NeoClouds for Sale: Can META Buy Nebius?

Next Year Will Be the Year of M&A for NeoClouds. What Happens to IREN, CIFR, and WULF?

Alphabet agreed to acquire Intersect Power for $4.75 billion, including assumed debt. This deal immediately raises questions about whether Google could eventually acquire partners like CIFR or WULF and when M&A activity among NeoClouds might actually begin. We’ll get to that in detail, but first it’s worth taking a closer look at the Intersect transaction itself.

Intersect Power develops, builds, and owns utility-scale energy infrastructure used to support large-scale data-center and hyperscaler demand. The company focuses on projects where power generation, storage, and data-center readiness are designed together, rather than added sequentially.

As of today, Intersect’s portfolio includes:

• large-scale solar generation assets,

• grid-level battery storage,

• and power-ready sites intended for direct co-location with hyperscale data centers.

Intersect controls a development pipeline of approximately 10.8 GW of projects expected to be delivered by 2028 across multiple U.S. states. This figure reflects generation capacity, storage, and grid-connected infrastructure already moving through permitting and construction stages.

Google had been involved with Intersect prior to the acquisition. In 2024, Google participated in an $800+ million funding round backing projects where power infrastructure and data-center capacity are developed in parallel.

By acquiring Intersect, Google gains direct control over:

• multi-gigawatt power development capacity,

• construction timelines,

• and long-dated access to energized sites for AI and data-center expansion.

The transaction immediately sharpened M&A discussions across the NeoCloud sector, where most operators remain dependent on external capital markets to finance power access, land, and build-outs at comparable scale.

After the HUT–Fluidstack deal, X immediately started joking about how fast HUT would be back in the market raising money same day or next.
This time it didn’t happen. But for NeoClouds, going back to the capital markets right after announcing a big deal has quietly become standard behavior.

When Nebius’ CEO called the Microsoft deal fundraising three months ago, people noticed. That framing stuck. Deals like these stopped being seen as purely strategic and started being read through a capital-markets lens.

Now analysts talk about potential takeouts next year, while the market keeps asking a simpler question: who could actually buy and at what price.

After the recent drawdown, timing matters. The conversation is no longer about whether consolidation might happen, but how it could realistically play out.

Five things that matter for NeoClouds right now

Trying to rank NeoClouds by how much debt they carry or how “levered” they look misses the point. At this stage of the cycle, balance sheets matter less than positioning. The market is no longer asking who has the cleanest structure on paper. It’s asking who can survive a tighter capital regime long enough to keep executing.

There are a few lenses that actually matter.

First, how soon does the company need to go back to the capital markets.
Not whether it has debt, but whether the market already expects another raise. Some companies are still given time. Others are treated as permanent fundraisers, where every announcement is immediately followed by speculation about dilution or a new convert. That expectation alone changes how the stock trades and how much flexibility management really has.

Second, what the market views as the company’s real asset.
This is often very different from what shows up in investor decks. Power access, land, interconnection, contracts, execution teams, or a software layer that can be separated from the balance sheet. If the value can be carved out, the company doesn’t need to be acquired to be monetized. That distinction matters a lot for M&A.

Third, whether the business can be bypassed instead of bought.
Most acquisitions don’t happen because something is attractive. They happen because something is hard to replicate. If a buyer can wait, partner, finance, or build around a company, there is no urgency to buy the equity, even after a large drawdown.

Fourth, how politically and strategically neutral the company is.
Some assets integrate cleanly. Others come with regulatory noise, geopolitical baggage, or unwanted visibility. The latter may be strategically important but still unattractive as outright acquisition targets, especially for large buyers who care about optics as much as economics.

Finally, whether management has time.
This is the outcome of all the above. Time is the most valuable asset in this market. Companies with time can wait for better pricing, better partners, or better conditions. Companies without time become the subject of takeover speculation, not because they want to sell, but because the market assumes they will eventually have to do something.

This is the framework through which NeoCloud M&A should be viewed. Not as a question of who looks cheap, but of who has leverage, patience, and optionality and who doesn’t.

IREN: Vertical Integration at Scale

IREN operates approximately 3 GW of power capacity, combining energy generation with datacenter infrastructure. With a market capitalization around $11 billion, the company has built its model on vertical integration - owning the power generation assets that supply its AI computing facilities.

In December 2024, IREN raised $2 billion through a convertible debt offering. This financing structure reflects the capital-intensive nature of gigawatt-scale infrastructure buildouts. The company relies heavily on debt financing to fund expansion, betting that long-term capacity contracts with hyperscalers like Microsoft will generate cash flow sufficient to service obligations while continuing to build out additional capacity.

Scenario 1: Full acquisition of IREN

A full takeout of IREN is feasible largely because, in absolute terms, it is not prohibitively expensive for a strategic buyer. Even with a premium, the transaction size would be manageable for a hyperscaler or a large financial sponsor.

In this scenario, the buyer:

• assumes existing convertible debt and future funding needs,

• consolidates execution risk and long-dated CapEx,

• gains full control over power, land, sites, and deployment.

The Microsoft contract plays a critical role here. Once fully realized, it can underwrite a significant portion of IREN’s debt and materially reduce cash-flow uncertainty. Remaining capacity can be monetized through colocation or long-term capacity agreements.

