'KGB' is how the AI spending boom will play out

• Big tech earnings season was dominated by AI capex projections that hit Amazon, Microsoft, and Google stocks.
• Bernard Golden believes there are three plausible scenarios for the future of hyperscaler capex.
• Golden is the CEO of an analysis, consulting, and investment firm focused on emerging technologies.

Big tech earnings season was dominated by AI capex projections that stunned investors and hit the stocks of Amazon, Microsoft, and Google, aka AMG.

This short-term drama makes it hard to think clearly about the long-term trajectory of AMG's businesses. To step back from the volatility, I use a framework called "KGB" that lays out three scenarios for the future of hyperscaler capex.

I've worked in the tech industry for decades and written several books on cloud-computing and open-source software. I've had hands-on experience with major cloud projects at VMware and Capital One. Here's my suggestion for how to think about this unprecedented AI investment wave.

K: Keeping Up With the Joneses

In this future, AMG are locked in an arms race, each spending aggressively to avoid falling behind. A more charitable version says executives believe failing to integrate AI is an existential threat, so they'll spend whatever it takes to avoid becoming the next digital Sears. Skeptics see this as reckless competition — burning cash for bragging rights, with tears and heavy losses inevitable.

G: Goldilocks

Here, AMG spending ends up "just right" to meet customer demand. These companies have unmatched visibility into future AI usage: real-time telemetry, signed long-term contracts, and ongoing enterprise negotiations. From this perspective, rising capex simply reflects confidence in durable demand and solid monetization.

B: Boat

As in Jaws, when the police chief finally sees the gigantic shark and mutters, "You're gonna need a bigger boat." In this scenario, no matter how much AI capacity AMG builds, it gets absorbed immediately. The supply of chips, servers, power, and data centers remains the binding constraint, not demand. The problem isn't overspending; it's the inability to spend fast enough.

The Scale Problem

Right now, the K and B camps are locked in a noisy brawl, while G adherents watch with bemusement. Part of the disagreement comes from how hard it is to grasp the scale of these businesses.

Amazon's cloud business, AWS, is running at roughly $142 billion in annual revenue, growing 24%. That implies more than $34 billion in incremental revenue over the next year alone. Microsoft Azure and Google Cloud are second and third, but are still huge businesses, backed by the financial firepower of their money-printing parent companies.

Most observers also fail to appreciate the enormous social and economic change underpinning this growth. The world is in the midst of a decades-long shift from analog to digital processes, which has plenty of future growth ahead of it. AI is just the latest addition to the trend, following the rise of the Internet, cloud computing, and enterprise software adoption. AMG are benefitting from a gigantic trend that will continue for years, if not decades. So maybe spending more than $600 billion on capex this year isn't foolhardy.

This isn't 2022

There's an obvious playbook to calm markets. In 2022, after pandemic-era overbuilding spooked investors, AMG executives promised financial discipline, made some cost tweaks, and watched their stocks recover spectacularly over the next two years. They could have done the same this quarter: keep a lid on capex, cite supply-chain constraints, and promise to revisit later. Their stocks likely would have popped on Goldilocks relief.

Obviously, AMG didn't do that. They ramped capex plans to breathtaking new levels. These companies are run by the same executives who lived through the 2022 drawdown. Their compensation is heavily equity-based, so they just took a big hit again. The fact that they chose to increase capex anyway should give observers pause. It suggests confidence based on knowledge that outsiders don't have.

Maybe this really is the Jaws moment. Maybe demand is so strong, and visibility so clear, that AMG know the real risk isn't overspending. It's showing up to hunt the shark without a big enough boat.

Bernard Golden is CEO of Navica, a Silicon Valley-based technology analysis, consulting, and investment firm.