Rs 200 Crore in RDI Fund Debentures to Agnikul: The Insurance Covenants Deeptech Borrowers Should Expect

On 25 September 2026, DD India reported that the Technology Development Board under the Department of Science and Technology had signed an agreement giving Rs 200 crore to Chennai-based Agnikul Cosmos under the Research Development and Innovation (RDI) Fund. The money is meant to take the reusable Agnibaan launch vehicle from TRL-4+ to TRL-8, which is the stretch between a validated component in a lab setting and a flight-qualified system.

The instrument matters more than the headline number. The funding is structured as Optionally Convertible Debentures (OCDs), not as a grant. A grant is spent and closed out against milestones. A debenture is a debt claim: it sits on the balance sheet, it carries terms the issuer has to keep, and it converts to equity only if the option is exercised. Until then, the holder behaves like a lender.

TDB Secretary Rajesh Kumar Pathak framed the rationale this way:

Reusability is an important technology frontier in this regard.

For a deeptech founder reading this as a template for their own RDI Fund application, the useful question is not whether reusability is strategic. It is what a debt-style holder will want to see protecting the asset it is financing, and what that implies for the company's insurance programme before the first tranche is drawn.

We do not know the specific covenants in the Agnikul agreement, and nothing published so far sets them out. What follows describes what lenders and debenture holders of this kind typically require, so founders can prepare for it. Treat each item as a likely request to negotiate, not a confirmed RDI Fund term.

A grant funder's main risk is that the money is wasted. A debt holder's main risk is that the asset backing the repayment or conversion value is destroyed and nobody pays for it. That is why debt documents in Indian project and venture lending usually contain an insurance clause with four recurring parts:

1. Maintenance of insurance: the borrower must keep named covers in force, with insurers of a stated standing, for amounts at least equal to replacement value or a fixed figure.
2. Loss payee or assignment: the holder is named on property and project policies so that claim money for the financed asset flows to it, or with its consent.
3. Reporting: copies of policies, renewal confirmations and a certificate of insurance within a set number of days, plus notice of any material claim or cancellation.
4. Events of default: lapse or material reduction of cover becomes a default, sometimes with a short cure period.

For a vehicle moving from TRL-4+ to TRL-8, the asset under finance changes shape every quarter: engines on test stands, integrated stages, ground support equipment, then a flight article. The insurance schedule has to follow that progression, or the covenant will be breached by the programme's own success.

On 12 January 2026, the PSLV-C62 mission failed. Deccan Herald reported a third-stage anomaly roughly 203 seconds after lift-off. WION reported that the failure destroyed six satellites belonging to Indian startups, and that there was no confirmation whether any of them were insured.

That uncertainty is the point. Six payloads from private companies were lost on a government launcher, and the reporting that followed could not say whether the owners would recover anything. For an equity-funded startup, an uninsured loss is a hit to runway. For a startup carrying debentures, the same loss can also be a covenant problem: the financed asset is gone, the holder was expecting a loss payee recovery, and the company may have to explain why there is none.

What it means for a launch developer rather than a payload owner

Agnikul and companies like it sit on the other side of the PSLV-C62 story. A launch developer faces three distinct exposures from a failed flight:

• Loss of its own vehicle and any reusable hardware it planned to recover.
• Liability to third parties on the ground or in the air, which the state requires the operator to insure under its authorisation.
• Contractual exposure to customers whose payloads were on board, governed by the launch services agreement and any cross-waivers in it.

A debt holder will care most about the first, because it is the asset the money built. The regulator cares most about the second. The customers care about the third. A deeptech borrower needs an answer for all three before the first integrated flight test.

The financed period for Agnibaan, TRL-4+ to TRL-8, is mostly ground work. That is where the insurance programme should start, not at the launch pad.

Ground test and integration

Static fires, stage tests and integration activity carry fire, explosion and machinery risks at the test facility. Indian insurers usually approach this through engineering lines rather than space-specific forms. Founders should ask brokers about:

• Erection all risks or contractor-style cover for integration of stages and ground support equipment, written on the build value as it rises.
• Machinery breakdown and electronic equipment cover for test stands, data acquisition systems and propellant handling plant.
• A fire and allied perils policy on the facility and stock, with explosion clearly within the insured perils given the propellants involved.

The engineering insurance market will want test procedures, abort criteria and the distance between test cells and other property. A debt holder will want the sum insured to track replacement value, which means agreeing a declaration mechanism so the figure can be raised as hardware is completed.

Pre-launch and launch

Pre-launch cover (transport to the range, integration at the pad) and launch cover (from ignition or intentional ignition through a defined point) are specialist placements, usually led from international space markets with Indian fronting. Capacity is thin and priced per mission. For a reusable vehicle, the wording must say clearly whether recovery operations after stage separation are inside the launch period, and how the recovered stage is valued if it is damaged on return. That question has no settled market answer yet, so it should be raised early with the broker, not at binding.

