Why the AI Capex Cycle Is a Tailwind for UK Grid and Waste Operators

The first move in 2026 was the AI derating of UK software. Sage has lost roughly 20% of its market value year-to-date; Rightmove is down more than 10%. The market is making a perfectly defensible bet that some unknowable share of seat-based SaaS revenue will be cannibalised by AI-native competitors over the next five years, and is repricing the multiple now to reflect that uncertainty.

What the market has been slower to do is reprice the other side of the same trade. The AI capex cycle is one of the largest non-cyclical infrastructure stories of the decade — IDC projects $758 billion of cumulative AI infrastructure spend through 2029 — and it is fully dependent on the electricity grid that delivers power to the data centres and the physical infrastructure that supports their operating footprint. Those businesses are owned by companies the market has, so far, valued as if they were going to be replaced by AI rather than fed by it.

That gap is the trade.

Grid Operators as the Pure-Play

UK data centres are currently estimated at around 6% of total electricity demand. The National Grid ESO base case puts that at up to 9% by 2035 — a 2.6 percentage-point rise that has to be physically delivered through copper, transformers and substations that take 18 to 24 months to commission. There is no software solution to the lead time.

[National Grid](/equity/NG) responded to this in 2024 with a £35 billion capital programme to upgrade UK transmission infrastructure ahead of the demand curve. The Q2 2026 update reported a record £5 billion deployed in the first half of the 2025/26 financial year — running ahead of the original £11 billion full-year run rate. The company has been a direct beneficiary of UK industrial-strategy alignment: the AI Energy Council has earmarked a further £30 billion of investment for designated AI growth zones in the North-East of England, with the Uxbridge Moor expansion targeted at completion by 2029 to support more than 12 new colocated data centres.

Beyond the capital programme, the partnership announced in late 2025 with Emerald AI is the more interesting structural development. Rather than building grid capacity ahead of demand, the joint approach allows hyperscaler customers to modulate consumption against grid availability in close to real time, materially improving asset utilisation on existing infrastructure. If that model scales, [National Grid](/equity/NG) captures a share of what is functionally a flexibility-trading revenue stream layered on top of regulated returns.

The underlying numbers support the case. The Openbook scoring framework reads [National Grid](/equity/NG) at an overall reward rating of 62, with 36.7% twelve-month total return and a 15.8% operating margin — modest growth, but evidence of disciplined execution against a substantially expanded capital base. CEO John Pettigrew has framed the capex programme as ultimately consumer-cost-reducing, which is the regulatory framing required to keep the cost-of-capital low through the cycle.

Waste Management: Contractually Insulated

Waste collection, transport and processing is a thinner story but a more defensible one. The core economic activity — moving physical material through a permitted network of trucks, transfer stations and disposal sites — is not addressable by software. Demand is mandated by law, customer concentration is favourable (local authorities sign 7-to-15 year contracts), and the regulatory direction of travel on landfill, recycling targets and circular-economy compliance has been a tailwind for the operators positioned to invest into it.

Veolia is the consolidator we would highlight inside that frame. In February 2026 the group secured a 15-year contract to operate two large water treatment plants in Mumbai expected to supply approximately 60% of the city's water needs at full ramp — a contract size that materially shifts the international revenue mix. Within the UK, Veolia now holds over £1 billion of active local-authority contracts across more than 50 councils, and has invested more than £1 billion into the UK over the last decade, including the largest UK plastic recycling facility and the world's first commercial-scale vehicle-to-grid trial returning battery energy from waste-collection fleets to the grid during peak periods. That last project is small in revenue terms but strategically interesting — it represents an operationalised intersection between two of the most heavily-funded infrastructure programmes of the decade.

Renewi and Biffa offer similar contracted-revenue exposure on the listed UK side, though without the international optionality.

The Speculative Bracket: Waste-to-Energy

There is a higher-risk subset of the waste sector that converts non-recyclable waste streams into hydrogen and other low-carbon fuels. [Powerhouse Energy](/equity/PHE) is the longest-standing UK-listed name in the category. Hydrogen Utopia International signed a memorandum of understanding with Hydrogen Systems LLC, a Saudi hydrogen infrastructure developer, at the start of 2026. Under the structure of the MoU, Hydrogen Systems provides EPC services and operates facilities converting plastics and waste streams into hydrogen, with sustainable aviation fuel as the targeted off-take. The sustainable aviation fuel market is projected to compound at roughly a 40% CAGR over the next decade, which gives the upside case oxygen.

The honest framing is that this is venture-scale risk in listed-equity clothing. None of these companies generate sustainable earnings yet. The investment case rests on whether the regulatory mandate on aviation fuel decarbonisation in the EU and UK translates into commercially viable off-take pricing in the second half of the decade. We would treat exposure here as separate from the defensive core thesis, sized accordingly.

Our View

The cleanest formulation of the trade is this: if the market is right that AI compresses software margins, it has to be roughly equally right that AI expands the regulated infrastructure that AI runs on. Both sides of that statement cannot simultaneously be priced into UK equities at current relative valuations.

We see the regulated utility and waste operators as the lower-volatility expression of the AI infrastructure thesis. They will not deliver the asymmetric upside of the hyperscalers, the picks-and-shovels semiconductor names or the speculative SMR developers. What they will deliver is contracted revenue growth attached to a non-cyclical capital programme, paid for out of regulated returns, in companies the market is currently valuing without giving the AI tailwind any credit at all.

That mismatch is the opportunity.

Educational information only. This is not personal financial advice.