📈  Chart of the Week 2/1/2026

By Michael Allison, CFA

Navigating AI Risk: Returns and Real Growth

I ran across a very interesting piece from The Carlyle Group entitled, “Bubbles as a Feature, Not a Bug” and I thought it was worth discussing.

As markets grapple with the scale and speed of artificial intelligence adoption, the real economic story is often obscured by headline-chasing narratives about bubbles and “irrational” valuations.

But from a strategic investment and macro lens, the Carlyle piece offers a compelling reframing: what looks like a price bubble may instead be the early pricing of a deep structural shift in economic productivity, a shift that ultimately routes value far downstream from the hardware and frontier infrastructure that has historically been bid up in early cycles.

1) Future Investment Returns: Where Value Actually Accrues

The historical analogies in the piece — from electrification to the internet — are instructive precisely because they show how broad, diffuse returns emerged over long term horizons:

• In the electrification era of the early 20th century, early electrification hardware stocks soared, then corrected. The much richer, sustained returns accrued to firms that harnessed electricity to reshape production and consumption. Durable goods manufacturers, for instance, delivered double-digit, decade-plus returns after broader infrastructure prices peaked.

• Similarly, the 1990’s internet boom’s real economic payoff was downstream, in efficiencies unlocked across logistics, retail distribution, and network effects that traditional valuation frameworks initially missed.

Applied to AI, this suggests that the near-term froth in frontier compute and platform equities is less an outright bubble than the market’s early attempt to price an expansive reshape of enterprise economics. Strong returns could actually come from:

• incumbents who integrate AI into core workflows and unlock persistent margin expansion,

• firms that leverage AI to restructure supply chains and data-enabled business models,

• new categories of services that monetize the attention, automation, and prediction layers above base compute.

From an investment perspective, one theme is pretty clear: half-life matters. Early hardware and semiconductor returns may be fleeting, but the more durable returns could come from structural adopters and business model reinventers, the downstream beneficiaries that capture long term productivity gains.

2) Productivity-Led Growth: Beyond the “Bubble” Narrative

Most productivity narratives around AI focus on automation lifting output or lowering labor costs. But the Carlyle analysis points to a broader lens on productivity:

• AI’s largest impact is not just efficiency but the redefinition of workflows, supply chains, and decision-making itself, the equivalent of factories moving from steam to electrified lines, or retail moving from bricks to connected digital ecosystems.

• Even when AI doesn’t directly transform results (e.g., basic data center demand), it catalyzes productivity improvements, accelerating digitization projects, data capture initiatives, and analytic infrastructures that firms had long postponed.

• Productivity growth isn’t uniform; it diffuses through sectors over years. That’s the pattern seen in electrification and early internet adoption: initial investment surges unlock a cascade of follow-on gains across manufacturing, distribution, and consumption platforms.

So perhaps the smarter way to think about AI’s economic lift for the economy is not as a capex supercycle, but as a multi-stage productivity wave: first infrastructure, then workflow transformation, then ecosystem expansion. The result is latent GDP growth driven as much by organizational redesign as by the underlying technology itself.

Bottom Line

I don’t view AI as a binary bubble or bust story. Like electricity and the internet before it, AI’s economic value will migrate downstream, creating years of compound returns and productivity expansion that only become obvious in hindsight. The near-term speculative froth reflects the market’s attempt to price innovation, but tomorrow’s durable winners will be those capturing the enhanced economic output unlocked by AI-driven productivity, not simply owning the frontier compute stack.

Source: Carlyle: Bubbles as a Feature Not a Bug (Navigating AI Risk)

Interested in reading more of Mike's weekly newsletters? Click below to view The Sunday Drive.

The Sunday Drive