Salesforce and Snowflake report earnings soon, and the numbers will do more than update two stock prices. According to The Information, these results are about to focus Wall Street’s attention on the question hanging over every enterprise software company right now: is AI a tailwind or a wrecking ball for the business models that built this industry?
That’s the real story here. Not the quarter itself, but what it signals about where software value is heading.
Why these two companies matter
Salesforce and Snowflake sit on opposite sides of the AI debate, which is exactly why investors are watching them together.
Salesforce sells software mostly by the seat. You pay per user, per login, per human doing the work. Snowflake charges by consumption. You pay for the data you process and the compute you burn. Those two pricing models react to AI in opposite directions.
- Seat-based (Salesforce): If AI agents start doing the work humans used to do, companies may need fewer seats. That’s the bear case. The bull case is that Salesforce sells its own agents on top, turning automation into a new revenue line.
- Consumption-based (Snowflake): AI workloads are hungry. More models, more queries, more data movement means more compute billed. In theory, AI feeds this model rather than starving it.
So one company has to prove AI won’t shrink its core. The other has to prove AI is actually showing up in the meter.
What’s changing and why it matters now
For two years, software executives have promised that AI would expand their markets. Talk was cheap. Earnings are not.
This is the moment those promises meet actual customer spending. Are enterprises paying extra for AI features, or just expecting them bundled in for free? Is agent adoption translating into bookings, or is it still pilots and press releases? The market wants evidence, and these reports are among the first clean reads we’ll get.
What stands out is the shift in the question itself. A year ago the worry was whether AI worked. Now the worry is whether AI pays, and specifically who captures the value: the software vendor, the cloud provider underneath, or the customer who pockets the productivity gain.
The broader dynamic
This tension runs through the entire SaaS sector. If AI genuinely replaces routine knowledge work, the per-seat model that defined a generation of software companies starts to look exposed. Vendors are scrambling to reprice around outcomes and usage instead of headcount. Salesforce’s push into agent pricing is a direct response to that fear.
Consumption players look better positioned on paper, but they carry their own risk. AI compute is expensive, and customers are getting sharper about cost. Heavy usage only helps if margins survive it.
Practical takeaways
For practitioners and business buyers, a few things are worth tracking as these results land:
- Watch the pricing language, not just the revenue. How a vendor talks about per-seat versus usage versus outcome-based pricing tells you how they expect AI to hit their model.
- Separate AI bookings from AI hype. Ask whether AI features are driving net-new spend or just defending existing renewals.
- If you’re a buyer, your leverage is rising. Vendors need proof points. That’s a good window to negotiate AI features into contracts rather than paying premium add-on fees.
- Map your own stack. If your software costs scale with seats, AI-driven automation could cut your bill. If they scale with consumption, budget for AI to push it up.
Looking ahead
Over the next year or two, expect the per-seat model to keep eroding at the edges while usage and outcome pricing gain ground. The companies that thrive will be the ones that turn AI from a feature they give away into a product customers choose to buy more of.
Salesforce and Snowflake are the early test cases. Their numbers won’t settle the debate, but they’ll tell us which way the wind is blowing. For the full breakdown, the original analysis is at The Information.