Why Big Companies Are Now Asking: Is All This AI Spending Actually Worth It?

Tiffany Luck is a venture investor who specializes in early-stage AI companies. In a recent podcast appearance, she discussed a shift happening right now in how large companies think about their AI investments.

The shift is this: two years ago, companies rushed to buy and build AI tools. Now they are asking whether they actually got anything valuable in return.

The Problem

Uber spent its entire year's AI budget in the first four months of 2026, mostly on cloud-based AI tools. When executives looked at the spending, they could not clearly connect the cost to any improvement for customers. They had no simple way to say: "We spent this much on AI and it made us this much money" or "We spent this much and now customers are happier."

Meta, another large tech company, built a system to track how much employees were using AI. When that data leaked to the public, they shut it down immediately. The incident revealed something: companies are both desperate to understand their AI spending and nervous about admitting they don't know if it's working.

Companies that bought AI tools quickly in 2024 and 2025 are now having difficult conversations during budget reviews. The question has changed from "Should we use AI?" to "Did the AI actually help us?"

Investors and startup founders see this as an opportunity. Companies that can show clear proof that AI made them faster or more productive will win more customers. Luck is backing startups that help other companies measure and track AI's real impact.

Personal AI Agents Are Coming

The next stage of AI technology involves "personal agents." Unlike today's AI assistants, which answer questions when you ask them, personal agents will work on your behalf without waiting for instructions. Think of it like hiring an assistant who can manage your calendar, write emails, and handle routine work — but the assistant is software, not a person.

Building this kind of AI requires new technical work: the AI needs to remember conversations over time, it needs permission to access your different tools and accounts, and it needs to be fast enough that it does not slow you down. These requirements mean new tools and services will need to be built, which is where investors like Luck are watching carefully.

What Happens Next

Several AI companies may go public in the next couple of years. That would tell us something important: are large investors willing to bet that AI software companies can make real money at scale. Right now, the answer is unclear.

Luck's message to founders is simple. You need to prove three things: that customers can find your product, that you have a real advantage competitors cannot copy, and that you can measure the actual benefit to customers. Those three questions are much harder to answer now than they were in 2024.