Forget the App. Buy the Pipes
Investors, engineers, and policymakers are converging on the same conclusion: the physical and digital systems beneath AI are the real prize
EDITOR’S NOTE
Forbes published a piece by the CEO of Apex Fintech Solutions arguing that the next decade’s biggest investment themes all share one thing: they are infrastructure stories. Inc. Magazine ran a cover-level argument that the next billion-dollar startups will not build apps, they will build the world’s pipes, grids, and cooling systems. And a new RMI report warned that electric vehicles are already outpacing the chargers needed to power them, threatening to turn a clean transport revolution into a grid capacity crisis.
What ties all of this together is not technology. It is the realisation, spreading from Wall Street to Silicon Valley to energy policy desks, that we spent a decade rewarding software and we are now about to spend a decade rewarding the systems that software depends on to actually function. That shift does not happen gradually. It tends to happen all at once, when the bottlenecks become impossible to ignore. We are at that moment right now.
TOP STORY
The Infrastructure Investment Thesis Just Went Mainstream and That Changes Everything
Bill Capuzzi, CEO of Apex Fintech Solutions, published a piece in Forbes this week that is the clearest single articulation of where serious money is moving: away from software bets and toward the systems that software, AI, and electrification depend on at scale. His argument is that, investors have spent years rewarding breakthrough technologies, and they are increasingly rewarding the physical and digital infrastructure those technologies need to scale. Data centres, power grids, fibre networks, cooling systems, payment rails, and EV charging corridors are no longer background assets. They are the investment. Inc. Magazine simultaneously published a companion argument aimed at entrepreneurs, making the case that the next generation of billion-dollar companies will be the ones building the world’s infrastructure layer, not the ones building products on top of it.
The reason this matters beyond investment portfolios is structural. When the capital allocation logic of the entire market shifts toward infrastructure, it changes what gets built, where, and for whom. Projects that previously struggled to attract private capital because they were too slow, too capital-intensive, or too boring for venture timelines become the most attractive assets in the room. Pension funds, sovereign wealth funds, and infrastructure-focused private equity firms have been making this argument quietly for years. Now the broader market is catching up. The Global Banking and Finance Review confirmed this week that digital infrastructure is being reclassified as a strategic asset across the financial services sector, with AI integration, cloud-native architectures, and interoperable ecosystems becoming the new baseline rather than premium differentiators. The infrastructure era is not coming. It is here, and the investors, entrepreneurs, and governments who recognised it two years ago are already sitting on the most valuable positions in the next economy.
TRENDS TO WATCH
1. EV Adoption Is Outrunning the Grid Built to Support It
RMI published a major planning analysis this week showing that electric vehicles are being deployed significantly faster than the charging infrastructure and grid upgrades needed to power them, with utilities in the United States facing a structural mismatch in which chargers can be installed in months while distribution system upgrades often take years, and EV load appearing quickly and in concentrated geographic hot spots that existing grid models were never designed to anticipate.
If this gap is not closed through better grid planning, demand forecasting, and flexible charging management, it threatens to produce exactly the kind of reliability crisis that undermines public confidence in the energy transition at the moment it needs the most political support.
The EV revolution will not be stopped by a shortage of vehicles or political will — it will be stopped by a grid that was never asked to plan for this.
Read the RMI analysis on planning for an electrified transportation future.
2. Liquid Cooling Is Becoming the Defining Design Choice of the AI Data Centre Era
UNICOM Engineering and Fourier announced this week a modular AI infrastructure solution featuring more than 600 MW of liquid-cooled deployment capacity for enterprise-scale AI data centres, with Futurum Research framing the announcement as evidence that cooling-defined infrastructure, rather than compute-defined infrastructure, is becoming the primary architectural logic of next-generation AI facilities.
The infrastructure implication is significant: as AI workloads grow denser and more energy-intensive, the facilities that can sustain the highest rack densities through advanced liquid cooling will be the ones that attract the most valuable AI tenants, making thermal management a strategic competitive advantage rather than an engineering afterthought.
In the AI data centre of 2030, the most important system in the building will not be the server — it will be the system keeping the server cool enough to run.
3. The Gulf States Are Positioning as the Global South’s AI Infrastructure Bridge
The Observer Research Foundation Middle East published analysis this week identifying the Gulf states alongside India, Brazil, Malaysia, Kenya, and South Africa as the Global South countries emerging as regional AI infrastructure hubs, arguing that countries that invest in AI-ready energy, connectivity, and compute infrastructure now will become the anchor points through which the AI economy reaches the rest of the developing world.
This matters because the alternative: a Global South that consumes AI services built and hosted elsewhere, reproduces the economic dependency that has defined commodity and digital markets for generations, and the window in which infrastructure investment can alter that trajectory is narrowing with every hyperscaler deal that locks in foreign-owned infrastructure.
The countries that build the AI infrastructure their regions need become the nodes that every neighbouring economy has to route through, and that is a form of geopolitical leverage worth building deliberately.
