Behind the defence-PSU rally, a ₹6 lakh crore procurement plan was hiding in plain sight

320 percent. Eighteen months. Fourteen names. Every analyst who filed it under "geopolitics premium" collected the right return and built the wrong mental model for what comes next.

The move was the price catching up, not a war bid. HAL's order book stood at ₹1.25 lakh crore at end-FY25. BEL's was ₹74,000 crore. MDL sat at ₹38,000 crore; BDL at ₹19,000 crore. All of it was disclosed in exchange filings, available to anyone who read them. The market simply took its customary 2-3 quarters longer than it should have to price what was already on the page.

That is the thesis. The rest is evidence.

Fourteen Names, One Number

The cohort: HAL, BEL, BDL, BEML, Mazagon Dock Shipbuilders (MDL), Garden Reach Shipbuilders (GRSE), Cochin Shipyard (CSL). Then the supplier tier: Astra Microwave, Apollo Micro Systems, Solar Industries, Premier Explosives, Data Patterns, MTAR Technologies, Paras Defence.

Cohort-weighted, the 18-month total return through early 2026 is approximately 320 percent. That is not a single-stock blowout driven by one unexpected contract. It is a coordinated re-rating across the entire value chain, from the prime contractor to the last-mile components maker. A war-premium bid does not move the ancillary supplier tier in lockstep with the primes; a procurement-cycle re-rating does.

The spread within the cohort is not uniform, and that dispersion matters more now than it did a year ago. The 320 percent figure is a cohort-weighted composite. It confirms that the re-rating was systematic. It says nothing about which individual name was priced correctly at the end of the move, and which was carried along on the sentiment tide.

The Frame Everyone Got Wrong

War-proximity bids are real. Every defence index in every country spikes on conflict headlines. They also fade within a quarter when no sustained procurement shift materialises. A sentiment premium has a half-life. A capex cycle does not.

The correct frame is the FY24-FY29 defence procurement plan, sized at roughly ₹6 lakh crore in the Ministry of Defence planning document. That is not a headlines-driven announcement. It is a five-year budget commitment with line items, indigenisation mandates, and a procurement pipeline that takes years to assemble and years more to unwind. You cannot reverse a ₹6 lakh crore capex plan with a ceasefire.

The lazy framing also fails on the timeline. The order books that anchored this rally were filed with exchanges in FY25. If the rally were purely a geopolitics bid, you would expect the price action to have preceded the disclosure, not trailed it by two to three quarters. The sequence is exactly backwards from the war-premium theory.

Sentiment can lift a stock for a week. A ₹6 lakh crore procurement plan lifts an order book for five years.

The Receipt Was Already Filed

The cynical version of this story says the move was unknowable until it happened. The filings say otherwise.

HAL disclosed ₹1.25 lakh crore in order book at end-FY25. BEL disclosed ₹74,000 crore. MDL disclosed ₹38,000 crore. BDL disclosed ₹19,000 crore. Combined, those four prime contractors alone reported aggregate order books exceeding ₹2.56 lakh crore (this desk's arithmetic from disclosed figures; each number is company-reported). What was missing was not the data. What was missing was the valuation framework to treat a public-sector order book with the same seriousness as a private-sector backlog.

That framework gap has now closed, and it closed fast.

The broader point is methodological. Defence-PSU order books are disclosed quarterly, the same as any other listed company. The information was not hidden. The reluctance was analytical: a history of programme delays, slow revenue conversion, and government payment-cycle uncertainty had trained the buy side to discount reported backlogs heavily. That discount narrowed as conversion evidence accumulated.

The Lag Is Closing, Not Opening

Defence-PSU price action has historically trailed order-book disclosure by two to three quarters. The mechanism is not mysterious. Order books at these companies build slowly, convert to revenue over multi-year delivery schedules, and sit at organisations whose project cycles are measured in years, not quarters. Buy-side revenue models take time to catch up because the timing of recognition is genuinely uncertain until delivery milestones are confirmed.

What happened over the past 18 months is that this lag closed faster than in prior cycles. The likely catalyst was the density of disclosures: HAL, BEL, MDL, and BDL all reported large backlog figures within the same reporting season, and the aggregate signal was harder to dismiss than any single company's announcement. When four prime contractors simultaneously post multi-year visibility, the discount rate on "scepticism about conversion" starts to look like an error rather than a precaution.

The question now is whether the lag closing means the trade is exhausted. That depends entirely on two variables: whether new orders continue to arrive at a pace that sustains the backlog, and whether the existing backlog converts to revenue on the schedule that current multiples have priced. The answer to both is in the FY26 quarterly cadence. Not in the next news cycle.

The Floor Nobody Wants to Model

75 percent. That is the indigenous content target embedded in MoD procurement policy for defence systems by FY27. It is the structural floor under the entire cohort, and most valuation models have not fully priced it.

The practical implication: any MoD-tendered platform with a foreign OEM in the prime role must now source 75 percent of its content domestically. The prime contractors in this cohort are among the few organisations currently certified, scaled, and capable of meeting that threshold. A private-sector new entrant does not cross 75 percent indigenisation overnight. That is years of supply-chain qualification, manufacturing buildout, and type-certification.

