If 2024 was the year of hype and 2025 the year of building, then 2026 will be remembered for something far more brutal: the year of accountability.
For the past two years, markets rewarded spending. Say “AI,” “GPU clusters,” or “data centre expansion,” and Wall Street handed out premiums. Companies weren’t judged on profitability; they were judged on how fast they could deploy capital.
The defining question then was simple: “How many GPUs did you buy?”
But after last month’s rate cut, a harder question has taken over:
“How much profit did those GPUs generate?”
While investors celebrate easier money, the “capex cliff” has quietly arrived.
Look at Oracle, Broadcom, and Astera Labs: just weeks before 2026, all three reported record earnings driven by AI, yet their stock prices got hammered by double digits immediately after. Not because AI is over, but because they delivered growth while exposing the cost. Margins shrank. Expenses surged. And for the first time, the market actually cared.
Over the next 12-18 months, analysts and boards will demand proof of monetization. That pressure will reshape where capital flows, and the winners won’t be the ones who spent the most… they’ll be the ones who can show what all that spending actually bought them.
Here’s where to look for…
Power is the new moat
Phase one of the AI boom rewarded chipmakers and semiconductor supply chains. But in 2026, the bottleneck shifts from silicon to electricity. If you read my last article about the triggers that could pop the AI bubble, you’ll remember the power bottleneck.
AI workloads consume staggering amounts of energy, and here’s the thing: power cannot be coded, downloaded, or compressed. It is the only constraint that money cannot instantly fix. This creates a new form of pricing power.
Utilities are no longer selling electricity; they’re selling certainty. Hyperscalers like Microsoft, Google, and Amazon don’t just want power… They are signing private, premium contracts for baseload capacity – energy that never disconnects. In the world of AI infrastructure, utilities start looking less like boring “safe income stocks” and more like digital landlords with monopoly corridors into the AI economy.
The progression is clear: 2023 was about GPUs (chips), 2024-2025 was about data centres (structure), and 2026 is about kilowatts (supply).
The M&A supercycle
In a world moving at AI speed, the real competitive asset is time. Companies are acquiring not from confidence, but from urgency. Why spend three years building when you can buy it on Monday?
Waiting has become the new risk.
The Netflix–Warner Bros deal wasn’t just a transaction, it was the signal. We already saw this urgency play out in 2024, when companies were so desperate they bought even with interest rates at 5%. AMD bought ZT Systems for $4.9 billion because they couldn’t wait three years to build server infrastructure. J&J bought Shockwave Medical for $13 billion to instantly plug a patent cliff. And to top it off, Meta recently bought Manus for over $2 billion, proving they’d rather write a big check than wait years to build their own autonomous agents.
If they were buying when money was expensive, imagine the deal flow now that money is cheaper. With rates settling near 3.5%, the “wait-and-see” crowd may finally move. Their prime targets:
- Mid-caps with irreplaceable IP
- Platforms with proprietary data
- Biotech firms with validated pipelines
- Industrial firms with network effects but no scale
2026 might be the year the giants go shopping; make sure you own what they want to buy.
Cash flow as judge and executioner
For over a decade, the market operated under a quiet luxury: capital had no cost.
That era is gone.
Yes, the Fed has started cutting rates, but make no mistake… we’re entering a 3-4% world, and we’re probably not returning to zero. Possibly ever. In this regime, the market is no longer chasing growth for its own sake. It’s evaluating who survives without external funding.
High growth with negative margins? No longer tolerated. Moderate growth with strong free cash flow? That’s the new leadership class. The question has changed from “How big can you become?” to “Can you fund the journey yourself?”
Growth is still rewarded, but for the first time in a long time, profit is required. So in 2026, look for these three archetypes:
- Toll booths: These companies take a cut of trillions of dollars in volume but don’t need to build factories or data centres to grow.
- Franchise kings: They own the brand; someone else pays for the building. This is the ultimate inflation hedge because they collect a percentage of top-line revenue while the franchisee pays for the operations.
- Niche monopolies: Boring companies that dominate a specific, essential corner of the economy. They can raise prices without losing customers because there is no alternative.
The industrialization of intelligence
Software dominated the last decade, but AI is now escaping the screen and entering the physical economy: robotics, autonomous factories, machine vision, smart logistics, automated warehousing, edge and industrial compute.
This isn’t about replacing white-collar workflows. It’s about rebuilding the supply chain, manufacturing base, and logistics backbone using intelligent systems. This is where AI stops being theoretical and becomes a robot arm, a forklift, a scanner, a factory brain.
The winners will be businesses sitting at the intersection of:
- Hardware + compute
- Automation + AI
- Real-world scale + optimisation
This is the next leg of the productivity cycle, and it’s happening in warehouses and factory floors, not on your laptop screen.
Healthcare: The spread too wide to ignore
For years, the argument against Big Pharma was simple: drug discovery is too slow, too expensive, and fails too often. But with AI compressing discovery timelines, improving trial precision, and accelerating regulatory data, the catalyst is finally here.
While the market obsessed over tech in 2024-25, healthcare quietly lagged. That lag is now your opportunity. Consider the valuation reality: Big Tech trades at 25-30x earnings, priced for perfection. Big Pharma trades at 12-15x earnings, priced for stability. That spread isn’t normal – it’s a margin of safety.
In 2026, we’re moving from “AI hype” to “clinical proof”:
- AstraZeneca is now using AI to simulate “placebo arms” in trials, potentially cutting the number of human patients needed by 30-50%.
- Pfizer reported compressing complex molecule computations from four months down to just days using supercomputing and AI.
You’re looking at a sector with guaranteed demand (aging population), trading at a discount, with its biggest cost bottleneck (R&D time) collapsing. Tech is priced as if nothing will go wrong. Healthcare is priced as if nothing will take off.
And the market isn’t paying attention yet…
The fifth perspective
2026 might be the year to align with operational reality: Premium baseload energy and nuclear utilities. Industrial automation and supply-chain intelligence. Businesses that compound free cash flow with durable moats and strong returns on capital. Dividend-growth equities becoming relevant again in a higher-yield environment. Mid-caps primed for acquisition bids. Healthcare opportunities shaped by valuation gaps and AI adoption.
The question isn’t what’s exciting?
It’s what survives the capex cliff and converts investment into return?
For almost two years, markets priced hope. Now they’ll price proof. The gold rush is ending and the companies that survive won’t be the ones who dug the fastest, but the ones who actually found gold.
Revenue is vanity. Profit is sanity. Cash is reality.
2026 demands accountability.
