VERDICT: TRACK (re-rated 2026-08-17 under the recalibrated rules; originally KILL)
Why:
- A sticky, capital-light business gaining share in a barely-touched market. Customers stay and spend 27% more each year; none of the ten biggest has left in three and a half years. Even deleting the largest customer entirely, growth was 24-25% against a market growing under 20%. The company holds about 3% of its serviceable market, and 90% of broadcast operations were still on old on-premise gear at the filing date.
- The money math is arriving. Losses fell from ₹321 crore to breakeven in three years. Each extra rupee of revenue drops 24-53 paise to profit, and the cost structure points to high-teens operating margins at maturity, one to two years out at the demonstrated pace, with no debt and almost no capital needed to grow.
- What a deep dive must answer: growth is narrowing (one customer is 14% of revenue and a third of growth; new customers add under 2%), the main division's growth halved in a year, and the pre-IPO profit leaned on items that won't repeat. Good business, decelerating trajectory; the deep dive decides which wins.
What the company does
Amagi's software lets TV and streaming companies run channels from the cloud instead of from broadcast facilities: create the channel, schedule it, deliver it to platforms like Samsung TV Plus and Roku, and insert ads into it. Customers sign three-year contracts that renew automatically. Almost three-quarters of revenue comes from the US. The work is done mostly from India, which is why gross margins are near 70%.
The model works like this: land a customer, then grow inside them. Existing customers spend 27% more each year, and none of the ten biggest has left in three and a half years. That stickiness is the real strength of this business.
The catch: the whole platform runs on Amazon's cloud. Amagi has committed to pay AWS ₹2,418 crore over six years, and two-thirds of the IPO's fresh money is earmarked to fund that bill. The money buys no asset. It pays a supplier.
Key numbers
| ₹ cr | FY23 | FY24 | FY25 | H1 FY26 |
|---|---|---|---|---|
| Revenue | 681 | 879 | 1,163 | 705 |
| Revenue growth % | n/a | 29 | 32 | 35 |
| Gross margin % | 65 | 69 | 69 | 70 |
| Profit after tax | (321) | (245) | (69) | 6.5 |
| Cash from operations | (245) | (183) | 34 | (201) |
| Division, ₹ cr | FY23 | FY24 | FY25 | H1 FY26 |
|---|---|---|---|---|
| Streaming Unification | 346 | 463 | 664 | 373 |
| Streaming growth % | n/a | 34 | 44 | 23 |
| Monetization and Marketplace | 197 | 203 | 281 | 178 |
| Monetization growth % | n/a | 3 | 38 | 53 |
| Cloud Modernization | 138 | 213 | 217 | 154 |
| Cloud growth % | n/a | 54 | 2 | 47 |
The division split is management's own cut; the audited accounts report a single segment. Half-year growth is measured against the same half a year earlier.
Growth is real and losses have shrunk fast. But note the last column: the "profitable" half-year actually consumed ₹201 crore of cash, and the company never explains why it cut sales spending 11% in the year before listing, which is where most of the margin improvement came from.
Scorecard
| Block | Rating | Why |
|---|---|---|
| Right to win | STANDOUT | Customers stay and spend more every year, which is proof the product matters. Even if you remove the largest customer entirely, Amagi still grew about 25% in a market growing under 20%, so it is gaining share. The doubts: the company's favourite claims (network effects, data advantage) are not supported by its own numbers, the filing names no competitor data at all, and Amazon sells a basic version of the same thing at commodity prices. |
| Industry and TAM | STANDOUT | The market is large, growing about 13% a year, and Amagi holds only about 3% of it; roughly 90% of broadcast operations were still on old on-premise gear at the filing date, and independent data showed the free-TV market still compounding. The company's paid market study has sizing errors and an overstated headline, but our own checks answer the room-to-grow question favourably, so those flaws are notes about the document, not the market. |
| Financial momentum | PASS | Revenue compounds above 30% at a 70% gross margin, the operating margin has improved three half-years running, and each extra rupee of revenue drops 24-53 paise to profit; on this cost structure the business lands at high-teens margins within a year or two, with no debt and almost no capital needed. Two open questions temper it: whether the sales-spending cut that drove most of the improvement holds, and whether the biggest customer keeps spending. |
| Risks, governance, RPTs | PASS | Clean where it matters: no fraud, no debt, no promoter loans, no group companies, tiny related-party numbers. Two blemishes worth knowing: in 2022 the company spent ₹113 crore buying back shares from two founders and one founder's wife, and the auditors have noted weak accounting-system controls three years running. Neither is disqualifying. |
| Promoter and cap table | PASS | The register is clean at listing: every convertible converted, nothing pledged, all subsidiaries wholly owned, founders selling nothing. Funding rounds priced steadily upward and the highest-priced investor is holding while the cheapest exits. The founders' pre-IPO share top-up was a milestone reward the investors agreed to in 2022 and paid for from their own shares, which is ordinary for a venture-backed company; the soft spot is that no outside money has come in since December 2022 and the one trade since was a third below the last round. |
| Offer structure | MARGINAL | No promoter sells. There is no debt. The main use of money is honest about what it is: paying the cloud bill, matched to a real contract and real spending history. The flags: two-thirds of the fresh money builds nothing, about a third has no named use, and the company's acquisition record is poor (one purchase written off within five months, another being wound up). |
Watch out for
- Ninety busy days before the IPO. Between April and July 2025 the company: doubled its AWS commitment, cashed out and restructured employee stock schemes (ending a big accounting charge), settled the founders' share entitlement at ₹25 a share against an internal valuation of ₹591, and got board approval to file. Each step is defensible; together they show a filing shaped for the window.
- The big customer. 14% of revenue, likely a US TV maker recently bought by a large retailer. Amagi has lost a big customer to "we'll build it ourselves" once before, in 2022.
- The clock. If the IPO doesn't complete by 31 July 2026, all the investors' special rights come back automatically.
- AWS is supplier, cost floor, and competitor at once.
- Two large investors keep board seats after listing through an arrangement public shareholders will be asked to approve.
Backtest artifact. Assessed from the filing only; verdict frozen before any post-IPO information was read. Valuation is deliberately not part of the verdict. Not a recommendation.