Table of Contents
Jensen Huang tells us that our AI agents will text us the most. Twilio’s $1.5B in quarterly receipts suggest he may be right and that they may be calling us even more than they are texting.
More interestingly, agentic AI may be reversing the industry’s multi-sourcing habit, which is also helping Twilio cross-sell. AI-native customers are starting with Voice, growing their traffic, attaching software, and then adding Messaging or another channel. Twilio’s dollar-based net expansion (DBNE) among AI-native customers is well above the company average.
On a numbers basis, revenue grew 22% (17% organically). Messaging grew 28%, though carrier pass-thru (see explanation here) supplied roughly ten points of that growth. Voice grew more than 20%, and software add-ons grew more than 25%. Twilio also raised full-year organic growth guidance from 9.5%–10.5% to 13%–13.5%.
The Good
Self-serve Voice grew more than 50%, and Verify accelerated above 30%. Revenue from customers spending more than $1M annually grew more than 20%.
The conversations layer is also reaching production. Car Finance 247 moved from private beta into a seven-figure agreement. Its AI assistant, Carla, has handled roughly 288,000 conversations, and customers who engage with it progress to approved leads at 1.6 times the rate of those who do not. Twilio says that translates into a multimillion-pound annual revenue uplift for the customer.
The redesigned Console launched in May. Twilio says it is seeing an uplift in conversion of more than 90%, compared with the old Console, as customers move through sign-up and first-product activation.
Twilio is also integrated into all the tools the cool kids are using, from Codex and Claude Code to Replit, Figma, and Cursor, putting itself inside the environments where agents are already being built.
That is an important competitive advantage. Twilio is not waiting for AI companies to mature before showing up. Twilio gets the developer early, gives it a low-friction way to experiment, and then has more products to sell as the workload grows. Ask Your Developer is alive and well.
Gross profit is keeping pace too. Non-GAAP gross profit grew 18%, slightly faster than organic revenue, marking the fifth consecutive quarter of accelerating gross-profit growth.
Some of that improvement came from decidedly old-fashioned telecom work. Twilio now has direct 10DLC and toll-free connections with every major US carrier. The AI story still depends on the plumbing.
The Interesting
AI may be reversing the CPaaS industry’s multi-sourcing habit.
Customers have historically spread traffic across providers for price, resilience, and negotiating leverage. Twilio says it is seeing the opposite among AI-native customers.
Conversation Memory and conversation orchestration become more useful when more interaction history sits in one place. That matters beyond cross-sell. If consolidating communications improves the context available to an agent, then single-sourcing can produce a better product rather than merely a better commercial arrangement.
Conversation Memory and Intelligence should reduce token spend because the model can use relevant customer history instead of searching through a much larger body of unrelated information.
Context is precious, and tokens are expensive.
AI-native companies are also contributing to Messaging growth, but Aidan said most of that volume still resembles traditional Messaging use cases. AI may create entirely new Voice interactions. In Messaging, at least for now, it appears to be creating more of the behavior we already know.
And humans are not disappearing any time soon. Khozema expects most interactions over the next couple years to remain human oriented, with AI increasingly augmenting them. High-value interactions in healthcare and financial services will still require identity, validation, governance, and human escalation.
That reinforces the identity thesis from Q4. AI does not make Verify or Branded Calling less valuable; it creates more interactions that need to be trusted.
The Unknown
I’ve been a skeptic of Twilio’s cross-selling initiative (see discussion here and here). I’ve maintained that for cross-selling to work, you need a clear, compelling use case and a simplified sales motion. And I did not see that in Twilio. I didn’t see it in 2012 to 2019 when I competed with them or from 2021 to 2025 when I advised others buying from them. But this quarter, Twilio may finally be on to a working formula.
Cross-selling is strong among AI-native and larger ISV customers. Separately, Twilio’s $1M+ customer cohort is growing more than 20%.
Here’s my theory: I was looking at cross-selling as a product-led motion when it may really be a procurement-led function. For large enough accounts, no one wants to go through new-vendor onboarding. So, for the right amount and the right product, it is easier to use an existing vendor for more of your communications needs.
If you’re a product manager inside the enterprise, that keeps your developer, the business user, the customer, and procurement happy. AI may be pushing the cross-sell, but for large-enough accounts, one purchase order could be doing more work than we think.
There is plenty still unknown: RCS remains small (and undisclosed), and carrier pass-thru fees can eat into your margins on short notice. But those are industry-wide questions, not Twilio-specific ones.
Twilio still does not disclose how much revenue comes from AI, and eventually the AI trade will cool. For now, it is doing what any rational company would do: Ride the demand while it is there without building the whole model around it.
Finally
At $1.5B a quarter, Twilio’s revenue is easily double that of its closest competitor, and it is doing this profitably. But the harder question is whether today’s Voice AI momentum becomes durable platform expansion or whether it remains concentrated in a handful of fast-growing cohorts.
Twilio today reminds me of Roger Federer circa 2005, when Rafa Nadal was just the King of Clay and Novak Djokovic had yet to reach a major final: effortless, dominant, and difficult to dislodge. The question now is whether Voice AI becomes just another winning shot or the thing that pulls the rest of the platform with it. Q2 suggests the latter.