — IIT Kharagpur, IIT Delhi, IIT Hyderabad, NIT Patna — filing in the dozens.

And the subject matter is exactly what you'd expect from that filer list: video coding (813 patents), 5G NR (1,191), sidelink (943), beamforming (887). These are standards-essential territory. They are the toll booths on any 5G or 6G network India builds, and they are overwhelmingly owned abroad.

The domestic sector is smaller than one Series B

Here is the number that reframes everything above. Across 386 tracked communications startups, total capital raised is $75.8 million. Median raise: $3.9 million. Series B or later rounds: zero.

For scale, a single well-funded startup in health or robotics has raised more than India's entire tracked communications startup sector combined.

There are real companies in it. Astrome Technologies ($14.7M) builds flat-panel antennas and E-band radios — a direct match to the antenna line running 69.3% Chinese. Pantherun ($19.6M) does FPGA-based encryption for critical infrastructure. RtBrick builds networking software for telecom and data centres. Velmenni works on Li-Fi optical wireless.

But four companies with meaningful funding do not substitute an $8 billion component import bill, and nobody involved thinks they do.

What we'd actually conclude

Three separate things are true at once, and they are usually discussed as if only the first exists:

• India successfully excluded Chinese vendors from finished network equipment. The data confirms it.
• India did not reduce dependence on Chinese components, which are more than half the parts bill and probably more than the official figures show.
• The intellectual property underlying whatever India builds instead is foreign-owned by a wide margin, and the domestic startup base is capitalised at a level that cannot change that this decade.

The first is a policy win. The second and third are the parts of the problem that a procurement rule was never capable of solving, and that no amount of assembly localisation reaches either.

If there is an investable opening here, it is narrower and less glamorous than "India's 5G opportunity": antennas, RF front-ends, and populated boards — physical components, currently majority-Chinese, where one or two domestic companies already exist and are visibly under-capitalised. That is a smaller claim than the sector's press coverage implies. It's also the one the data actually supports.

Methodology

Trade figures are FY2024-25 from India's Trade Intelligence & Analytics Portal at HS8 level. The finished-versus-parts split groups HS 851711/13/14/18 (handsets) and 851761/62 (network apparatus) as finished equipment, against 851771 (aerials) and 851779 (other parts) as components — a standard reading of the heading structure, but a judgment call we're stating openly rather than burying.

Trend and cross-check figures come from CEPII BACI (2017-2024, India as importer, HS6, mirror-reconciled). The two sources are methodologically different — calendar versus fiscal year, HS6 versus HS8, mirror-reconciled versus as-reported — so we use TIA for levels and BACI only for direction and cross-validation. The divergence discussed above is presented as an observation with a plausible explanation, not a proven finding.

Patent, startup and funding figures are Deeptech Navigator's registry as of 2026-07-22. Patent coverage is centred on 2022-onward filings. "Foreign filer" reflects the filing entity's country as recorded, which does not always track ultimate ownership.

In our data

Sources

• China supplies over 30% of industrial goods; overdependence critical: GTRI — Business Standard ↗
• Booming AI Chip Trade Seals Hong Kong's Role as Gateway to China — Bloomberg ↗