Every Software CEO Is Narrating an AI Inflection. Few Can Prove It.
The earnings season is almost behind us, and the AI SaaS apocalypse — promised for two years running — has once again failed to show up. Software companies are not being hollowed out by AI. They are adapting to it. It is still too early to declare winners and losers. It is not too early to identify who is moving faster.
The hard part is that the income statement no longer tells you who. Every management team narrates an AI inflection on every call, and most can point to a reported growth number that seems to back it up. The discipline that separates signal from story is simple to state and tedious to apply: strip out the acquisitions and the lumpy multi-year deals, then check whether the acceleration survives — and whether the consumption and expansion metrics that can't be financially engineered agree with it. For one company, it survives, and a half-dozen independent metrics confirm it. For the other four, the reported acceleration thins to nothing once you remove what was borrowed.
Showing It in the Numbers
Snowflake
Snowflake is the one name where the acceleration is real and visible from several directions at once.
Product revenue YoY growth has turned up: 32% (Q2 FY26) → 29% (Q3 FY26) → 30% (Q4 FY26) → 34% (Q1 FY27). The Q1 FY27 sequential dollar addition of $107.7M dwarfed prior quarters' ~$68M. Critically, this is not a single-metric story. $1M+ customers nearly doubled year over year (46 crossings vs. 26). Net revenue retention ticked up to 126%, breaking a multi-quarter flat line. Cortex Code reached 7,100 accounts in a single quarter post-launch, and Snowflake Intelligence accounts doubled quarter over quarter. Non-GAAP operating margin expanded at the same time revenue accelerated — the combination most companies cannot manufacture. And management raised FY27 product revenue guidance mid-Q1, from $5,660M to $5,840M (+$180M, +3.2%).
That is what an inflection looks like when it is real: the headline number, the customer-cohort data, the retention metric, and the margin line all move in the same direction in the same quarter. None of it is borrowed from an acquisition.
Still Telling the Story
The next four companies all report a number that looks like acceleration. In each case, the acceleration lives somewhere other than the organic, recurring business — in an acquisition, in multi-year backlog, or in a handful of lumpy deals. The consumption metrics, where AI demand would show up first and where it cannot be financially engineered, sit flat.
ServiceNow
ServiceNow's subscription revenue growth (constant currency) fell from 21.5% in Q2'25 to 19.0% in Q1'26. Adjust for the ~100bps of guided Moveworks contribution and organic growth is closer to 18%. The Q2'26 guide of 21–21.5% CC appears to show reacceleration, but it includes 125bps from the Armis acquisition; the implied organic rate is ~19.75–20.25% — stabilization, not genuine acceleration. Management has progressively leaned on M&A to mask the organic trend. The one genuinely reassuring data point is cRPO, stable at 21% CC, which says bookings are holding.
Salesforce
Salesforce's organic growth is roughly 8% in constant currency, and has been for four consecutive quarters. The reported line tells a more flattering story because Informatica closed in November 2025 — a partial quarter of contribution in Q4 FY26, the first full quarter in Q1 FY27, adding roughly $444M to total revenue, about 4.5 points of growth.
Sequence the last four quarters of CC growth and the entire thesis fits in one table:
| Q-3 | Q-2 | Q-1 | Latest | |
|---|---|---|---|---|
| Reported CC revenue growth | 9% | 8% | 10% | 12% |
| Organic (ex-Informatica) | 9% | 8% | 6% | 8% |
| Reported cRPO growth | 10% | 11% | 13% | 13% |
| Organic cRPO | ~9–10% | ~9–10% | ~9–10% | ~9–10% |
Reported, it looks like acceleration into the low teens. Organic, revenue growth has been stuck at 8% the entire time, and organic cRPO — the forward-bookings tell — never left the 9–10% band. The Q4 jump in reported cRPO is almost entirely Informatica.
Agentforce, the engine management is pointing to, is still tiny. ARR moved $540M → $800M → $1.2B over three quarters — real momentum, but $1.2B is only about 2.6% of FY27 revenue. Even if Agentforce doubles over the next year to $2.4B — which only requires sustaining roughly $300M per quarter of net new ARR, in line with its recent pace — it adds about 3 points of total growth. With the core clouds at 7% CC, that takes total organic growth from 8% to maybe 10–11%. Real, but not the inflection management is signaling.
MongoDB
MongoDB is a high-quality business growing a little more slowly on the margin — don't be fooled by the 25% headline and the beat-and-raise. The engine is Atlas (75% of revenue), and Atlas has been impressively flat at ~29–30% YoY for four quarters. It is durable, but it is not accelerating. The splashy Q4 strength and +97% RPO were lumpy multi-year enterprise deals, and management guided that line back to low-single-digit growth for the coming year. Management's own FY27 guide implies a step-down to ~19–20% even before its habitual sandbagging. And the heavily promoted AI angle remains, by management's own admission, immaterial.
Elastic
Elastic's total revenue growth (CC) printed 18% → 15% → 16% → 14%, with Q4 the year's low and sequential revenue essentially flat (+0.2%). The metric management leads with — sales-led subscription — decelerates from FY26's 20% to an FY27 guide of 16.9%. Elastic Cloud, the historic engine, decelerated all four quarters to 20% and went slightly negative QoQ. The single most damning fact against the "AI is accelerating consumption" narrative repeated on every call: Net Expansion Rate never budged from ~112%.
Management points to cRPO accelerating to 20% CC in Q4 (from 15%) and RPO to 28% (from 22%). That acceleration is real — but it is concentrated in non-current, multi-year backlog (+43% YoY) and a handful of lumpy eight-figure deals (a CISA federal contract, a Fortune 50 SOC). It is backlog, not consumption.
What separates the camps
Across all five names, the analytical move is identical, and it is the only move that matters this season: take the reported acceleration and ask where it actually comes from.
Snowflake's survives the test. Strip nothing out and the growth still turns up, corroborated by customer crossings, retention, product adoption, and margin — four independent witnesses telling the same story. The other four fail it, each in its own way. ServiceNow and Salesforce borrow their reported acceleration from acquisitions — Moveworks and Armis, Informatica. MongoDB and Elastic borrow theirs from lumpy multi-year deals and backlog. And in every one of those four cases, the metrics that can't be financially engineered — organic cRPO, Atlas growth, Net Expansion Rate — are sitting still.
None of this is a call on where these stocks go from here. It is a record of what the filings actually changed this quarter, written down so the read can be checked against what happens next. The companies still telling the story are not broken — several are excellent businesses. They simply have not yet produced the evidence that the AI inflection is in the numbers rather than in the narration.