Companies that build the most capable AI models will run out of electricity, cooling capacity, and the chips and cables that connect them in 2026. According to Bridgewater, giants like Alphabet, Amazon, Meta, and Microsoft are set to do so. Investing a total of $650 billion in AI-related capital spending in 2026 (up from $410 billion in 2025). These are commitments on a scale that only makes sense if you believe that the physical layer is where competitive advantage is actually built.
Most founders and investors are still optimizing the layer above. They build products that rely entirely on infrastructure they don’t own, don’t control, and can’t take for granted. This is a structural weakness.
So where does the value go from here? Who will end up controlling what the AI economy actually runs on? The answers are starting to come together.
Could infrastructure in 2026 become a bubble?
Every time capital moves quickly in a certain direction, someone calls it a bubble. We’ve seen this before with dotcoms, with cryptocurrencies, and, most recently, with a wave of AI startups that were little more than a thin layer of code layered on top of ChatGPT.
Today, the word “infrastructure” includes the GPU clusters, fiber networks, and power grids that feed modern data centers. This is a big leap from what it once meant: roads, bridges and pipelines. When a word starts to include many different origins at once, the bubble question becomes inevitable.
A bubble, in essence, is a demand that does not yet exist. Capital is poured into something that may or may not be achieved. As for infrastructure, the reality is the opposite. You can’t train a modern AI model without massive amounts of electricity, store sensitive data securely without a defined network, and run more computing without chips. These are the hard physical limits that define what is actually possible today. From Texas to Northern Virginia, data center expansion is already running up against local power constraints and grid capacity debates. Even the supply of advanced GPUs — dominated by companies like Nvidia — has become a strategic constraint.
Naturally, large capital will respond to this. PitchBook data appears Infrastructure investment rose by 44 percent Year after year. Moreover, it is a completely different type of investment compared to what fueled the AI boom. The infrastructure is bound by strict physical constraints, and demand cannot be manufactured arbitrarily.
Driving factors behind the infrastructure boom
If capital is flowing into infrastructure on this scale, there must be powerful forces driving it. Geopolitics is one of the most important.
Governments around the world are withdrawing from central clouds over which they have no control. If your most sensitive data resides on servers owned by a foreign company and subject to foreign law, true sovereignty is in question. This realization has helped accelerate what is now called “sovereign AI” — the idea that a nation’s AI capabilities must run on existing infrastructure within its borders. Building this infrastructure requires massive investments, which is why geopolitics has become a major driver of current prosperity.
Then comes the cost of capital. When interest rates rise and uncertainty rises, investors become more selective and prefer assets with tangible value. The infrastructure fits this profile. It is tied to physical capacity and contracted demand, which are assets that exist regardless of market sentiment. Compared to supporting a startup that may pivot several times before finding its footing, infrastructure can seem remarkably stable.
Energy completes the picture. A modern AI model can be trained You need approximately ten times More power than traditional computing workloads, and the appetite continues to grow. Network access has quietly become one of the most controversial benefits in technology.
This dynamic is already evident from major financial movements. For example, BlackRock Launched $100 billion A fund specifically dedicated to AI energy infrastructure. The logic is clear and straightforward: who controls the energy supply ultimately influences how quickly the AI economy expands.
Why the app era has run out of the way
Infrastructure did not suddenly become attractive overnight. During the app era, startups built an app, acquired users, grew quickly and subsequently handled margins. The infrastructure was already in place as well. Cloud computing was relatively cheap, smartphones were ubiquitous, and capital was patient enough to wait for profitability. In other words, these conditions were almost ideal for application layer bets.
But markets eventually become saturated. Suddenly, a new food delivery app couldn’t solve new problems, and could only fight for leftovers in a crowded space where profit margins were shrinking, and attracting new users was expensive.
At the same time, AI and Web3 introduced a level of computational and architectural complexity that legacy mainframe architectures were not designed to handle. After all, you can’t run a parametric model on infrastructure designed for a passenger transportation application. That was the real turning point, when infrastructure became the most attractive bet.
The next decade belongs to the builders below
In the age of apps, the winners controlled the distribution. Google-owned search, Apple-owned phones, and Amazon have dominated the online storefront. Everyone built on top and paid for access. Now the gatekeepers move down the layer. Computing, power, and connectivity are becoming critical bottlenecks because every AI product ultimately depends on them.
This shift changes the shape of a winning company. In the infrastructure cycle, success becomes defined by the ability to deliver real power: more compute, lower costs, and better control of data. Buyers ultimately pay for reliability and operational control.
For investors, the picture favors absorptive capacity. Infrastructure and networks that are expensive to replicate, difficult to replace, and have already been pulled down by real demand, tend to persist.
The first opportunities appear in two areas. One is decentralized physical infrastructure networks (DePIN), where computing and connectivity are distributed outside the largest cloud providers. The other involves hybrid operators, that is, teams that own physical hardware while also running the software stack required to keep those machines productive.
This is not a quick-building business, and that’s precisely the point. Institutional capital is often more comfortable investing in sectors with long timelines, operational complexity, and scarce physical assets. In the AI economy, value is likely to be less concentrated in the stack, as each application ultimately relies on reliable compute, power, and connectivity.
