SB Energy Wants a $50 Billion Valuation. Nearly 99% of Its Contracted Data-Center Capacity Depends on OpenAI. What Exactly Are Investors Valuing?

SB Energy has no operating data-center revenue, a roughly $439 billion backlog stretching decades into the future, and almost its entire contracted capacity tied to OpenAI. Nvidia is investing, OpenAI holds billions in warrants, and SoftBank sits on multiple sides of the transaction. The IPO exposes a new AI risk: enormous valuations built on concentrated future demand before the underlying economics have been independently proven.

There is a number in SB Energy's IPO story that deserves considerably more attention than its potential $50 billion valuation.

99%.

Reuters reported that approximately 8.753 gigawatts of SB Energy's 8.8 gigawatts of contracted data-center capacity is associated with OpenAI.

Technically, SB Energy has two contracted data-center customers:

SoftBank

and

OpenAI.

But economically, this is overwhelmingly an OpenAI story.

And yet SB Energy may seek a valuation of more than $50 billion as it enters public markets.

That alone would make customer concentration worth examining.

Then the rest of the structure appears.

SB Energy currently generates no revenue from operating data centers.

Its first-half 2026 revenue of approximately $139 million came from its existing energy operations.

It reported a $3.21 billion net loss during the same period.

Its enormous data-center backlog is approximately $439 billion.

Only about 10% of that backlog is expected to materialize within six years.

Most of it comes later.

Much later.

And the overwhelming majority of the contracted capacity depends on one company:

OpenAI.

That creates a very different question from:

How big is the backlog?

The better question is:

How good is the demand inside the backlog?

A $439 Billion Backlog Is an Extraordinary Number

Backlogs matter.

They can show:

future revenue visibility,

customer commitment,

contracted demand,

and business momentum.

Investors understandably value them.

But a backlog is not cash.

It is a claim on future economic activity.

Its value depends on:

who owes the money,

when they owe it,

whether they remain solvent,

whether the contract remains economically attractive,

whether infrastructure is delivered,

whether demand materializes,

and whether the underlying industry evolves as expected.

Those questions become particularly important when the backlog extends decades into the future.

AI may be one of the fastest-changing industries ever created.

Very few people can confidently predict:

AI architecture five years from now,

compute economics ten years from now,

model providers twenty years from now,

or data-center requirements several technological generations from today.

Yet portions of SB Energy's valuation proposition depend on assumptions extending across precisely those horizons.

That is Backlog Quality Risk.

Backlog Size and Backlog Quality Are Not the Same Thing

Imagine two companies.

Company A has $20 billion in contracted future revenue spread across:

hundreds of customers,

multiple industries,

different geographies,

and relatively short contract periods.

Company B has $400 billion in contractual commitments stretching decades into the future, with almost everything dependent on one dominant customer.

Which backlog deserves the higher confidence?

The larger number does not automatically represent the safer business.

Because concentration changes everything.

A large backlog can create the appearance of enormous diversification in time while remaining extraordinarily concentrated in counterparty risk.

That is exactly what investors need to examine here.

OpenAI Is Not an Ordinary Customer

OpenAI is one of the fastest-growing and strategically important AI companies in the world.

That can make it an extraordinary customer.

But it also means SB Energy's dominant customer is participating in one of the most capital-intensive expansion strategies in corporate history.

OpenAI is simultaneously trying to secure:

models,

chips,

data centers,

electricity,

cloud capacity,

custom silicon,

and global infrastructure.

The size of those commitments is enormous.

That means SB Energy investors are not merely underwriting SB Energy.

They are indirectly underwriting assumptions about:

OpenAI growth,

OpenAI revenue,

OpenAI financing,

OpenAI compute requirements,

and OpenAI's ability to monetize AI at sufficient scale for decades.

That is AI Valuation Dependency Risk.

AI Valuation Dependency Risk

I define AI Valuation Dependency Risk as:

The risk that one company's valuation increasingly depends on future revenue from a dominant customer whose own valuation, financing and ability to pay depend on the same expanding AI ecosystem.

The structure can look like:

OpenAI expects enormous AI demand

→ OpenAI commits to enormous compute infrastructure

→ SB Energy receives enormous contracted backlog

→ SB Energy's future revenue appears enormous

→ SB Energy seeks a higher valuation

→ investors finance SB Energy

→ SB Energy builds infrastructure for OpenAI

→ OpenAI gains additional capacity

→ OpenAI's growth expectations rise

→ additional infrastructure becomes justifiable.

