The current earnings cycle offers an unusually clear window into the health of the agentic AI market. Three companies in particular — Salesforce, Snowflake, and Marvell — function as gauges in the dashboard of artificial intelligence, each representing a distinct layer of what can usefully be visualized as an AI layer cake. Reading their results in concert tells a more complete story than any single report could on its own.
The SaaS Apocalypse Question
Few narratives have shadowed enterprise software more persistently than the so-called SaaS apocalypse. Once tools like Claude made it conceivable that SaaS applications could be designed by anyone, anywhere, much of the wind went out of the sails of incumbents like Salesforce. The stock has paid the price, trading near three-year lows even as it earned its place in the Dow Jones Industrial Average. Some analysts believe it could fall further after earnings.
The counter-thesis is that SaaS is not only still alive but is being reborn through agentic platforms. Salesforce is building precisely such a platform, and the earnings print will reveal how much of its large enterprise customer base is actually climbing onto the agentic bandwagon. If those customers are deploying agentic AI effectively, the bear narrative collapses; if they aren't, the implications ripple outward well beyond a single ticker. For investors, the depressed price level itself may represent an opportunity rather than a warning — assuming agentic adoption shows up in the numbers.
The Data Layer and Why Snowflake Matters
Agentic AI is only as good as the data over which its agents can reason. That single sentence captures why Snowflake belongs in this conversation despite operating in a very different niche from Salesforce. The company is no longer best understood as a data warehouse play; it has become a platform purpose-built for agentic workloads, and that repositioning makes it a direct proxy for whether AI return on investment is actually being realized in the marketplace.
The software sector broadly has been beaten down, but Snowflake's exposure to AI demand looks structurally favorable. Any short-term decline likely reflects profit-taking and tactical positioning rather than a deterioration of fundamentals. As another canary in the coal mine for AI overall, the company appears well-positioned to capture demand as it materializes.
The Hardware Layer and Marvell's Position
At the hardware base of the cake sits Marvell, whose stock has surged roughly 228% over the past year and now trades at 52-week highs. Expectations going into earnings are obviously high, but expectations are high across the board for anyone with hardware exposure to AI. Marvell's recent positioning is unusually strong: Nvidia invested $2 billion and brought the company into NV Link Fusion, anchoring it firmly in the center of the AI hardware universe alongside Broadcom and Nvidia itself.
Hesitancy about owning Marvell at these levels is understandable but likely misplaced. Adjustments will eventually arrive across the broader AI space, but the partnerships and structural placement Marvell has secured suggest the company still has meaningful room to run.
When the Layers Disagree
The most informative signal of this earnings cycle may not come from any single beat or miss, but from divergence across the layers. If Marvell delivers strong results while Salesforce guides cautiously, that combination tells a story worth listening to carefully. Weakness on the Salesforce side would be a leading indicator of trouble ahead for the hardware sector, because the industry has frontloaded enormous quantities of hardware and software inventory in anticipation of demand that ultimately must be validated at the application layer.
Where the rubber meets the road is whether Salesforce customers are successfully deploying agentic AI. That is the moment of truth — the point at which all the upstream capital expenditure on chips, networking gear, and data infrastructure either justifies itself or doesn't. A cautious tone from the application layer would be a leading signal that the hardware cycle is overheating.
A Diversified View
Taken together, these three companies form a reasonable basket for diversified exposure across the AI stack. If agentic AI ultimately fails to deliver on its ROI promise, far more than these three names will suffer; the entire ecosystem from chipmakers to data platforms to enterprise software vendors will face a reckoning. But that moment does not appear to be here yet. Tailwinds remain intact across all three layers of the cake, and the upcoming earnings reports should reinforce — not undermine — the broader thesis that the agentic AI transition is real and underway.
The signals to watch are not the headline numbers in isolation but the relationships between them. Each company is a gauge; the dashboard only makes sense when all three are read together.