Snowflake just settled a $285 billion argument for SaaS and Agentic AI.

For two years, every board I talk to braced for the same ending, where agents replace seats, seat pricing collapses, and software dies. The market believed it enough to erase roughly $285 billion in software value on that single bet. Then the numbers landed in late May, and the bet fell apart.

Snowflake reported revenue up 33 percent to $1.39 billion , signed a fresh $6 billion deal with Amazon, and watched its stock run almost 50 percent in four days, logging its best day ever in the middle of the run. Salesforce, still selling mostly by the seat, slipped on soft guidance the same week. The whole sector caught fire, and software posted its best month since 2001.

For a year the talk had been SaaSpocalypse , the fear that vibe coding would let anyone build an app in an afternoon while the seat-based giants quietly bled out. Now we are hearing different messages and see job roles pick up in Agentic AI.

Jensen Huang Says Agentic AI Makes Snowflake Stronger

Then Jensen Huang walked on stage at GTC Taipei and said the opposite of what everyone feared. People had been warning him that agentic AI would wipe out software companies, and he told them it was exactly the opposite, calling it an incredible time to be a software company. He backed it with numbers the doomers had skipped.

GitHub commits jumped from 500 million in 2025 to nearly 1.4 billion in the first months of 2026, roughly a tripling, alongside 90 million pull requests merged and 20 million new repositories created every month.

By Huang’s math, the same 30 to 40 million developers who cost the industry about $3 trillion in salaries now generate closer to $9 trillion in output, which is why he argues the agent era keeps hiring engineers.

The line that makes a difference is this one: “There are going to be so many agents, the world is no longer limited by the number of people. Those agents are going to use more tools than ever.”

More agents doing work means more software running underneath them, and that is the whole bull case in one sentence.

Agentic AI? The Numbers Behind Snowflake’s Sector

Here is how far the named players moved in the final week of May, the stretch when the rally broke open.

  • Snowflake: nearly +50% for the week, with its best day ever, a 36% single session, in the middle of it
  • Okta: +30% on May 29, on an earnings beat with $765 million in revenue
  • ServiceNow: over +20% for the week
  • Atlassian: +26% for the week
  • Oracle: about +16% for the week
  • Microsoft: nearly +8% for the week, though still down about 7% on the year

Now look at who won on results, beyond the names that simply bounced off the floor, and a pattern shows up fast: the leaders all charge the same way, by how much you use.

Snowflake and Okta meter the work, so when agents multiply, their revenue multiplies right alongside. Snowflake alone added 616 net new customers in the quarter and now counts 779 accounts spending over a million dollars a year, with net revenue retention at 126 percent and remaining performance obligations of $9.21 billion, up 38 percent, which is what a company accelerating into the shift looks like.

The flat per-seat names jumped too, though that move read as relief after a brutal year down as much as 60 percent.

What Snowflake Just Taught Every CEO And CFO About Agentic AI

For chief executives, the lesson is positioning. The companies winning this cycle have made their platforms the place agents do their work, then priced what those agents consume, because a seat stops measuring value once a single employee directs ten agents.

The larger advantage compounds through the ecosystem, where every partner integration that routes agent activity through a platform expands revenue without expanding the sales force, a dynamic seat licenses never offered.

Oracle and ServiceNow are pressing that advantage from the infrastructure and workflow layers, each racing Snowflake to own the layer agents run through. Salesforce illustrates the harder path, since Agentforce is a credible response while the seat-heavy core beneath it continues to draw the toughest questions from investors.

For finance chiefs, the dashboard itself changes. Consumption, net revenue retention, and remaining performance obligations move to the center of the story, and the upside they capture comes paired with real volatility, since usage falls as quickly as it rises when customers optimize.

Running the model well calls for tighter guidance, scenario ranges, and early warning when the largest accounts decelerate. The rally also deserves proportion, because even after the best month since 2001, the iShares software ETF sits down roughly 4 percent on the year while the Nasdaq has gained 18 percent, marking a sector that is clawing its way back with plenty left to prove.

The June earnings wave will test all of it, because momentum and a durable model are different things, and the next several quarters will settle which one this rally represents. The direction, though, is already set. Snowflake has shown every software leader where value accrues in the agentic ai era, and the companies that bill for the work delivered are the ones the market has started to pay for.