What a hospital bed actually costs to build, according to the people building them

In its June 2026 quarter call, Medanta described revising its Guwahati project. The original plan, as described on the call, was about ₹500 crore for 400 beds. The revised plan is ₹970 crore for 650 beds. Capex per bed went from roughly ₹1.25 crore to roughly ₹1.49 crore, and management was explicit about why: they doubled the operating theatres and cath labs.

That is the most useful thing in this article. Capex per bed is a function of what you intend to treat, far more than of where you build.

Two things worth saying about the numbers you will read elsewhere

First, no rating agency publishes a capex-per-bed figure. It is tempting to divide ICRA's ₹30,000 to ₹32,000 crore of sector investment by its bed-addition forecast, and the answer looks plausible. It is wrong: the investment figure is a four to five year cumulative and the bed additions are two years. We checked, and the arithmetic does not hold. Where a per-bed number appears in this article, a company said it.

Second, buying is not building. CRISIL put inorganic acquisitions across FY24 to FY26 at roughly ₹11,000 crore for about 4,300 beds, at valuation premiums of around 2.2 times. Manipal's August 2026 agreement for the 100-bed Kinder Women's Hospital in Bengaluru at ₹130 crore is a cleaner single data point at about ₹1.3 crore per bed.

What those beds then earn

Revenue per occupied bed per day, from the June 2026 quarter: Max ₹81,900, Fortis about ₹74,200, Medanta ₹70,244, Narayana about ₹52,600, Yatharth ₹34,758. Rainbow separates mature from new units and reports roughly ₹70,000 against ₹59,000, an 18 per cent gap that is worth budgeting for.

Be careful comparing those with the sector figure. CRISIL projects around ₹52,200 across 98 private hospitals including smaller and standalone units. It is not measuring the same universe as the listed chains, and neither number is wrong.

Sector occupancy sat at 63.5 per cent in FY2026 on ICRA's sample, with operating margins around 24 per cent. Hospital-only margins at the listed chains cluster between 21 and 26 per cent once you strip out pharmacy and diagnostics.

How fast it turns

This has genuinely changed. CRISIL now puts breakeven for a new facility at 12 to 18 months, improved from three to four years. The disclosed cases are faster still: Yatharth's 400-bed Faridabad Sector 20 unit reached EBITDA breakeven in nine months, and was running at 49 per cent occupancy; Aster reported nine months at Kasaragod; KIMS reported under seven months at Mahadevapura.

One caveat matters and nobody states it clearly. Every one of those is EBITDA breakeven. Not one chain published a cash or PAT breakeven timeline for a greenfield unit this quarter. If your funding model needs a cash-breakeven month, that number does not exist in public disclosure and you will have to build it yourself.

What this means for your hospital

Decide the case mix before you size the building. The Guwahati revision is what happens when the clinical ambition is settled after the civil plan, and it cost 94 per cent more capex for 63 per cent more beds.

Model the ramp explicitly. A new unit runs a lower realisation and a lower margin for its first years, and the disclosed drag is between two and five percentage points of EBITDA margin at chains with far better negotiating positions than a single-site hospital has.

And be honest about which benchmark you are using. ₹0.6 crore per bed and ₹2.1 crore per bed are both real numbers from the same quarter. The difference between them is a decision you are about to make, not a market you happen to be in.