Long-Dated Capex Is Being Financed on Shortening Money

A capital expenditure cycle and a credit cycle are usually discussed as one story, on the assumption that strong headline credit growth means the capex is funded. They are two stories, and in India through 2026 they are running on different clocks.

The FY2026-27 Union Budget raised capital expenditure to ₹12.22 lakh crore, with effective capex at ₹17.15 lakh crore - roughly 10% above the previous year’s ₹11.1 lakh crore, and weighted toward highways, corridors and high-speed rail. What follows concerns the tenor of the money financing that build-out, using credit data to 31 July 2026.

The Spending Is Long-Dated by Construction

Economic corridors and rail links are not working-capital assets. They draw funding across multiple years before producing a cash flow, and the private capital that co-finances them - contractors, developers, equipment suppliers, the non-bank lenders behind them - inherits that horizon whether or not its own liabilities match it.

That is the ordinary condition of infrastructure finance and not in itself a problem. It becomes one only when the liability side shortens while the asset side does not.

The Funding Base Is Shortening

System deposits grew 15.4% year-on-year to 31 July 2026 against advances at 19.1%, putting the banking system’s loan-to-deposit ratio at 80.3%. A system funding a widening share of its book from certificates of deposit and market borrowings is funding long assets with instruments that reprice and roll far more often than the assets do.

Non-banks sit at the sharp end of this. An RBI working paper found policy transmission to NBFCs incomplete: one percentage point on the repo moves NBFC weighted average borrowing rates by 0.24 points over three quarters. In an easing cycle that means the relief mostly does not arrive, which is why NBFCs have seen only marginal relief in borrowing costs despite the cuts already delivered.

Banks Are Reclaiming Share Because They Transmit Faster

The competitive consequence follows directly. Banks passing rate cuts through more quickly than non-banks can take share on price, and India Ratings has flagged NBFCs as a drag on system credit growth even while the headline number stays healthy.

This is where the aggregate misleads. Healthy system credit growth composed of banks gaining share from non-banks is not the same signal as healthy growth across both. The borrowers who cannot follow the share shift - smaller developers, thinner-balance-sheet contractors, the tail of the supply chain - do not experience the easing at all. They experience a lender that has become more expensive and more selective.

Mismatches Surface at Refinancing, Not at Origination

The important property of a maturity mismatch is that it is invisible while funding rolls. Nothing breaks at origination. Nothing breaks while the wholesale market is open. The gap shows up at the refinancing date - as a wider spread, a shorter tenor offered, a covenant, or no bid at all - and by then the asset is part-built and the borrower has no alternative.

This is a claim about sequence, not about timing. It says where stress appears and in what order, not when. A funding base that shortens while the asset base lengthens produces refinancing pressure, then balance-sheet stress, then consolidation, in that order, and the interval between them is set by how long the wholesale market stays accommodating.

What This Means for Allocators

Underwrite the liability tenor, not the headline credit growth. For any lender financing multi-year assets, the question is how much of the book rolls inside twelve months and at what spread.

Read system credit growth compositionally. Growth driven by share transfer between lender types conceals which borrowers have lost access, and those are the borrowers where the mismatch actually sits.

And date the exposure rather than rating it. The capex is real. The funding is shorter. The gap between them comes due on a schedule that is already knowable.