The customer strengthens the supplier's valuation.

The supplier strengthens the customer's infrastructure story.

Each reinforces expectations around the other.

That does not mean the demand is fake.

It means the demand is interdependent.

Now Look at Who Else Is Inside the Transaction

Reuters' reporting reveals an unusually dense network of relationships.

SoftBank is SB Energy's:

controlling shareholder,

parent,

customer,

guarantor,

and trademark licensor.

OpenAI is:

the overwhelmingly dominant tenant,

a software supplier,

and holder of warrants associated with its agreement.

Nvidia is:

a critical chip supplier,

an investor,

and a guarantor of part of OpenAI's lease obligations.

That is not a simple:

company → customer

relationship.

It is an ecosystem.

And when multiple parties occupy several economic roles simultaneously, traditional market signals become harder to interpret.

Who Is Actually Taking the Risk?

Normally, commercial relationships create useful separation.

Supplier sells.

Customer buys.

Investor invests.

Bank finances.

Guarantor guarantees.

Each has a relatively clear role.

Here those roles increasingly overlap.

One participant can simultaneously be:

customer,

investor,

supplier,

guarantor,

or beneficiary.

That changes the risk architecture.

Because one company's distress can propagate through multiple relationships simultaneously.

Nvidia Adds Another Loop

Reuters reported that Nvidia has committed $1.5 billion to SB Energy through a private placement associated with the IPO.

Half of Nvidia's shares reportedly will be purchased at a discount to the IPO price.

Meanwhile, Nvidia is also central to the hardware needed for SB Energy's massive Ohio data center serving OpenAI.

Nvidia is therefore connected to:

the infrastructure,

the customer,

the hardware,

the financing ecosystem,

and the equity.

That matters.

Because this increasingly resembles the broader AI ecosystem I have been tracking:

supplier → investor → guarantor → customer → infrastructure → revenue → valuation → more financing.

OpenAI Has Warrants Worth Billions

Reuters reported that SB Energy disclosed OpenAI had been issued warrants valued at approximately $5.5 billion.

That means the dominant customer can also participate financially in the upside of the company providing its infrastructure.

Again, there can be legitimate strategic reasons for this.

Warrants can align interests.

They can encourage long-term commitments.

They can strengthen partnerships.

But they also change the meaning of the customer relationship.

OpenAI is not merely buying infrastructure.

It can benefit financially if SB Energy's valuation rises.

Now ask:

How independent is the demand signal when the dominant customer also has financial upside in the supplier?

That is not an accusation.

It is a valuation question.

Customer Demand Can Become Investor Demand

This is where AI markets are becoming structurally unusual.

OpenAI commits to infrastructure.

The commitment creates backlog.

The backlog supports SB Energy's growth narrative.

The growth narrative can support a larger IPO valuation.

A larger valuation increases the value of equity and warrants.

Participants tied to the ecosystem potentially benefit.

Public investors then enter.

What began as:

customer demand

has become:

valuation support.

That transmission mechanism deserves scrutiny.

The IPO Is Asking Public Investors to Take the Next Position in the Chain

Before the IPO, much of this ecosystem is supported by:

SoftBank,

OpenAI,

Nvidia,

private capital,

and strategic financing.

An IPO introduces a new participant:

the public shareholder.

Public-market investors are being asked to value future cash flows created by relationships that already involve many of the companies financially supporting one another.

That is why the IPO matters beyond SB Energy.

It potentially transfers part of the AI infrastructure wager from concentrated strategic investors into broad public markets.

The Company Has No Operating Data-Center Revenue Yet

This deserves repeating.

According to Reuters, SB Energy currently has no revenue from operating data centers.

The roughly $139 million of revenue reported in the first half of 2026 comes from legacy energy operations.

That means investors considering the company's proposed AI infrastructure future cannot rely heavily on historical data-center operating performance.

There isn't any.

They are underwriting:

contracts,

construction,

future capacity,

future AI demand,

future OpenAI requirements,

and future infrastructure economics.

In other words:

the valuation is predominantly forward-looking.

That Does Not Make the Valuation Wrong

Future-growth companies are valued on future opportunity all the time.

Amazon was once valued far ahead of current profits.

Telecommunications networks required enormous capital before producing returns.

