The air in the Bangalore convention center smelled like high-end espresso and collective anxiety. It’s a familiar scent. We’re at the India AI Impact Summit, a title that suggests something heavy is hitting the ground, though mostly it’s just the sound of venture capital evaporating. The room was packed with IT leaders—the people who actually have to keep the servers humming while the C-suite chases ghosts—and the mood was less "brave new world" and more "how do I explain this bill to the CFO?"

The central tension? SaaS is dying. Or rather, the version of SaaS we’ve been overpaying for since 2010 is getting a forced facelift it didn't ask for. For a decade, the deal was simple: you pay for a "seat," someone sits in it, and they use the software. Now, the seat is empty because a Large Language Model is doing the heavy lifting, yet the software bill is somehow doubling.

"It’s a margin grab," one CTO told me over a plate of lukewarm samosas. He requested anonymity because his company just signed a multi-million dollar renewal with a major CRM provider. "They're charging us for the 'Copilot' features on top of the seat license. We're paying for the human and the ghost in the machine. It’s double-dipping, plain and simple."

This is the specific friction point that nobody in the keynote speeches wanted to touch. If AI actually makes us more efficient, the traditional per-user pricing model makes zero sense. If a marketing department can do the same work with five people instead of fifty, the SaaS vendor loses 90% of its revenue. Their solution? The "AI Tax." We’re seeing a frantic pivot toward consumption-based pricing—charging by the "token" or the "output." It’s like moving from a buffet to an à la carte menu where every glass of water costs five bucks.

The India-specific angle adds another layer of irony. This country has long been the engine room of global enterprise software. But if the engine room is replaced by a cluster of H100s in a data center in Virginia, what’s left for the "Impact" summit to discuss? The panels talked a lot about "reskilling," which is corporate-speak for "find a new job before the script finishes running."

There’s a deep cynicism growing among the boots-on-the-ground IT managers. They’re tired of "wrappers." You know the ones. Startups that raise $50 million to put a slightly different shade of blue on an OpenAI API call and call it an "intelligent vertical solution." At the summit, these wrappers were everywhere. They look like real companies. They talk like real companies. But they have no moat, no proprietary data, and their entire existence depends on Sam Altman not having a bad Tuesday.

One vendor tried to sell me on a "generative coding assistant" that allegedly saves 40% of developer time. The price tag? An extra $30 per month, per head. I asked him what happens when the AI hallucinates a security flaw into the production branch. He blinked. He talked about "human-in-the-loop." That’s another one—the idea that we’ll pay premium prices for AI, then pay humans to spend all day double-checking that the AI isn't lying to us. It’s a cycle of inefficiency sold as progress.

The reality of the "Impact" is a mess of hidden costs. It’s not just the license fees; it’s the data cleaning. You can’t just point a chatbot at a decade of messy, contradictory spreadsheets and expect magic. Most of these IT leaders are spending their budgets on "data hygiene"—a janitorial task that should have happened years ago—just so they can turn on an AI feature that will probably tell a customer to eat glue.

We aren't seeing a shift in how work gets done; we're seeing a shift in who gets the rent. The "Impact" in the summit's name isn't about productivity. It’s about the collision between old-school enterprise budgets and the reality of compute costs. Running these models is expensive. Nvidia wants its cut. Microsoft wants its cut. The SaaS middleman, terrified of becoming irrelevant, is just trying to pass those costs down to the IT guy who just wanted a database that didn't crash on Fridays.

Walking out into the Bangalore heat, I watched a group of engineers arguing over a laptop. They weren't talking about "the power of the future" or whatever the glossy brochure said. They were trying to figure out why their API costs had spiked 300% in a week without adding a single new customer.

If this is the impact, who exactly is being hit?