Ask the broker to show the debt holder the launch cover structure before the covenant is signed. If the market will not write launch cover at replacement value for a first flight, the covenant should say so, rather than requiring a sum insured the company cannot buy.

Third-party liability is the one cover a launch operator cannot treat as optional. Private launches in India require authorisation through IN-SPACe, and the insurance obligation attaches to the operator through that authorisation. Our earlier analysis of third-party liability cover for launch vehicles sets out how the required limit is sized against the mission profile rather than the vehicle's cost.

For a borrower, the interaction with debt terms is worth thinking through. A lender will typically want evidence that every regulatory condition on the borrower's licences is met, because losing an authorisation is a direct threat to the business. So a liability policy that does not satisfy the IN-SPACe condition is not only a regulatory problem. Under a typical compliance-with-law covenant, it can also become a lender problem.

Points to settle with the broker:

• The limit required for each mission and whether the policy is per-occurrence or aggregate. An aggregate limit eroded by an early claim can leave later test flights uninsured.
• Whether the government and range operator need to be named as additional insureds.
• The relationship between the policy and any indemnity the operator gives the state in the authorisation, because the policy will not automatically respond to every contractual promise.

As the space launch subsidy analysis noted, government money that reduces the cost of launching does not reduce the liability the operator carries. The same applies to RDI Fund money: it finances development, and the third-party risk stays with the company.

Beyond the space-specific lines, a debt holder will review the ordinary corporate programme, and for a company of Agnikul's profile the ordinary programme is not small.

Property and equipment. Manufacturing facilities, additive-manufacturing machines, clean rooms and stock of long-lead components are the tangible assets the debenture money is building. Expect a request for property insurance at reinstatement value, with the holder noted as loss payee. Check the average clause: underinsured stock on a fast-growing build programme is the usual reason a property claim pays less than expected.

Key-person cover. Deeptech value is concentrated in a small number of engineers and founders. Debt holders often ask for key-person life or disability cover on named individuals, assigned to the holder or with the holder as beneficiary up to the outstanding amount. It is cheap relative to the exposure, and founders should agree the named list and the assignment mechanics early so it does not hold up a drawdown.

D&O. Directors and officers liability cover becomes more relevant once a public-sector holder sits in the capital structure. If the debentures convert, a government-linked shareholder will expect board governance and disclosure to match. Even before conversion, information covenants and milestone reporting create a stream of representations that directors are personally signing. Founders should check whether the D&O policy covers claims brought by a debenture holder or major creditor, since some wordings exclude claims by significant shareholders and that exclusion may apply after conversion.

The clauses that cause friction in practice are rarely the list of required covers. They are the mechanics. Points a deeptech borrower should negotiate before signing, based on how these clauses usually work:

1. Scope of loss payee: limit loss payee status to the specific assets financed, not every policy the company holds. A blanket clause can send liability or key-person proceeds to the holder when the company needs them for operations.
2. Threshold for direct payment: agree a claim size below which proceeds go straight to the company for repair, so a damaged test stand does not wait for holder consent.
3. Reinstatement option: give the company the right to use proceeds to rebuild the financed asset, which matches what both parties want on a development programme.
4. Reporting calendar: align insurance reporting with the existing milestone reporting, so renewal confirmations arrive in the same pack and do not become a separate compliance task.
5. Cure period for lapse: ask for a reasonable cure window, because per-campaign covers and specialist placements sometimes bind a few days late through no fault of the company.

Every change to the financed asset base, such as a new test site, a new engine variant, or a stage moved to a range, should trigger a check of whether the property and engineering schedules still match the covenant. A simple quarterly reconciliation between the asset register, the policy schedules and the covenant schedule prevents most surprises.

The Agnikul agreement suggests the RDI Fund will use debt-like instruments for deeptech. If that holds, other companies applying for these instruments should expect insurance diligence to be part of the process. A sensible preparation sequence:

1. Map the asset base across the funded TRL range: what exists now, what will be built, and when each item becomes valuable enough to insure separately.
2. Get a broker to quote ground test, engineering and property covers on the current base and the projected base at each milestone, with a declaration mechanism for rising values.
3. Establish what the market will and will not write for launch and recovery of a reusable vehicle, in writing, before the covenant fixes a requirement.
4. Confirm the IN-SPACe third-party liability requirement per mission and how the policy will meet it across a series of test flights.
5. Put key-person and D&O cover in place with wordings reviewed for creditor and shareholder claims.
6. Draft the loss-payee, reinstatement and reporting mechanics yourself and offer them in negotiation, rather than accepting a generic lender template.

The PSLV-C62 loss in January 2026 showed what happens when nobody can say afterwards whether a destroyed asset was insured. A company holding public debentures cannot afford that ambiguity. The space launch startup insurance guide covers the mission-level placements in more detail; the covenant work described here is the corporate layer that has to sit around them.