Railroads were financed before cities grew around them.

Infrastructure has always required conviction about future demand.

SB Energy could become enormously valuable.

OpenAI could require astonishing amounts of infrastructure.

AI demand could exceed today's projections.

The company could successfully diversify.

But public investors still need to distinguish:

possible future scale

from

presently validated economics.

That distinction becomes especially important around IPOs.

The $50 Billion Question

Reuters has reported that SB Energy could seek a valuation above $50 billion.

Compare that with the company's existing operating economics.

First-half revenue:

approximately $139 million.

First-half net loss:

approximately $3.21 billion.

Current operating data-center revenue:

zero.

Potential valuation:

$50 billion-plus.

That valuation cannot principally be about the business SB Energy operates today.

It is about the business investors expect it to become.

And the bridge between those two businesses is overwhelmingly:

OpenAI.

Investors Are Effectively Pricing OpenAI Twice

This produces an interesting possibility.

OpenAI has its own enormous private valuation.

Now companies whose growth depends primarily on OpenAI can receive enormous valuations too.

Think about the chain.

OpenAI valuation reflects expectations about future AI demand.

SB Energy valuation reflects expectations about future OpenAI infrastructure demand.

Nvidia valuation reflects expectations about AI compute demand from companies including OpenAI.

Cloud-company valuations reflect expected AI workloads.

Data-center valuations reflect expected AI capacity demand.

Energy infrastructure valuations reflect expected data-center power demand.

Several separate valuations may therefore depend on variations of the same underlying economic assumption:

OpenAI and the wider AI market will generate enormous amounts of economically productive demand.

That raises another risk.

This Is AI Valuation Stacking

Call it:

AI Valuation Stacking.

It occurs when the same underlying expectation of future AI demand supports valuations across multiple companies positioned at different points in the same economic chain.

The sequence can become:

OpenAI expected growth

→ OpenAI valuation

→ OpenAI infrastructure commitments

→ SB Energy backlog

→ SB Energy valuation

→ Nvidia hardware demand

→ Nvidia valuation

→ data-center investment

→ energy infrastructure investment

→ additional AI infrastructure valuations.

Multiple valuations.

One underlying growth thesis.

If that thesis succeeds, enormous wealth can be created.

If it disappoints, several supposedly separate investments can reprice simultaneously.

Diversification Can Become an Illusion

An investor may believe they have diversified by owning:

AI model companies,

chip companies,

data centers,

energy infrastructure,

cloud providers,

and AI financing businesses.

But if all those assets depend on the same few frontier labs generating enormous future cash flows, the portfolio may be less diversified than it appears.

Different ticker symbols do not necessarily mean different economic exposures.

That is hidden correlation risk.

The Backlog Is Long-Dated

Reuters Breakingviews highlighted another major issue.

Only around 10% of SB Energy's enormous backlog is expected within the next six years.

Most comes afterward.

That matters enormously.

Because every additional year introduces uncertainty.

Technology changes.

Customers change.

Capital costs change.

Model architectures change.

Electricity markets change.

Chip economics change.

Regulation changes.

Competitors appear.

A 20-year AI infrastructure contract may be contractually reassuring.

Economically, it asks investors to predict an industry that has difficulty predicting itself 24 months ahead.

Contract Duration Can Hide Technology Risk

Long contracts often reduce business uncertainty.

In rapidly evolving technology markets, they can also introduce another kind of risk.

Suppose AI compute becomes dramatically more efficient.

Suppose new chips require different infrastructure.

Suppose distributed inference reduces centralized data-center requirements.

Suppose new model architectures need less compute.

Suppose AI margins compress.

Suppose regulation changes.

Suppose OpenAI loses market share.

The contract still exists.

But the economics surrounding the contract can change dramatically.

Contractual certainty does not equal economic certainty.

What Happens If OpenAI Pulls Back?

This is the scenario investors need to model.

Not because it is necessarily likely.

Because concentration makes it material.

If OpenAI:

reduces infrastructure spending,

faces financing constraints,

changes architectural strategy,

shifts providers,

renegotiates capacity,

experiences slower demand,

or finds dramatically more efficient compute,

SB Energy cannot simply replace 99% of contracted capacity overnight with thousands of smaller customers.

That is the essence of concentration risk.

When one customer dominates the economics, that customer becomes part of the company's business model.

