The Facility Was Right. The Product Was Wrong.
Thirty kitchens, six months empty, and what it cost to find out what we were selling.
The first real decision was a building in Gurugram. Ten thousand square feet, thirty kitchens, the largest single site we could find and sign. It was the correct bet on every dimension I knew how to measure — location, cost per square foot, proximity to the delivery radius that mattered. We signed it. We built it. We turned it on.
Then we sat with thirty empty kitchens for six months, during the most bullish period the category has ever had.
What we thought we had built
Delivery infrastructure.
The pitch was speed. An enterprise F&B brand with an existing menu and existing demand could take a kitchen, plug into our systems, and be selling in under thirty days. No site search, no build, no equipment procurement. We had removed every reason a brand takes eighteen months to enter a new market.
It was a good product. Brands liked it. Almost none of them signed.
The trade nobody names
Here is the thing about a cloud kitchen that does not appear in the sector’s own account of itself.
You remove the room. No frontage, no dining floor, no front of house. Capex drops by an order of magnitude and rent stops being a fixed cost you spend three years praying to outgrow.
Then you hand the customer relationship to an aggregator, and the aggregator takes twenty to twenty-five per cent of gross. We partnered our way down to sixteen. In a dine-in P&L, sixteen per cent off the top is not a line item. It is the difference between a business and a hobby.
That is the entire model. You are trading fixed cost for variable cost. Capex down, commission up. Nobody selling you a cloud kitchen puts it that way, because structurally superior sells better than different trade. But it is a trade, and a trade has terms.
The terms are volume. A lower breakeven beats a lower margin only above a certain throughput. Below it the arithmetic inverts, and the room you removed starts to look like the thing that was quietly subsidising you.
In 2020, volume was not the constraint.
So why didn’t it fill
Everything the sector said about that period was true.
Delivery velocity rose exponentially. Ordering food to your door stopped being a preference and became the only available option, which is the only mechanism by which consumer habit reliably changes — not because a product got better, but because for a period the alternative was removed. When restaurants reopened, the habit had already formed. It did not reverse.
All of that was happening. None of it filled the facility.
It reduces to one sentence. A market moving is not a business winning. The tailwind moved the demand curve. It never touched the sales cycle. Every enterprise brand still had a procurement process, a capex committee, an incumbent way of doing things, and somewhere else they could have gone. Exponential category growth does not shorten a single one of those by a day.
Anyone who built during that period and calls the result validation is reasoning backwards from the fact that they survived.
The pivot
It was a burger brand, and they did not want a kitchen.
They walked in asking whether we could give them a commissary. Central production. One facility prepping for outlets across a city, rather than one kitchen selling into a delivery radius.
I want to be honest about this: it was not my insight. It was a customer describing a problem I was not selling into, and me being slow enough to listen.
Their alternative was to build it themselves. Fifty lakhs to a crore, four to six months, and a lease on a facility they would own for a decade. What we could offer instead was a rental deposit, some custom work on the site from our side, and a move-in inside month one.
Same square footage. Same infrastructure. Entirely different buyer.
A cloud kitchen is a distribution asset, and it is sold to whoever owns the delivery P&L — often a growth or marketing budget, structurally suspicious of fixed commitments. A commissary is a production asset. It goes to the people who plan capacity, who already had build central kitchen somewhere on a five-year plan, and who would very much prefer not to.
We had built the second thing and spent six months selling it as the first.
The hub nobody would fund
By then, hub-and-spoke was the consensus format in Indian F&B. Everyone had a slide. Central production, distributed outlets, better margins, tighter control.
The spokes were easy — small footprints, low capex, quick to open. The hub was the part nobody wanted to pay for. It was the crore and the six months and the decade-long lease sitting between a brand and a strategy it had already agreed to.
We rented it to them.
That was the business. Not delivery. Not kitchens. We had converted the most capital-intensive component of the format everyone claimed to want into an operating expense, and it took an empty building and a burger brand to explain that to me.
What the numbers conceal
We went on to build a hundred and twenty-two kitchens across three cities in under twenty-four months, and reached ninety per cent occupancy inside twelve, across more than a hundred brands.
Those numbers are true, and they are the wrong ones to learn from.
The number that taught me something was six months of empty kitchens in Gurugram — a facility that was correct in every respect I had tested. I had run the substantive work on unit economics. Cost per square foot, breakeven throughput, commission structure, payback period. All of it held.
None of it examined what the buyer thought they were purchasing.
You can test a business exhaustively and still be testing the wrong thing. That is not a failure of rigour. It is what rigour cannot reach, and the only thing that reaches it is a customer telling you, plainly, that you have misunderstood your own asset.