Saturday Spotlight: #AIProfitabilityProblem

The AI Boom: A System on the Brink

The artificial intelligence boom is increasingly looking less like a technological revolution and more like a structurally unsustainable system built on circular financing. It resembles a self-consuming ouroboros where “profits” keep recycling through the same entities, while colliding with infrastructure bottlenecks and hard capacity constraints.

The curated AI narrative is increasingly straining against facts and basic arithmetic, and it remains unclear whether broader demand is sufficient to justify the scale of investment.

Circular Funding

For two years, Nvidia has led the aggressive AI investing surge, becoming one of the world’s most valuable companies as Microsoft, Amazon, Google, and Meta committed hundreds of billions to AI infrastructure. Goldman Sachs estimates total spend could hit $7.6 trillion by 2031, with hyperscalers nearing $1 trillion a year in AI capital expenditures.

Most AI compute demand is heavily concentrated among a small number of firms, including OpenAI, Anthropic, Meta, and xAI. Each relies heavily on hyperscaler subsidies, including discounted cloud computing and infrastructure financing.

Microsoft effectively funds OpenAI through Azure, a flexible cloud-computing platform, while Amazon (~$25 billion) and Google (~$40 billion) support Anthropic at scale. Meta is spending up to ~$145 billion on AI despite weak near-term profits, with monetization still limited to a small number of high-usage applications.

This also creates a closed loop where money flows between investors, AI startups, cloud providers, and Nvidia. Intercompany sales amplify the demand while the underlying end-user adoption and revenue are significantly lower.

OpenAI is burning billions annually and faces a possible 2027 bankruptcy, Anthropic remains deeply unprofitable, and much of Microsoft’s and Amazon’s “AI growth” is internally subsidized rather than demand driven, while AI tools and computing are getting cheaper, making profitability harder.

When profit motives align, they can drive this warped accounting spiral, producing a self-reinforcing system that sidesteps operational and market realities. Yet the deeper issue is not financial structure, but failed arithmetic.

Infrastructure Limits

Since 2023, companies have announced mega-AI data center projects requiring gigawatt-scale electricity, vast cooling systems, and major transmission upgrades, but many remain unbuilt or nonoperational, face local opposition, and depend on subsidies and permitting timelines that move far slower than capital deployment.

Even if these facilities are completed, they face three unavoidable issues:

• Chip obsolescence – outpaces deployment. Nvidia’s rapid cycles mean graphics processing units (GPU chips) will likely be outdated before data center deployment. If infrastructure buildout is slower than chip turnover, this will tie up capital in quickly depreciating chips.

• Resource consumption – exceeds local capacity. AI data centers require city-scale amounts of electricity and massive water supplies, straining already stressed grids and competing with other users. This demand grows faster than data centers and grid capacity can expand.

• Infrastructure – is inadequate. The electrical grid is structurally unprepared for projected demand and increasingly vulnerable to electromagnetic disturbances.

Grid Threats

The US grid is already far behind pre-AI-demand forecasts, with multiyear interconnection queues, transformer shortages (lead times as long as four years), and transmission bottlenecks. New capacity takes decades to permit and build — far slower than AI’s investor-projected growth cycle.

Also, according to the Harvard-Smithsonian Center for Astrophysics and the US Magnetotelluric Array, geomagnetic storms can induce destruction across already stressed power grids — the same infrastructure AI depends on — and that vulnerability is increasing as grid complexity and load rise.

The 1989 Quebec blackout is an example of how these events can overwhelm grids and destroy transformers at scale.

As AI-linked data centers further integrate with the grid, a severe solar storm could cascade across linked power, satellites, communications, finance, logistics, and cloud systems, creating a far more destabilizing disruption than a conventional outage.

Expansion Paradox

Put all together, the arithmetic is stark:

• GPU deployment is faster than data centers can be built.
• Data center demand is growing faster than the power grid.
• Power and water needs exceed local infrastructure supply.
• Chips become obsolete faster than facilities can fully use them.
• Revenue growth is not keeping up with the scale of expansion.

That means even before financial returns are considered, the entire system is already pushed to its limits under basic physical constraints.

Market Detachment

Despite this reality, markets are exhibiting the same willful myopia that sparked the 2008 financial crisis, once again ignoring fundamentals while pricing AI as if frictionless growth were guaranteed.