But Isn't the Contract Supposed to Protect SB Energy?

Long-term leases provide real protection.

That should not be dismissed.

But every contract eventually depends on the counterparty's willingness and ability to perform.

The stronger the customer, the stronger the contract appears.

OpenAI is currently one of the most valuable private companies in the world.

Yet OpenAI is simultaneously making extraordinary infrastructure commitments across the AI ecosystem.

Investors therefore need to analyze not simply:

Does OpenAI have a contract?

but:

What does OpenAI's total obligation structure look like across all of its infrastructure commitments?

That is counterparty intelligence.

OpenAI Is Becoming an Economic Hub

This is another emerging pattern.

OpenAI is increasingly connected to:

data centers,

chips,

cloud providers,

energy companies,

financing structures,

strategic investors,

and software ecosystems.

A single company's future growth assumptions now influence extraordinary amounts of capital outside the company itself.

That means OpenAI's failure radius extends far beyond OpenAI.

If it grows faster than expected, enormous parts of the ecosystem benefit.

If it grows slower than expected, the consequences propagate outward.

That makes OpenAI increasingly systemically important to the AI investment cycle.

SB Energy Is Therefore More Than an IPO

It is a market test.

Public investors are being asked to answer:

How much value should be assigned today to future AI demand concentrated overwhelmingly in one frontier company?

That answer could establish precedent.

If investors enthusiastically accept this structure, other AI infrastructure companies may attempt similar listings.

Huge backlog.

Few customers.

Long contracts.

Heavy capital needs.

Strategic investors.

Enormous valuation.

The market could begin treating future AI infrastructure commitments almost like established revenue.

That changes capital allocation.

Backlog Can Become a Valuation Currency

Once public markets reward enormous contracted backlogs, companies have strong incentives to secure increasingly large long-term commitments.

Those commitments support valuations.

Higher valuations support financing.

Financing supports additional construction.

Construction supports larger commitments.

This can create another feedback loop:

future demand

→ backlog

→ valuation

→ capital

→ infrastructure

→ additional contracted capacity

→ larger backlog

→ higher valuation.

Again, nothing about this requires fraud.

Markets routinely finance expected growth.

The risk is reflexivity.

Expectations can begin supporting the capacity intended to fulfill those same expectations.

Now Connect It to Ghost Demand

This is where the story connects directly with another developing AI infrastructure problem.

Utilities across the United States are discovering that portions of projected data-center electricity demand disappear when stronger financial commitments are required.

The industry calls some of it:

ghost demand.

SB Energy's contracts are obviously much more substantive than speculative grid applications.

But the broader question is the same:

What quality of demand is being counted?

Requested demand.

Contracted demand.

Financed demand.

Guaranteed demand.

Supplier-supported demand.

Related-party demand.

Independent end-market demand.

Those categories should not automatically be treated as interchangeable.

Then Connect It to AI Demand Circularity Risk

This also connects directly to AI Demand Circularity Risk.

Nvidia has:

invested in AI companies,

helped finance customers,

guaranteed infrastructure obligations,

and explored revenue-sharing arrangements with companies buying Nvidia hardware.

Now Nvidia is also investing in SB Energy while helping support OpenAI infrastructure whose data centers require Nvidia chips.

SoftBank sits across multiple roles.

OpenAI is customer and equity beneficiary.

SB Energy becomes infrastructure provider.

The deeper question remains:

How much of the ecosystem's apparent growth is independently generated demand—and how much is growth supported by capital relationships among the same participants?

Reuters itself has now highlighted the difficulty of distinguishing those two categories.

That is becoming one of the most important financial-intelligence questions in AI.

This Is Not Evidence of Artificial Valuation

That distinction is essential.

There is no evidence that SB Energy's valuation is fraudulent.

There is no evidence that its backlog is fictitious.

There is no evidence that OpenAI does not intend to honor its agreements.

There is no evidence that Nvidia or SoftBank are improperly manipulating the IPO.

The more important issue does not require any of those claims.

The structure is objectively highly concentrated.

The business is highly forward-looking.

The infrastructure is capital-intensive.

The relationships overlap.

And the valuation depends significantly on future demand from one dominant customer.

Those facts alone justify scrutiny.

A $50 Billion IPO Should Be Able to Survive One Question

Remove OpenAI.

What is SB Energy worth?

Not because OpenAI is going away.

