Recently Larry Fink has mentioned that computing power is going to be a commodity. He added there's no such thing as an AI bubble. He called both the flip side of the same coin.

They are in charge of $13.9 trillion, an assertion that they aren't releasing when the CEO stands on stage and announces yet another new asset class around the corner for one of the largest asset managers on Earth by assets under supervision.

It is your north star of where the money will soon be pouring in.

Larry Fink: At some point "you are going to have a new asset class buying futures of compute" statement Tuesday at Milken Institute Global Conference in Beverly Hills, he continued, saying: "We just do not have enough compute power right now."

Both sentences are connected. Therefore, you have 2 definitions: the first is that of a financial instrument, and other is for scarcity.

The scarcity argument

Fink is on the opposite side of a question regarding an AI bubble, that valuations of AI do not make sense along with revenue growth now. So extreme and constraining in fact, are you able to really discuss it as a bubble without missing the point? "It's not an AI bubble," he said. "There is the opposite."

And by this, he means that the bottleneck is in supply and not demand. In other words, the US simply doesn't have the chips, the memory or data center real estate, nor power to put them in useful operation - to deliver on workloads enterprise and government customers are already locked into running.

It was 80x bigger than in Q1 according to Anthropic, and still couldn't keep pace with demand. The five largest hyperscalers are on track to dump over $800 billion into AI infrastructure in 2026 alone. All the large model labs ration compute. That is not bubble behavior. That is a supply shortage.

On the same Milken panel, Brookfield CEO Bruce Flatt was less diplomatic: "We're going to over the next 10 years start to rewire the global economy.”

Then comes the fu­ture idea of computing, which has emerged as an implication of this constraint. You have no way to lock in future computing capacity as a financial instrument, the way an airline can commit to a price of fuel, or how a food manufacturer can lock in the cost of wheat today.

Fink comments: that gap is going to be closing. For example, a futures market for compute could let data heavy industries hedge against erratic price spikes of GPGPU time, memory access and inference capacity. It would also enable infrastructure providers to use forward contracts on capacity they have not yet built as a financing source for construction. The model is not new. Commodity markets, so it has been for a century.

What BlackRock is actually doing

Fink is not just theorising. BlackRock has an entire division here called Global Infrastructure Partners, which is deploying capital into the physical layer today at scale.

The firm is also conducting a roughly $40 billion acquisition of Aligned Data Centers. An AES Corporation $10.7B contract to set aside power for AI workloads $30B initial funding target over the AI Infrastructure Partnership with Microsoft and Abu Dhabi's MGX of up to $100B (gross). Those include a consortium of partners that boasts Nvidia and xAI, among others.

Speaking to the Milken audience Tuesday, Fink revealed that BlackRock would announce a partnership with a yet-to-be-named hyperscaler on building data centers later this week. He declined to name the counterparty until after an announcement about the deal was formalised.

The one thing that they all have in common with each of their patterns of movement. BlackRock not only is a financier in AI related infrastructure but also an owner of the physical layer; the land, the power connections, cooling elements and buildings where AI systems exist.

What no one spoke about

NUMBERS: A poll of enterprise executives conducted shortly before the conference found that 80% privately confessed their corporate AI transformation narrative was ahead of true progress inside, not exactly shocking news.

C-suite executives (78%) were more likely to admit it than those in the VP-level role. Companies are saying one thing to investors and doing something else entirely inside the company; there is a chasm between what they claim to be doing for their customers and what they actually do.

That tension does not undermine the infrastructure claim of Mr. Fink, if anything, it corroborates that assertion. The AI-washers of the day will have to deliver proper, functioning AI. Whenever and if they do that, they are going to need Fink to own and deliver the FI compute. The shortage increases in prognostication.

Fink is not betting on the winner in the AI models sweepstakes. He is betting on the pipes.

OpenAI ships voice models that reason, translate, and transcribe live

OpenAI released a new generation of voice models in its API on Wednesday, giving developers tools to build apps that can reason through spoken requests, translate across +70 languages, and transcribe speech as it happens.

Apple is hiring 20,000 workers while the rest of Big Tech is cutting. 

While Meta, Microsoft, Oracle, and Amazon have collectively cut tens of thousands of jobs this year, Apple committed in early 2025 to add 20,000 workers over four years. The reason is structural. Apple spent $4.3 billion on capex in the first half of its fiscal year.

Microsoft is spending $190 billion this year. Meta just raised its 2026 capex guidance to between $125 billion and $145 billion. Apple did not over-hire during the pandemic and is not running an AI infrastructure arms race, which means it has nothing to cut and no debt to service.

Critics say that makes it a bystander in the most important technology build-out in a generation. Apple settled a $250 million lawsuit this week over allegedly misleading marketing of its Apple Intelligence features, suggesting the AI product itself is not yet delivering on the promises made to sell the hardware. Whether disciplined capital allocation is wisdom or missed opportunity depends entirely on whether the compute arms race produces returns.

Join the Conversation!

We'd love to hear your thoughts and comments. Join our community and stay updated with the latest trends and discussions in crypto.

Twitter | Instagram | Telegram Channel | Linkedin | Facebook