The iDEX scheme, launched in 2018, has been the other structural driver. It channelled private-defence demand into the listed PSU supply chain by creating a regulated procurement on-ramp for smaller innovators to supply into the indigenisation pipeline. The ancillary cohort (Astra Microwave, Data Patterns, Solar Industries, Premier Explosives and others) has built order books in part because iDEX created structured demand they could not previously access.

These policy foundations do not evaporate when headlines fade. That is the floor.

The Export Layer: Under-Modelled, Not Unreal

Exports are the layer the sell side consistently under-models for this cohort. The historical reason is credibility: India's defence exports were negligible a decade ago, and the instinct was to assign near-zero probability to any scale-up.

The programmes now referenced are not aspirational. BrahMos missiles have been contracted to the Philippines. Pinaka multi-barrel rocket systems are headed to Armenia. ATAGS howitzers have been delivered to Bhutan. Tejas conversations are reportedly in various stages with Latin American buyers.

The desk's standard: none of these export programmes should be modelled as recognised revenue until orders are formally signed and delivery milestones confirmed. That standard is stated here explicitly. But the directional point holds: India's defence production base is now export-capable in categories that matter (cruise missiles, artillery systems, light combat aircraft), and a valuation model that assigns zero probability to export scale-up over a five-year horizon is leaving genuine optionality off the table.

This is not a claim that India is now a top-tier arms exporter. It is a narrower claim: a cohort priced as a domestic-only supplier should carry some non-trivial export probability in a reasonable forward model, and current sell-side consensus largely does not.

Conversion Is Where the Cohort Splits

A large order book is a necessary condition for re-rating. It is not sufficient to justify identical multiples across fourteen names.

BEL and HAL have shown efficient conversion in recent reporting cycles: order-to-revenue timelines that are reasonably predictable, margins that have been stable, and manufacturing capacity that scales within the existing footprint. Their backlogs can be modelled with some confidence as deferred revenue with known delivery cadence.

BEML is different. Its defence mobility segment converts lumpily, with large equipment orders followed by extended manufacturing and acceptance cycles. The metro and mining segments introduce additional demand variability. Investors who apply a straight-line re-rating to BEML on the back of the defence capex story will find the quarterly result sequence frustrating.

GRSE and CSL occupy a distinct position. Naval platforms (frigates, offshore patrol vessels, aircraft carriers) carry the longest build cycles in the cohort. Their order books offer duration; they do not offer near-term conversion speed. For anyone tracking FY26 quarterly disclosures as the verification mechanism, the shipyard names will be the slowest to provide revenue confirmation.

No call on which name is preferred here. That is not this desk's mandate today. The dispersion is the story.

The Risk Register, Written Straight

Three named risks. None is remote.

Budget rationalisation. India's fiscal arithmetic has accommodated successive rounds of defence spending, but a serious deficit-control cycle will eventually force choices about which capex commitments accelerate and which are deferred. The ₹6 lakh crore FY24-FY29 plan is a planning document, not an appropriation. Each year's Budget translates it into actual expenditure. A meaningful deviation from the five-year trajectory is the single largest systemic risk to the entire cohort's order-book assumptions.

Programme delay. AMCA, the Advanced Medium Combat Aircraft, is the current live example. Technical and industrial complexity has pushed its timelines before, and the risk of further slippage is real. HAL carries AMCA as a long-run pipeline asset; the concern is that continued delay compresses long-term order-book expectations without an offsetting acceleration in near-term contracts. AMCA is not in the current-cycle revenue bridge, but it is embedded in the multiple that extrapolates HAL's trajectory beyond FY29.

Private-public JV margin pressure. The indigenisation push is simultaneously an invitation to private-sector players with foreign OEM partnerships to compete for contracts that PSUs have historically held. JV structures can compress PSU content-share on a given programme and, with it, margins. The PSU incumbency advantage is real: scale, government relationships, and existing type-certifications are not easily replicated. But the advantage is not permanent, and the margin compression from a well-structured private JV winning a significant programme would be material.

None of these risks has a precise timeline. All warrant a line in every forward model.

What FY26 Actually Tests

Forget the next news headline. The FY26 quarterly cadence is the real stress test for this cohort.

The question is operationally simple: do the order books at HAL, BEL, MDL, and BDL begin converting to revenue at the pace that current market multiples have priced in? Order-book-to-revenue conversion ratio, tracked across Q1 and Q2 FY26 results, is the metric that matters now.

If conversion tracks, the re-rating is validated and the multiple holds. If revenue recognition slips (programme delays, supply-chain bottlenecks, government payment cycles), the valuation buffer gets tested quickly. A large order book is a ceiling on the upside and a narrative floor. It is not a floor under the stock price if the conversion timeline stretches.

This is also where the cohort split becomes visible in the data rather than in the theory. The names that convert efficiently will separate cleanly from those that do not. The market will re-rate within the cohort, not just across it.

The next real inflection for this cohort is not a contract announcement. It is Q1 FY26 results season, and the order-book-to-revenue ratio it prints.

That is the number to watch.