O'Leary Filter: Grid Access Decides Which Campuses Survive

A Builder Amid the Selloff

AI shares just suffered a brutal week. MarketBeat invited Kevin O'Leary to discuss the fallout. O'Leary became famous through Shark Tank and started his career in software. Today he backs giant data center campuses in Utah and Alberta.

Two Verticals Thesis

O'Leary reaches for an internet analogy. In the early web era everyone counted eyeballs. Infrastructure owners captured the durable gains. He sees two verticals now: the power grid plus processing and storage capacity. Those form the theme for the next five years.

Cloud and storage already run a proven setup. AWS, Google and Microsoft keep expanding. The giant hyperscalers keep adding capacity. Much of that demand sits outside AI. Phones, corporate records and telecom networks all need space. So the investment case never rested on a single product.

His message to shaken retail holders stays calm. Demand runs near three times supply. He sees no slowdown across the next three years. Grids need more capacity while gas keeps converting into electricity. Tenants hunt for low-latency capacity near fiber. For equity investors that setup signals patience rather than panic.

Data Centers Through a Landlord Lens

O'Leary treats data centers as property deals. The physical shell plus power defines the asset. The theme fills 20 to 30 percent of his book. He spreads positions across four countries. The rule stays simple: build where energy costs least. Saudi kilowatt-hours cost around one cent, yet raising funds there proves difficult today. Cheap power alone never closes a deal.

Budgets run enormous. One large campus costs near $16B before opening. One gigawatt of AI processing needs roughly $15B. Demand approaches 45 GW while construction covers near 5 GW. No spare grid capacity waits nearby. Developers must therefore arrive with their own electricity. Stranded gas feeding on-site turbines leads the workaround. Nuclear could help, though nuclear needs around 15 years to matter.

Why Most Campuses Never Break Ground

O'Leary claims 60 to 65 percent of announced projects will never rise. He frames the figure as his own estimate, not measured fact. His list of flaws runs long: missing grid power, stalled permits, scarce water. Local resistance and equipment bottlenecks add pressure. DataCenterDynamics carried his warning to a wide audience. AInvest reports nearly half of planned projects stalling. He argues most groups grabbed land without grasping the requirements. For stock pickers that screen comes first.

Outside numbers point the same way. CNN cited a JPMorgan calculation that around 60 percent of 2027 capacity never started construction. Berkeley Lab found over 70 percent of grid connection requests withdrawn. Goldman Sachs expects power demand to climb 165 percent by 2030. McKinsey sees $5.2T of required funding by 2030. Figures confirm the direction without matching exactly. A single source never settles an investment decision.

Ballot calendars sharpen the conflict. Permits turned into political footballs across many states. Rising household electricity bills fuel resentment. Local approval risk keeps climbing. O'Leary also invokes the race against China. Falling behind, he argues, carries strategic costs.

His liquid wager is Bitzero, ticker $AIBZ on Nasdaq. The company operates across Finland and Norway. Sub-six-cent hydropower underpins the pitch. Land, permits and water arrive as one package. CoinDesk reported the firm reserving over 13,000 acres for AI and crypto infrastructure. Transparency note: this interview was produced with Bitzero sponsorship, so company claims deserve a skeptical read. Treat single-source narratives with caution.

A Screen for Winners

Retail investors get a four-question screen. Is grid connection guaranteed. Is the equipment schedule firm. Are cooling and water solved. Is local approval secured. MarketWatch describes this mindset as a picks-and-shovels approach. Backing infrastructure beats betting on AI software alone. The question is never which ticker first, but which foundation stands ready.

The long horizon still glows. Corporate profits keep setting records. Productivity gains run 10 to 20 percent. All 11 industries now deploy AI tools. O'Leary expects the wave to continue. That continuation forms the core equity thesis.

He reserves a dedicated slice of his portfolio for the theme. Listed companies supply liquidity. Private projects lock funds for years. Balancing both forms his method. Spreading positions divides risk. This content is not investment advice; consult a qualified professional before deciding.