IN OTHER NEWS
1. Marvell Is Betting That Memory and Storage Will Define Agentic AI Performance
Marvell Technology is showcasing its advanced AI memory and storage infrastructure portfolio at the Flash Memory Summit this week, positioning the company as the enabler of the agentic AI era’s most overlooked constraint: not the processing chip, but the memory architecture that determines how fast an AI agent can access and act on the data it needs. This is worth watching closely because the companies that solve AI’s memory bottleneck will become as structurally important to the AI economy as the companies that solved its compute bottleneck.
2. AI Cannot Replace the Electrician and That Is the Most Important Infrastructure Fact of 2026
CTech published a sharp analysis this week making the point that while AI is replacing programmers at scale, it cannot replace the electricians, welders, HVAC technicians, and pipefitters who physically build and maintain the infrastructure that AI runs on, directly reinforcing last week’s Bechtel CEO argument and underlining that the infrastructure workforce shortage is not a temporary labour market fluctuation but a structural constraint on the entire AI buildout. Every week this argument is made by a new credible voice, it becomes harder for policymakers and employers to avoid addressing it seriously.
3. Siemens Is Building the Operating System for Critical Infrastructure Resilience
Siemens published its framework this week for making critical infrastructure systems future-ready through highly resilient architectures designed to withstand cyberattacks, reduce operational downtime, and adapt to evolving demand — framing infrastructure control and operations as a strategic layer that sits above physical assets and determines whether those assets function as designed under stress. As AI drives more complex, interconnected, and interdependent infrastructure systems, the firms that can provide trusted, resilient operating platforms for those systems will become essential partners to every grid operator, water authority, and transport network on the planet.
THE WEEK’S INFRASTRUCTURE NUMBER
600 MW
That is the liquid-cooled AI infrastructure capacity announced by UNICOM Engineering and Fourier in a single modular deployment solution, a number that illustrates how rapidly the scale of individual AI infrastructure investments has escalated — from the megawatt-level deployments of five years ago to the gigawatt-era ambitions of today, with thermal management now the binding constraint on how far and how fast that scale can go.
Read the Futurum Research analysis on cooling-defined AI infrastructure.
INFRASTRUCTURE FINANCE LEARNING
The Forbes piece on infrastructure investing carries one lesson that every infrastructure finance professional should write on their wall: the market is currently mispricing the systems layer. When investors still associate infrastructure with bridges and highways while the real capital is flowing into data centres, cooling systems, payment rails, and EV charging corridors, there is a pricing gap — and pricing gaps are where the most durable returns are made. If you are evaluating infrastructure assets right now, ask whether your valuation methodology accounts for the AI and electrification demand curves, not just the historical revenue profile of the asset class.
TRANSACTION TEARDOWN
The most instructive financial structure this week sits beneath the UNICOM Engineering and Fourier modular liquid-cooled AI infrastructure announcement. While the full deal terms have not been made public, the transaction architecture of a 600 MW modular AI data centre deployment at enterprise scale follows a well-established pattern in the current market worth understanding.
A deployment of this scale is almost certainly structured as a project finance transaction rather than a balance sheet deal, meaning the debt is raised against the cash flows of the specific asset — the contracted revenue from anchor tenants, typically hyperscalers or enterprise AI operators — rather than against the general credit of the developer. Debt-to-equity ratios in this space currently run between 60:40 and 70:30, with lenders accepting higher leverage where there are long-duration power purchase agreements or take-or-pay data centre contracts that provide revenue certainty. The financing instruments likely include a combination of construction finance, which converts to term debt on commissioning, and potentially green bonds or sustainability-linked debt where the liquid cooling technology can demonstrate a measurable reduction in power usage effectiveness relative to conventional air-cooled equivalents. Tenor on infrastructure debt of this type typically runs 15 to 20 years, matched against the useful life of the facility. The key risk the lenders are stress-testing is power supply: a 600 MW facility needs a credible, contracted power source before debt can be committed, and the availability of that power is currently the single largest constraint on closing AI data centre financing across all major markets. What this transaction signals more broadly is that the infrastructure finance market has absorbed the AI data centre as a bankable asset class with well-understood risk parameters, and the competition is now entirely about who can secure the power, the land, and the anchor tenant first.
ON OUR RADAR
Watch for the first major utility filings responding to RMI’s electrified transportation planning framework over the coming months, because the quality and ambition of those grid investment plans will determine whether the United States avoids a charging infrastructure crisis in the late 2020s or spends the next decade retrofitting a grid that was never designed for the EV adoption curve it is now facing.
Benjamin Yaw Manu
Global Infrastructure Analyst / Researcher / Capacity Building Specialist
Infrastructure keynote speaker, conference host, executive briefing at Tomorrow’s Infrastructure
Author of Thriving in Uncertainty, The Business Transformation Blueprint for Resilience, Relevance, and Growth
Have feedback or a topic you would like to see covered? Reply to this message and let me know. Warm regards