The AI Boom Isn't in the Numbers Yet

Show notes

Three days apart, the AI economy published its two truths. Nvidia booked $89 billion of data-center revenue in a single quarter — up 117% — while McKinsey's annual survey found the share of companies that can trace any earnings impact to AI stuck at 37%, unchanged in a year. The boom is unquestionably in somebody's numbers. This week: what the economy-wide data actually shows, the beautiful slide being built on top of it, and how to stop borrowing the economy's numbers and start producing your own.

This episode of Above the Noise — the unbiased AI brief for enterprise leaders:

• News Brief: The new Fed Chair used his first Jackson Hole keynote to call AI a potential new factor of production — $100B+ in annualized token sales from the two leading labs — and admitted the central bank can't yet tell whether all that machine output complements your people or competes with them. If the Fed won't call it, your strategy deck doesn't have to pretend it has · Nvidia's historic quarter: $96.2B in revenue, $89B of it data center, $108B guided next quarter — the cleanest measurement in AI, as long as you remember capex is a fact and ROI is still a forecast · McKinsey's State of AI survey: deployment up everywhere, earnings impact flat at 37%, true high performers at 6% — and 39% now expect AI-driven job cuts, up from 32%. Expectations are outrunning evidence inside the same buildings.
• Expose a Lie ⭐: "Productivity is booming. That's AI. The transformation is already in the numbers." The chart is real — the Kansas City Fed measured the productivity pickup at ~2.5 points annualized, double the 2010s pace. The claim on top of it is borrowed. Three cracks: the top four industries driving the boom are the same four that led before a single chatbot shipped ("the leaderboard is the old names, in bold") · fewer than one in five businesses tells the Census Bureau it uses AI at all, and the Fed found adoption explains little of the aggregate gain · and the earnings floor — record receipts on the spend side, flat returns on the buy side. Right now the AI economy is measured in receipts, not returns.
• The Playbook: How to find your own concentration instead of quoting the economy's average. Three moves: ban the macro slide in both directions (no national statistics in any AI business case — replace it with your register, a number each system must move, and whether it moved) · hunt concentration, not coverage (pick the four workflows where machine work is worth the most; fund the redesign; starve the vanity pilots) · invest like it's real, report like it's early (every workflow gets a number, a review date, and a kill-or-scale fate). The Line for the Meeting: "That's the economy's number. Show me which of ours moved."
• The question to sit with: If every AI system in your company stopped tonight, which line of your P&L would move — and how long before anyone noticed? If the honest answer is "nothing measurable, for weeks," your AI budget is currently funded by conviction, not results. Conviction is fine. Just know which one is writing the checks.

No vendors. No hype. Just the signal.

👉 Follow Above the Noise wherever you listen — new episode every two weeks. Send it to one leader who just forwarded you a national productivity chart as a business case, and leave a rating so more people find it. Between episodes, find Shaun on LinkedIn — come argue with him there. Especially if your P&L would notice tonight. He wants to see the line item.