Case 056Comps and relative valueHard
Compare an AI model developer burning cash to grow 150% a year with a mature, highly profitable enterprise software company. Which multiple would you use for each, what drives value, and where would you look for growth?
1The situation
Nuvexa Labs builds and sells access to large AI models. Revenue is Rs 2,000 crore, growing 150% a year. Gross margin is 45%. Compute is its biggest cost at 60% of revenue: 40% is inference, the chips that serve customer requests, which sits in cost of revenue, and 20% is training new models, which sits in operating costs. Salaries and other costs take another 105% of revenue, so EBITDA is minus 80% of revenue, a loss of Rs 1,600 crore.
Serolin Systems sells enterprise software on subscription. Revenue is Rs 20,000 crore, growing 12%, with a 70% gross margin and a 40% EBITDA margin, Rs 8,000 crore of EBITDA.
2Your task
Which multiple would you use for each company, what drives each one's value, and where would you look for evidence of growth?
Quick check
Why is EV/EBITDA useless for Nuvexa today?
Worked solution
Try it on paper, then open one step at a time.
30-second answerThe answer to give first
Value Serolin on cash earnings, EV/EBITDA or free cash flow yield; value Nuvexa on forward revenue, backed by a model of the path to a mature margin. Serolin's value rests on retention and margin. Nuvexa's rests on three things: how fast revenue compounds, how far compute cost per unit falls, and how much new equity it must raise before it turns profitable, about Rs 3,750 crore on these numbers.
Step 1Why do these two companies need different multiples?
Compare a new restaurant in its first year with a café that has run for twenty. The café you value on what it earns, because those earnings are steady. The new place has no earnings yet, so you look at how full it is, how fast the queue is growing, and whether the food costs leave room for a profit once it is busy. A multiple only works on a number that is positive and representative of the future; Serolin's EBITDA is both, and Nuvexa's is neither.
So Serolin goes on EV/EBITDA, EV/free cash flow or P/E, compared with mature software peers. Nuvexa goes on EV/forward revenue, compared with other high-growth peers, and every revenue multiple is shorthand for a story about future margins. Two companies at the same revenue multiple can be worth very different amounts if one has a 45% gross margin and the other has 80%, because the gross margin caps the profit the revenue can ever turn into.
Step 2What actually drives each company's value?
For Serolin: net revenue retentionRevenue this year from last year’s customers divided by their revenue last year. Above 100% means existing customers spend more over time even after some leave., price increases, and margin held at 40%. Its 12% growth is mostly existing customers buying more. For Nuvexa, the value lives in the gross margin line. Inference compute is 40% of revenue today. If the cost of serving each request falls faster than prices, gross margin climbs toward software levels; if prices fall as fast as costs, it never does. That one race decides whether Nuvexa is a software company or a utility reselling chips.
Model the path to make the revenue multiple honest. Assume growth slows from 150% to 30% by year 5 and EBITDA margin climbs from minus 80% to plus 25%. Revenue reaches Rs 29,120 crore and EBITDA Rs 7,280 crore in year 5. Losses in years 1 and 2 add up to Rs 3,750 crore, which must be raised from new investors.
Step 3What does that path say about the revenue multiple?
Put an illustrative mature multiple of 18x on year-5 EBITDA, the same multiple applied to Serolin today, and discount five years at 20% for the risk that the path fails. That gives about Rs 52,662 crore today before the dilution from new funding, roughly 26x current revenue. Serolin at 18x is about Rs 144,000 crore, 7.2x revenue. Nuvexa can justify a revenue multiple several times Serolin's, but only if the mature margin arrives: at 15% instead of 25%, the value falls by about Rs 21,065 crore.
| Measure | Nuvexa Labs | Serolin Systems |
|---|---|---|
| Revenue, Rs crore | 2,000 | 20,000 |
| Growth | 150% | 12% |
| Gross margin | 45% | 70% |
| EBITDA margin | -80% | 40% |
| Growth plus EBITDA margin | 70 | 52 |
| Multiple to use | EV / forward revenue | EV / EBITDA, FCF yield |
| Illustrative value / revenue | 26.3x | 7.2x |
Step 4Where would you look for growth in each?
For Nuvexa: usage volumes, the number of paying enterprise customers, revenue per customer, the price per unit of model output and the compute cost per unit, quarter by quarter. A rising gross margin while volumes grow is the single best sign. For Serolin: net revenue retention, new modules sold into the base, price rises at renewal, and whether it can add AI features that customers pay extra for. Close the comparison with the risk on each side: Nuvexa can run out of funding or margin; Serolin can be disrupted by companies like Nuvexa. That is the judgement the interviewer is waiting for.
Where candidates lose it
The common miss is picking EV/revenue for Nuvexa and stopping there. A revenue multiple with no view on the margin it implies is a guess; the follow-up is always what margin you are assuming and why.
The second is forgetting dilution. Nuvexa's losses are funded with new shares, so a per-share value today has to allow for the equity still to be raised, or it overstates what current holders own.
What the interviewer asks next
- Nuvexa's compute supplier raises prices 20%. Which line moves, and what happens to your valuation?
- How would you value Nuvexa with a DCF, and which assumption would you sensitise first?
- Serolin's net revenue retention falls from 115% to 102%. What does that tell you?
- Why might an acquirer of Nuvexa pay more than a financial investor would?
Asked at Evercore, Mergers and Acquisitions, New York, 2026 (Wall Street Oasis): I was asked questions about which multiple to use, how to value the business, where would I look to see for growth drivers.
Company names and figures are illustrative.