New railroad Barons of AI would Own The Track and Tax  The Traffic

BlackRock, Nvidia, Microsoft, and Elon Musk’s xAI are buying a mutual data center for $40 billion, shows that data centers have turned from just a tech tool into a money-making asset, by controlling massive data centers that power AI systems, these companies are setting themselves up to earn profit from every company that needs computing power to run its AI. 

This deal marks a significant shift from AI being about algorithms and breakthroughs to being about who owns the servers. The inclusion of a finance management company, Black Rock, signals the financialization of AI. The AI data centers are now being treated like real estate or oil field assets.  

While Microsoft and Nvidia already dominate software and hardware, they now extend their ownership to infrastructure as well. It seems like they’re soon gonna introduce a new “rent economy” where every company chasing their AI pursuit would have to borrow the computing power from these tech giants. 

Expect A Tool Tax

By controlling the servers that power everything from OpenAI to Google, BlackRock and its partners (Nvidia, Microsoft, and xAI) are turning computing power into a tool road.  It’s a modern version of railroad barons owning the tracks; no matter who wins this race of AI, the house always gets paid. 

This is capitalism’s finest form for AI, owning the pipes, not the product. The role breakdown is also pretty clear here: BlackRock turns computation into an investment vehicle, Nvidia sells the shovels, Microsoft provides the cloud, and Musk’s xAI drives the demand. 

Instead of open innovation (which was always the promise of AI) we are heading towards an arrangement of computer monopolies where access to intelligence would depend on who can afford to pay the tool tax. 

AI’s Power Problem

Running 5 gigawatts of data centers takes as much power as several nuclear plants. BlackRock’s AI expansion means local communities will face higher energy demand and environmental pressure, while profits go to distant investors. 

They’re also smart enough to build these centers in cheaper places with looser regulations, such as Texas, that has a reputation of business-friendly regulations to avoid rising costs but leave pollution, grid strain, and rising local bills behind without any accountability. 

That’s another example of the dark side of AI’s infrastructure boom, where profit is privatized and impact is socialized. Energy-hungry AI data centers that promise “innovation” would consume resources that could power millions of homes.  

Competitive Foreclosure

By joining forces to buy massive data centers, Nvidia and Microsoft gain special access to computing powers that their rivals also need. This means smaller or independent AI companies could struggle to find enough capacity, not owing to a worse technology, but because the competitors now own the servers. 

Hence, we see that OpenAI’s is building separate partnerships with Nvidia and AMD, plus Meta and CoreWeave are also building proprietary facilities. This represents defensive responses in recognition that relying on competitor-owned infrastructure surrenders strategic autonomy.

In this set-up, what was once  What was once about advancing technology is fast becoming a rent-seeking system designed to enrich financial gatekeepers, not fuel progress.