Twilio Inc. TWLO

Key data

The business

Twilio sells communication as plumbing. A developer drops a few lines of code into an app and it can send a text, place a call, verify a phone number, or fire off an email. Twilio charges per message and per minute, then keeps a margin over what it pays the carriers. No long contracts, no seats: you pay for what you use, the way you pay for electricity. That usage model is the whole story, good and bad. It means revenue tracks real activity in real time, and it means there is very little committed backlog to lean on: remaining contracted revenue over a year out is just $138M against $5.6B of trailing-twelve-month sales. The moat is the switching cost of code already embedded in ten thousand apps and the carrier relationships underneath. The new wrinkle is AI agents that talk to customers by voice and text, which run on exactly these pipes.

The numbers

The quarters show a business re-accelerating after two flat years. Revenue growth ran in the high single digits through 2024, then stepped up to 20% in Q1 2026 and 22% in Q2.

Read the bottom two rows carefully. Q2's $1.07B net income is mostly a $992M one-time release of tax assets, not earnings; strip it and the quarter made about $76M pretax. The real signal is revenue climbing and operating income turning durably positive, not the headline EPS.

The five-year arc is the classic bloat-then-diet: peak losses of $1.3B in 2022, headcount and stock grants cut hard, first full-year profit in 2025. The operating leverage is the part that matters.

Operating margin went from negative to 6.6% in eighteen months while gross margin drifted down: growth is coming from higher-volume, lower-margin usage, and the profit is being manufactured below the gross line by spending discipline. Twilio throws off about $1.1B of free cash flow on $5.6B of trailing sales, a 20% margin, but stock compensation eats 10.6% of revenue, so the cash the owner actually keeps is closer to 9%. What this memo believes that the multiple does not fully price: the re-acceleration is real usage, not a pull-forward, and the single print that settles it is Q3 2026 organic growth landing north of 20% with margin still rising.

Management

The insider tape is one-sided: zero buys, 67 sales, $475M sold in twelve months. The bulk, about $438M across three sales, is Andrew Stafman, the Sachem Head board seat unwinding a fund position rather than an operator cashing out, and the Form 4s do not disclose plan status either way. Weigh that against capital return, where the record is genuinely good: $2.3B of buybacks in 2024 and $869M in 2025 were spent when the stock traded well below current levels, and share count is now slowly shrinking. Founder Jeff Lawson stepped back; Khozema Shipchandler has run the cost discipline that produced the profit turn. Pay is still heavy on stock, which is why the 10.6% dilution charge belongs in every margin you look at.

How it fails or surprises you

Usage re-accelerates on AI (right tail). If AI voice and text agents move real volume onto Twilio in 2027, growth holds in the low-to-mid 20s and 6.6% operating margin doubles on the same cost base. The market is paying for a 13% grower, not a 22% one; the tell is two more quarters above 20% with gross margin stabilizing.

The usage model cuts down fast. With almost no committed backlog and pay-as-you-go pricing, a customer pullback shows up in weeks, not at renewal. No single customer is over 10% of revenue, which caps the damage, but a soft macro quarter would drop growth back to single digits quickly. Watch sequential revenue and net expansion.

The earnings are thinner than they look. Normalize away the $992M tax gain and Twilio runs at about $380M of annual operating income, so the honest multiple is closer to 90x real operating earnings than 31x reported. If margin expansion stalls, the stock is expensive on what it actually makes, and that is the fact this bull read explains least.

Closing thoughts

This is a name where one print resolves the argument. The market has priced Twilio as a low-teens grower that finally stopped bleeding; the last two quarters say it might be a 20%-plus grower with operating leverage still ahead. Q3 2026 is the print: above 20% growth with margin rising confirms the re-rate, a slip back toward the mid-teens says the acceleration was a bump and the 6.3-times-sales tag is too rich. The prior watch here, growth falling below 17%, has not been tested yet because Q3 has not printed, and Q2's jump to 22% points the wrong way for the bears. The fatter tail is up, because the downside is capped by a clean balance sheet, real cash flow, and no customer concentration, while the upside is a full multi-year operating-leverage cycle the multiple only half-believes.

The bet is still that the world keeps sending more texts, calls, and login codes through Twilio's pipes, that the growth is genuinely back, and that companies are wiring AI agents into those same pipes. What breaks it is the pair to watch every quarter, revenue growth and operating margin moving together: if growth slips under 17% or margin flattens while gross margin keeps eroding, the re-acceleration thesis is wrong and you are paying up for a utility.