MathCo’s founders had an advantage when they started the company in 2016. They had already helped build Mu Sigma. Sayandeb Banerjee was its first employee, and the founding team knew how large analytics accounts were won, how delivery teams were built and how young engineers were trained for Fortune 500 clients.

That experience helped MathCo grow quickly. Operating revenue reached ₹621 crore in FY26, up 24% from ₹502 crore a year earlier. The worrying number sits elsewhere. Profit fell from about ₹64 crore to ₹3.8 crore.

MathCo added roughly ₹120 crore of revenue during the year. Employee costs increased by about ₹123 crore, from ₹374 crore to ₹497 crore. Salaries alone absorbed all the incremental revenue before software, travel, offices and other costs were paid.

Employee expenses went from roughly 75% of revenue to 80%. For every ₹100 spent on employees, MathCo generated about ₹134 of operating revenue in FY25. That fell to ₹125 in FY26. That is awkward for a company selling AI.

AI is supposed to help analytics firms do more work with fewer people. MathCo, at least in FY26, moved in the opposite direction.

Some of this may be temporary. The company hired heavily across data science, AI and engineering. New employees can spend months in training before becoming fully billable, while experienced AI talent is expensive. If utilisation improves, margins could recover quickly.

But there is a harder question. MathCo may be expanding with the old analytics-services model at the same time AI is making that model less useful.

The old equation was simple: more clients meant more projects, which meant more analysts and more billable hours. AI weakens that link. A project that once needed ten people may now need six.

That helps margins only if MathCo keeps much of the productivity gain. If clients continue paying mainly for people and hours, fewer employees can also mean lower revenue.

Fractal offers a useful comparison. Its FY26 revenue grew about 19% to roughly ₹3,300 crore, while profit increased around 30%. Revenue per billable employee rose about 5% to roughly ₹75 lakh. Fractal said pricing, productivity and more output-based work helped margins.

MathCo needs a similar change. NucliOS matters only if it changes the economics: fewer engineers per project, more reusable work, faster delivery and more recurring revenue.

Cash adds another warning. Cash and bank balances fell from roughly ₹53 crore to ₹15 crore during FY26. With nearly ₹500 crore of annual employee costs, poor collections can make growth itself expensive.

The opportunity is still large. If employee costs fell from 80% of revenue to around 70% on ₹700-750 crore of revenue, MathCo could free up roughly ₹70-75 crore.

Let’s go through what else is happening in Indian startup world - Grab your simmering cup of StartupChai.in and unwind with our hand-brewed memes.

Anthropic has launched Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, 20% below Opus 5 pricing.

OpenAI has also released GPT-6 Sol and Luna with API prices 50% below the promotional pricing of its previous GPT-5.6 models.

Read more here

PayGlocal has received in-principle approval from IFSCA to set up a wholly owned PSP subsidiary in GIFT-IFSC.

The move will strengthen its cross-border payments operations.

Read more here

Mastercard has exited Pine Labs by selling its entire 4.31% stake for ₹934 Cr through a bulk deal.

ICICI Prudential Life Insurance was the largest buyer, with Societe Generale and Citigroup Global Markets Singapore among the other purchasers.

Read more here

Health Care at Home has appointed Sudhir Bahl as CEO and a member of its board. Bahl previously served as COO of ICICI Venture and India operations director at TVM Capital Healthcare, and also founded healthcare venture LifeForce.

The Quadria Capital-backed company operates in more than 70 cities with over 2,000 specialists and says it has served more than five million patients since 2012.

Read more here