RingCentral Doubled in 90 Days. The Revenue Line Still Says 5.9%.

A record-margin quarter, gross debt under $1 billion, and AI-attached ARR at 15% — the tape paid up for the option, not the P&L.

RNG · Earnings Call · 2026-10-22

The gap between the tape and the P&L

RingCentral's stock is up 108.7% over the last 90 trading days, closing at $82.42 on October 21 — its high for the period, just 3.6% off that mark going into the print. For a company whose entire history from the February 2021 peak of $443.29 reads as a 82.1% drawdown, that is a violent re-rating. And yet the quarter underneath it grew total revenue 5.9% to roughly $676 million. Subscription revenue was approximately $653 million, up 6%. That is not a growth story. So what did the market buy? Two things. First, a valuation reset so deep that almost any operating progress looks like a re-rating: price-to-revenue sits at 1.3x, up 20% year over year but 96% below the 29.0x peak of 2020Q4. Second, a genuinely new disclosure: the AI attach rate finally has a denominator.

The AI flywheel, now with numbers

Management's framing for six quarters has been agentic voice AI. The novelty in Q3 is that it can now be measured as a share of the business rather than as a customer count. CFO Vaibhav Agarwal put it plainly: “Customers who are using 1 or more paid AI products are now approaching 15% of ARR. Doubling year over year.” — Vaibhav Agarwal · 2026-10-22 That is over $400 million of ARR influenced by paid AI, against the "almost 10% of our ARR, so in the $250 million range" Vlad Shmunis described in February — a step-change in both numerator and disclosure quality in eight months. The product-level numbers: AI Receptionist above 22,000 customers, up nearly 4x year over year; ACE approaching 7,000, up 60%; and the customer engagement bundle, CEB, past 14,000 paying customers, up nearly 50% quarter over quarter, with roughly two-thirds of those buyers also taking at least one AI product. That last statistic is the one that matters — it is not a new logo count, it is an attach rate, and it is the mechanism by which a 6% grower creates a faster-growing sub-segment inside a sticky base. The related customer engagement bundle is the on-ramp; AirPRO is the upsell.

So there are 2 fundamental growth drivers for RingCentral. 1 is more calls. And the other 1 is more AI. So they tend to go together... our traffic has growth has been strong, but linear. Our AI growth has been explosive, obviously. But overall penetration of AI and AI answered calls on our network is still very, very small percentage wise.
Vladimir G. Shmunis · 2026-10-22

That is the honest version of the bull case: the base business is a metered utility that grows linearly, and AI is an attach layer whose penetration is still in the low single digits. It also reframes the margin question. Agarwal's claim that “our gross margins on AI products are generally holding in line with the corporate average as well” — Vaibhav Agarwal · 2026-10-22 is the single most load-bearing sentence on the call — the bears' case has been that AI inference costs would eat the ~80% subscription gross margin. So far, they have not.

Free cash flow per share is the actual product

The number management wants you to underwrite is not revenue. It is free cash flow generation. Q3 free cash flow was $157 million, up 21%, and the full-year outlook was raised to $625 million at the midpoint — nearly 24% of revenue, over $7 per share, up about 23%. Free cash flow margin (less SBC) reached 21.1%, up 5.8 points year over year, and the improvement is visible in the multi-year shape rather than a single print. The balance sheet work is real: gross debt cut below $1 billion ahead of the stated target, net leverage at 1.3x, no maturities until 2030, $305 million undrawn. Effective net cash improved to -$1.0 billion from -$1.5 billion a year ago, alongside a share count down 5% to about 87.7 million and $300 million-plus left on the buyback. One accounting asterisk deserves more attention than it got. GAAP EPS of $6.50 included a $545 million noncash release of a deferred tax valuation allowance; excluding it, EPS was $0.28. No analyst pressed on it in Q&A. The release is itself evidence of durable GAAP profitability — operating margin hit a record 7.7% GAAP in the latest filed quarter — but the headline optics flatter a quarter that was actually about SBC down 10% plus steady execution.

The contrarian tell: RingCentral versus the software tape

Here is the dissonance worth sitting with. Across the market's own price tape over 360 days, the software vocabulary is being sold: net new ARR is a decliner, as are "system of record," "cloud revenue growth," and agentic operating system. The AI trade has rotated away from pure software narrative names and even away from AI ARR as a standalone slogan. RingCentral doubled into that rotation. The reason is that RingCentral is not selling ARR growth — it is selling a cash-return story with an AI option attached, priced at 1.3x revenue and roughly 8x free cash flow. P/FCF sits at 8.2x, actually down 23% year over year. Investors did not pay for AI; they paid for a de-risked balance sheet and got the AI optionality free.

What is still being deferred

The channel story has slipped again, quietly. A quarter ago Shmunis told investors “GSP, RCAI in GSP story is probably not so much for this year, but '27, '28 from that.” — Vladimir Shmunis, Founder, Chairman and CEO · 2026-05-08 In Q3 the company extended BT Business, added Digital Island in New Zealand and Iliad Italia — real progress, but the language had not changed: “as a word of caution when we say that it is still pretty early... through the channel in particular with GSPs. They just tend to move a little slower.” — Vladimir G. Shmunis · 2026-10-22 Seventeen global service providers remain an option, not a growth engine. That is the one claim a reader should discount hardest. The other thing management stopped leaning on is M&A. Asked about buying versus building, Shmunis was unambiguous: “we are able to deliver the results without having to go outside” — Vladimir G. Shmunis · 2026-10-22 — with over $250 million a year of product spend in-house. Given the FCF per-share framing, that is coherent. Buybacks at 8x cash flow beat acquisitions at software multiples. So the story is this: a 5.9% grower with a 24.7% non-GAAP operating margin, a sub-$1 billion gross debt load, and a quantified AI attach rate at 15% of ARR and doubling. The 109% move was not a forecast of accelerating revenue — Q4 guidance implies more of the same. It was the market admitting it had priced the balance sheet, not the platform.