Executives gathered at Fortune’s Brainstorm Tech conference last week delivered a blunt message. Most companies chasing returns from artificial intelligence investments miss the mark. They treat the technology as a plug-and-play solution. Success demands something harder. It requires breaking problems down to core truths and rebuilding operations around them.
Jeremy Kahn reported the details for Fortune. Panelists stressed that shortcuts lead nowhere. Firms race to deploy chatbots or summarization tools. Few pause to examine underlying processes first. The result? Scattered pilots. Modest gains. And growing skepticism from boards demanding proof of payback.
One in ten companies sees real financial returns from AI initiatives. That statistic circulated widely on social platforms this week. It comes alongside findings that 94% of executives view the technology as critical. Only 17% feel prepared. The gap reveals itself in budgets. Billions pour into infrastructure. Measurable uplift stays rare.
Big Tech leads the spending spree. Alphabet, Amazon, Meta and Microsoft together plan as much as $665 billion in AI-related capital expenditures this year. The figure dwarfs prior outlays. Yet questions mount. Hardware depreciates fast. Energy costs climb. And many deployments fail to move core business metrics.
Meta’s Mark Zuckerberg addressed ROI directly during earnings calls. He called it a technical question. The company raised its 2026 capital spending forecast. Investors pressed for clarity on returns from massive data center builds. Zuckerberg pointed to improved understanding of user preferences through AI. Still, analysts remain cautious.
But first-principles thinking changes the equation. Executives at the conference described it as starting with fundamental questions. What customer problem does this solve? Which steps in the current workflow create friction? How does data flow today, and where could intelligence alter outcomes? Answers rarely point to bolting AI onto existing systems.
Process reinvention follows. Companies that succeed redesign workflows from the ground up. They identify bottlenecks. They train models on proprietary data. They measure impact against baseline performance before and after changes. This approach takes longer. It delivers larger gains.
Consulting firm EY outlined a path out of the trap in a May report. Tech companies achieve better results through end-to-end transformation. They set KPIs tailored to specific initiatives. Strong governance keeps efforts aligned with business goals. EY analysts noted that treating AI as a strategic investment rather than an experiment matters most. Discipline separates winners from those burning cash on pilots that fade.
IBM’s 2026 CEO Study reinforces the point. Leaders who adopt an AI-first mindset scale 10% more initiatives across the enterprise. They rethink C-suite structures for faster decisions. Custom models built on proprietary data and intellectual property yield higher expected revenue from new offerings. CEOs incorporating these elements forecast 13% more of their 2030 sales from products that don’t exist today. The study, available at IBM, shows clear separation between experimenters and transformers.
CFOs face particular pressure. Generative AI scrambles traditional ROI calculations. Earlier technologies allowed years to prove value. This wave demands faster assessment. Costs often exceed projections by orders of magnitude. Nisha Bhandare, a vice president at Gartner, highlighted the uncertainty in a conference session covered by CFO Dive. Organizations learn spending patterns in real time. Many lack visibility into token consumption or compute allocation.
Some leaders push back against immediate ROI demands. Nvidia’s Jensen Huang compared it to requiring a child to produce a business plan for a hobby. Experimentation needs room to breathe. Yet boards grow impatient. Kyndryl’s readiness report found 61% of senior leaders feel greater pressure to demonstrate returns than a year ago. Investor surveys show half expect positive results within six months.
The fork in the road appears. Companies can chase short-term cost cuts through headcount reduction. Or they pursue genuine operational transformation. The latter demands first-principles work. It means mapping processes. Cleaning data. Building custom agents. Training staff. And accepting that some experiments fail.
Data governance forms the foundation. Without clean, accessible information, models produce unreliable outputs. Security and privacy controls cannot lag. Ethics reviews belong early in the cycle. ESI ThoughtLab’s study of 1,200 organizations, conducted with partners including Deloitte, identified these practices among high performers. Those with strong foundations report better returns. The analysis appears on the ThoughtLab site.
Token spending comes under fresh scrutiny. Several companies exhausted annual budgets for AI coding tools in months. Uber’s experience, shared with The Information, showed how internal competitions to increase usage can accelerate costs without corresponding value. Leaders now seek metrics beyond activity. They track cycle time compression. Customer satisfaction scores. Margin expansion. Mean time to repair.
Business unit owners hold accountability for outcomes. Technology leaders manage platforms and budgets. This separation prevents enthusiasm from outrunning results. Podcasts like AI to ROI, hosted by Benchmarkit’s Ray Rike, feature executives who insist on defining value metrics before writing code.
Stanford’s 2026 AI Index Report, released this week, captures the scale. U.S. private AI investment hit $285.9 billion in 2025. Adoption spreads quickly among consumers who use free tools daily. Enterprise results trail. The full report from Stanford HAI shows the United States dominates funding and startup creation. Talent inflows, however, have slowed dramatically.
So what separates the few who capture value? They start with questions a child might ask. Why does this process work this way? What assumptions underpin it? Where does waste hide? AI then amplifies solutions built on those answers. Not the other way around.
Expect more discipline in 2026. Boards want numbers. Investors track capex against revenue growth. Early wins build credibility. Quick pilots that deliver measurable gains in weeks create momentum. Moonshots come later.
The executives at Brainstorm Tech agreed on one truth. AI will not deliver returns through magic. It rewards rigorous thought, organizational courage and patience. Those who apply first-principles thinking now position themselves to lead. Others risk watching their investments depreciate alongside yesterday’s hardware.
WebProNews is an iEntry Publication