Highlights:
- The key constraint on AI growth is now physical infrastructure, including power, cooling, grid, and data centre capacity.
- Investors should broaden their AI investment playbook beyond chips and towards electrical equipment, cooling systems, utilities, and independent power producers.
- Hyperscalers that are able to self-fund large-scale infrastructure and energy projects are likely to hold a growing advantage.
- The AI boom is creating a barbell economy, where cash-rich technology giants continue expanding despite high interest rates, while the broader economy absorbs rising infrastructure and energy costs.
As the May 2026 earnings season draws to a close, a new reality is becoming harder to ignore: the bottleneck for artificial intelligence (AI) is no longer the chip. It is the grid.
The focus is shifting from who can build the smartest AI model to who can power and cool it. The numbers are difficult to ignore: generative AI queries consume significantly more electricity than traditional internet searches. The Magnificent Seven are expected to spend more than $650 billion in capital expenditure this year alone, with over $500 billion earmarked exclusively for AI infrastructure.
The message from the market is becoming clearer: we have officially moved past the chip era of this bull market and entered the infrastructure era.
The Thirst for Cooling, Power, and Grid
In major AI hubs such as Northern Virginia and Dublin, the sheer concentration of AI workloads is pushing local grids to their limits. That pressure is driving increasingly unconventional solutions for players around the world.
In late April 2026, Meta signed an agreement to explore space-based solar energy for future data centre operations, reserving access to 1 gigawatt of potential orbital solar power generation. Across Europe, large-scale AI campuses — such as the Pantheon project — are increasingly pairing data centres with dedicated renewable energy and battery storage to reduce reliance on public grids.
The scale of the challenge is substantial. US data centre electricity demand is projected to rise so sharply that the country could face a 49 GW power shortfall by 2028. At the same time, oil prices remain elevated, with WTI hovering near $100 a barrel, adding further pressure to electricity and industrial costs.
For investors, this is broadening the AI trade beyond semiconductors. Utilities, grid equipment manufacturers, cooling technology providers, copper producers, and independent power companies are increasingly being viewed as second-order beneficiaries of AI infrastructure demand.
Energy reliability is no longer just an operational issue. It is becoming a strategic advantage.
Why High Interest Rates Aren't Stopping Big Tech
Central banks are starting to feel the inflationary shockwave of the AI buildout.
Paradoxically, AI may prove deflationary for white-collar labour as automation improves productivity and reduces operating costs. Yet at the same time, it is becoming inflationary for the physical economy powering it. The more AI expands, the greater the demand for copper, cooling systems, power equipment, construction materials, and grid upgrades.
Under normal market conditions, high interest rates would slow large-scale construction and infrastructure investment. This cycle may be different. Despite elevated global borrowing costs, recent developments show just how aggressively hyperscalers are pushing ahead. Alphabet recently tapped the euro bond market after raising billions earlier this year to support AI expansion, while Meta and Microsoft have continued lifting capital expenditure forecasts as competition for compute capacity intensifies.
The result is what increasingly resembles a barbell economy. On one side are cash-rich technology giants capable of funding trillion-dollar infrastructure projects regardless of interest rates. On the other are smaller businesses and rate-sensitive sectors facing tighter capital, higher energy costs, and rising financing pressure.
This leaves central banks in a difficult position. Keeping rates elevated to contain AI-driven inflation risks slowing the broader economy. Cutting rates too early, however, could fuel even greater excess in a sector already consuming enormous amounts of capital, energy, and industrial resources.
For now, Big Tech appears willing to spend through the cycle. The question is no longer whether the AI buildout continues, but whether global infrastructure can keep pace with it.
The Outlook: From Silicon to Steel
As investors look towards the second half of 2026, attention may increasingly shift away from the screen and towards the substation.
The next winners are likely companies with control over supply, power, cooling, and long-term infrastructure capacity. For example, firms such as Alphabet and Microsoft have the scale, cash reserves, and operational reach to secure long-term energy solutions and continue expanding even in a high-rate environment.
In the next stage of the AI boom, the market has simply stopped asking "what can AI do?" and started asking "how much does it cost?" The answer to that question will define the next decade of global finance.
By Ross Maxwell, Global Strategy Operations Lead, VT Markets