Software stocks rose 21% in May — the best monthly performance since October 2001 , according to CNBC. Snowflake closed 36% higher Thursday for its best day ever , dragging the entire sector out of what many had been calling a "SaaSpocalypse." The fear was simple: if AI agents can build apps in minutes, why pay for enterprise software? Snowflake gained nearly 50% in four trading days following Memorial Day, announcing a $6 billion cloud and chip deal with Amazon as customers gravitate toward more AI tools .
The rally matters because it contradicts the thesis that dominated the first four months of 2026. Software names have been hit particularly hard over the past year due to the boom in so-called vibe coding, with users able to now build apps and websites in minutes thanks to offerings from Anthropic, OpenAI and others . Software shares have struggled, as investors bet AI will undercut the previous profitability of software-as-a-service (SaaS) companies , Axios reported. But the Snowflake earnings beat suggests something more nuanced is happening: AI isn't replacing software companies. It's forcing them to evolve or die.
Can Incumbents Survive the Shift to Agents?
At ServiceNow's Knowledge 2026 conference earlier this month, executives didn't bother defending the SaaS model — they declared it over. "The era of sidecar AI is over," president and COO Amit Zavery told Fortune. "Customers don't want to cobble pieces together — they want outcomes." What ServiceNow is betting on instead is its Context Engine: a governance layer built on 100 billion workflows and 7 trillion annual transactions .
That's the pattern emerging across the sector. Software doesn't die. As Fortune examined in March, Wall Street's conviction that AI will kill SaaS runs up against a stubborn historical pattern — platform shifts tend to enrich incumbents who adapt, not destroy them. Economists and technology historians argued that existing vendors with deep customer relationships and proprietary data are better positioned than newcomers to capture the upside of agentic AI .
The developer tooling market is living proof. In 2026, 95% of developers use AI coding assistants at least once a week, and 75% of developers use an AI coding assistant for more than half of their coding work , according to a survey cited by Orbilontech. Three tools dominate: GitHub Copilot, Cursor, and Claude Code. 84% of developers use or plan to use AI tools, according to the 2025 Stack Overflow Developer Survey. Among professionals, 51% use them daily. The market is real but uneven: 84% adoption, 51% daily use among professionals, 31% monthly agent use, 29% trust , Scrimba reported.
The trust gap is the story. Developers use these tools constantly but don't fully believe them. Developer satisfaction has decoupled from market share. 46% of senior developers name Claude Code their "most loved" tool vs. 19% Cursor and 9% Copilot — yet Copilot still commands the largest installed base , according to a usage analysis cited by Uvik.
What Happens When the Meter Starts Running?
Tomorrow changes everything for GitHub Copilot users. All GitHub Copilot plans will transition to usage-based billing on June 1, 2026. Instead of counting premium requests, every Copilot plan will include a monthly allotment of GitHub AI Credits, with the option for paid plans to purchase additional usage. Usage will be calculated based on token consumption, including input, output, and cached tokens, using the listed API rates for each model , GitHub announced in late April.
Copilot Pro: $10/month, including $10 in monthly AI Credits. Copilot Pro+: $39/month, including $39 in monthly AI Credits . The base subscription prices aren't changing. What's changing is the ceiling. One developer in the community thread estimated that agentic coding sessions — where Copilot plans, researches, and executes multi-step tasks — routinely consume $30 to $40 per session. A Pro user with $10 per month in credits hits their ceiling in a single working session , Enterprise DNA reported.
The backlash has been swift. The golden age of Microsoft's Github Copilot appears to be at an end — for the little guy, at least. The company is switching its billing system from a flat subscription rate to a token-usage system that has the potential to bill users at a significantly higher rate. Bigger enterprises may still have the juice for it, but smaller companies and workers could find themselves wondering how they're supposed to balance the monthly budget , TechCrunch reported Friday.
GitHub Copilot users are pushing back on GitHub's planned move to usage-based billing, with some developers arguing the change will make the AI coding assistant less predictable and less valuable even though base subscription prices are staying the same , Visual Studio Magazine noted. The complaint isn't just about cost. It's about the psychological shift from "unlimited" to metered. For years, Copilot's pitch was deliciously simple: pay a predictable monthly fee and get a coding assistant that seemed to grow more capable every quarter. That simplicity mattered. Developers do not love procurement theater, and small teams especially embraced Copilot because it felt like a tool rather than another cloud meter humming in the background. The new model breaks that spell , a Windows Forum analysis observed.
The timing is awkward. GitHub spent two years promoting agentic workflows — long-running, autonomous coding sessions that span multiple files and execute complex refactors. Now it's billing those workflows like raw compute. GitHub spent the last two years making Copilot feel more autonomous, more persistent, and more ambitious. It promoted agentic development, repository-aware assistance, code review automation, and multi-step flows as the future of programming. When a vendor trains users to offload more work to the assistant, then turns around and says the assistant's long-running behavior must now be priced like raw compute, users will feel baited even if the economics are defensible .