Nvidia keeps posting record results, fueled in part by hyperscaler-backed internal spending. OpenAI and Anthropic continue raising tens of billions. Pension funds, insurers, BlackRock, and private credit are being increasingly exposed to AI infrastructure debt, while banks are trying to offload AI risk at discounts.

While AI adoption is real and shows growth in specific sectors, monetization is still small relative to infrastructure spending.

Advocates draw false parallels to railroads and the early internet: steel rails and broadband last decades, while AI hardware becomes obsolete in months, creating a nonviable infrastructure equation.

Deployment Dilemma

The bearish conclusion is not just that AI may be financially overvalued — it is that the current AI deployment model is on course to fail basic monetization, ROI, and adoption viability tests.

When you combine accelerating chip obsolescence, limited local resources, and a constrained electrical grid, it’s not just an economic bubble — it’s a deployment model that struggles to exist at the scale being priced in.

Hashtag Picks

Tiny Data Centers May Be Coming Into the Homes of Americans in the Future

From CNBC: “Data centers are gobbling up land, driving up electric bills, and becoming a lightning rod for public discontent over big tech’s power in society. … At the same time, the idea of putting data centers closer to consumers, even onto and into their homes, is gaining traction in real estate circles. Major players in housing, including homebuilder PulteGroup, are in early testing with Nvidia and California-based startup Span to install small fractional data center ‘nodes’ on the exterior walls of newly built homes.”

The $670 Billion Question: Is AI Demand Real, or Are We Building on Subsidized Sand?

The author writes, “Microsoft, Meta, Amazon, and Alphabet are planning to spend up to $670 billion on AI infrastructure in 2026. A WSJ comparison chart puts that figure at roughly 2.1% of U.S. GDP (based on their estimate of nominal GDP). For context, U.S. railroad investment in the 1850s averaged roughly 1.7% of GDP, with a peak around 2.6% in 1854. The scale is comparable to the largest infrastructure booms in American history.”

Inside the Dirty, Dystopian World of AI Data Centers

From The Atlantic: “The race to power AI is already remaking the physical world.”

Why AI Is Slowing Down in 2026

The author writes, “A curious gap has emerged between the breathless predictions of artificial intelligence enthusiasts and the reality unfolding on the ground. The technology itself continues advancing at a remarkable pace. The research papers keep coming. The models keep improving. Yet something is holding back the full realization of AI’s potential, and the reasons have almost nothing to do with the debates dominating public discourse. The bottlenecks are physical, structural, and in some cases almost absurdly mundane. Understanding where the friction actually lies reveals a very different picture of the AI landscape than what most commentators present.”

Nvidia Faces Rising Competition as AI Chip Funding Surges

The author writes, “Nvidia still sits at the center of the AI boom, but the money is starting to flow toward companies trying to challenge that position. Investors have poured roughly $8.3 billion into AI chip startups this year, a sign that the market is widening beyond just one dominant player. The shift is not really about replacing Nvidia overnight. It is more about what comes next.”

Scientists Found AI’s Fatal Flaw — The Most Advanced Models Are Failing Basic Logic Tests

The author writes, “In a new paper that’s making waves, scientists from Stanford, Cal Tech, and Carleton College have combined existing research with new ideas to look at the reasoning failures of large language models (LLMs) like ChatGPT and Claude. Those who rely on LLMs for intellectual labor often cite the models’ reasoning ability as a major draw, despite the evidence that this ability is limited, even when dealing with simple questions. So, what’s the truth of the matter?”

Financing the AI Boom: From Cash Flows To Debt

From the Bank for International Settlements: “Rapid advances in artificial intelligence (AI) appear set to reshape economies, industries and financial markets and AI firms have been a major driver of equity market developments over the past year. Yet AI-related innovations demand not only groundbreaking research but also substantial investment in infrastructure. At the heart of this transformation lies a surge in capital expenditures to build the physical infrastructure for AI, e.g. data centres, and related technological infrastructure such as computer servers, networking hardware, cooling systems, grid connections and power stations. These investments are key in supporting the enormous demand for computational resources and data storage facilities to train and operate AI models. The need to finance these investments is causing a shift in sourcing the financing from cash flows to debt.”

Saturday Hashtag: #AIProfitabilityProblem originally appeared on CryptoLiveDaily