Because concentration analysis requires understanding how much enterprise value depends on one counterparty.

If the answer changes dramatically when one customer's economics change, then investors are effectively buying exposure to that customer through another company's stock.

That should be priced accordingly.

The Real IPO Question Is Independence

Investors usually ask:

How large is the market?

How large is the backlog?

How quickly will revenue grow?

What margins can the company achieve?

Those questions still matter.

But AI infrastructure increasingly requires another:

How independent is this company's economic engine?

Independent of:

one customer,

one supplier,

one parent,

one guarantor,

one financing ecosystem,

one technology architecture,

and one AI growth thesis.

Because enormous scale does not necessarily create independence.

Sometimes scale creates greater concentration.

The Strategic Questions

Public-market investors should be asking:

What percentage of SB Energy's future data-center revenue ultimately depends on OpenAI?

What portion of the $439 billion backlog is legally firm versus conditional?

What termination, renegotiation or capacity-adjustment rights exist?

What happens if OpenAI's compute requirements are materially lower ten years from now?

What assumptions support the $50 billion-plus valuation?

How much of SB Energy's future value comes from contracts beginning more than six years from now?

How should investors discount 20-year AI infrastructure commitments in a technology market changing every few months?

How much of OpenAI's ability to satisfy those contracts depends on future financing?

What happens if OpenAI simultaneously reduces infrastructure commitments across multiple providers?

What part of SB Energy's backlog represents independently originated customer demand?

How do Nvidia's investment and guarantees affect risk perception?

How should OpenAI's warrants affect interpretation of the customer relationship?

How much economic exposure is being replicated across OpenAI, Nvidia, SB Energy, SoftBank and other AI infrastructure investments?

And the simplest question:

If one customer represents nearly the entire future data-center business, are investors buying SB Energy—or another leveraged expression of the OpenAI growth thesis?

The Strategic Conclusion

SB Energy may become an enormously successful AI infrastructure company.

Its contracts may perform.

OpenAI may require every gigawatt currently projected and more.

Its $439 billion backlog may eventually generate extraordinary cash flow.

A $50 billion valuation may ultimately look cheap.

But none of those possibilities eliminates the current risk architecture.

Today:

SB Energy has no operating data-center revenue.

Its historical revenue comes from legacy energy operations.

Its proposed future is overwhelmingly dependent on AI infrastructure.

Nearly all of its contracted data-center capacity is tied to OpenAI.

Its backlog stretches deep into a technologically uncertain future.

SoftBank occupies several sides of the economic relationship.

OpenAI is both dominant customer and warrant holder.

Nvidia is supplier, investor and guarantor within the broader infrastructure chain.

And public investors are being asked to assign a valuation today to cash flows that may arrive across decades.

That is not simply a growth-company IPO.

It is an experiment in how public markets value AI dependency.

This creates at least three distinct risks:

AI Valuation Dependency Risk

when one company's valuation depends heavily on another AI company's future economics.

Backlog Quality Risk

when enormous contracted numbers conceal concentration, duration and counterparty uncertainty.

And AI Valuation Stacking

when the same underlying expectation of future AI demand supports valuations across multiple companies in the same economic chain.

That last one may be the most important.

Because the market can contain dozens of companies—

OpenAI.

Nvidia.

SB Energy.

Cloud providers.

Data-center operators.

Power companies.

Financing vehicles.

Chip suppliers.

—and still be making essentially the same investment repeatedly:

AI demand will become enormous, profitable and durable enough to support all of us.

If that assumption is correct, the wealth creation could be extraordinary.

If it is wrong about the magnitude or merely wrong about the timing, the repricing will not necessarily stay inside one company.

The same underlying expectation is already embedded across multiple balance sheets and valuations.

That is why the number investors should focus on may not be SB Energy's proposed $50 billion valuation.

Or even its $439 billion backlog.

It may be:

99%.

Because when almost the entire future business depends on one customer, the most important question isn't how large the opportunity looks.

It is:

How much of the valuation survives if the customer driving it does not become exactly as large as everybody currently assumes?

I write about AI failure intelligence, ROI exposure, high-stakes decision architecture, and the hidden pathways through which AI incidents become financial and institutional consequences.

Follow me and subscribe to my work if you are responsible for investing in, acquiring, governing, insuring, or protecting strategically important AI systems and need to understand what technical failure can become after it leaves the engineering team.