The downside is structural: the buyer pays today for upside that is still being built. All future execution, funding, and scaling risk is absorbed upfront. This only makes sense if the buyer is confident in long-term power scarcity and views IREN as core infrastructure rather than an option.

---

Scenario 2: Minority stake (~20%) with control features

A minority investment with control rights looks more efficient under current market conditions.

In this structure, the buyer acquires:

• ~20% equity,

• governance rights,

• options or convertibles,

• and commercial agreements (capacity rights, exclusivity, ROFR).

This allows the buyer to:

• start financing the build-out without full consolidation,

• lower IREN’s cost of capital,

• secure de facto control over critical assets,

• preserve flexibility on timing and valuation.

Debt tied to MSFT-backed project is effectively covered by contracted cash flows. Remaining sites and future capacity can be developed and monetized as colocation or capacity-as-a-service.

This structure suits both hyperscalers and private funds. Hyperscalers secure capacity and control bottlenecks without taking full equity risk. Financial sponsors focus on capital structure optimization and execution de-risking, keeping the option of a full takeout later.

Key takeaway

This is not a question of if IREN is valuable.
It is a question of how and when to take risk.

• Full acquisition locks everything in today.

• Minority control buys influence, time, and optionality.

In the current market, the second option is the more natural first step

How the two IREN scenarios map to the five-point framework

1. Time pressure
IREN is one of the few NeoClouds where the market is not yet expecting an immediate capital raise. That pressure will inevitably increase as GPU capacity scales.
→ This is why a minority stake makes more sense: it removes future timing risk without forcing a full takeout.

2. What the market sees as the real asset
For IREN, the real assets are not equity, but:

• power,

• permitted sites,

• MSFT-backed capacity.
  → A full acquisition buys everything at once.
  → A partial deal acquires exactly what the market values, without overpaying for the rest.

3. Can it be bypassed instead of bought
IREN cannot simply be bypassed or quickly replicated — the power and sites already exist.
But it can be structured rather than acquired outright.
→ This naturally points to carve-outs, JVs, and minority control structures instead of full M&A.

4. Political and strategic neutrality
IREN is a large, public, capital-intensive asset. A full takeout would be:

• highly visible,

• complex,

• and require a clear strategic justification.
  → A partial acquisition reduces both political and reputational exposure for the buyer.

5. Management time
IREN still has time.
That availability of time is precisely what makes a structured deal preferable to a forced transaction.
→ A full takeout is a bet on certainty.
→ A minority stake is a bet on optionality.

Who is most likely to step in

If you follow recent deals closely, the most realistic first buyers are private capital, not hyperscalers.

Google’s move with Intersect Power is a good reference point. Rather than buying platforms outright, hyperscalers are increasingly acquiring assets in pieces: power, land, development pipelines, long-term capacity rights. They avoid equity risk, public-market noise, and future CapEx commitments where possible.

That dynamic naturally favors private funds as the initial counterparties.

Private capital is structurally better positioned to:

• underwrite capital-intensive build-outs,

• absorb construction and execution risk,

• restructure balance sheets,

• and lower WACC through patient capital and project-level financing.

For funds like Apollo, KKR, Brookfield, Blackstone, a minority stake in IREN combined with governance rights and commercial agreements is a familiar structure. It allows them to:

• deploy large amounts of capital efficiently,

• monetize contracted cash flows (such as MSFT-backed capacity),

• and keep optionality on a full exit later.

Hyperscalers, by contrast, can stay disciplined:

• letting private capital do the heavy lifting upfront,

• stepping in later through long-term capacity contracts,

• or acquiring specific assets once execution risk is removed.

This is how the puzzle fits together.

Private funds finance, structure, and stabilize.
Hyperscalers buy capacity, control bottlenecks, or acquire assets selectively.

In that context, a minority stake with control features is not a compromise — it is the cleanest way to align incentives across all parties under current market conditions.

A full takeout of IREN may still happen one day.
But if it does, it is far more likely to come after private capital has already helped de-risk execution and scale the platform.

CIFR: a cleaner M&A case built around sites, pipeline, and team

Compared to IREN, CIFR looks structurally simpler to underwrite from an M&A perspective.

Today, CIFR controls a development pipeline of roughly 3.4GW across 8 sites, built around land acquisition, power access, and standardized, AI-ready designs. The company does not try to own the entire stack or solve every layer internally. Its value is much more concentrated around site selection, permitting, and execution speed.

That logic was reinforced by today’s acquisition of a new 200MW site in Ohio. The site includes 195 acres, secured power agreements with AEP Ohio, access to PJM, and energization targeted for Q4 2027. This is exactly the type of asset buyers care about: large, power-secured, and timed for the next wave of HPC demand.

From an M&A standpoint, CIFR is easier to think about because:

• the asset base is modular,

• sites can be financed or acquired individually,

• and execution risk is concentrated at the development level, not the platform level.

One important detail that often gets overlooked: CIFR is not just assembling land and power. The company explicitly offers end-to-end data center development, including site preparation, power integration, and full build-out through energization.

In other words, CIFR is positioning itself not only as a land-and-power aggregator, but as a turnkey delivery partner for hyperscalers and large AI customers. That matters for M&A, because it collapses multiple steps of execution into a single acquisition.

This is where the logic becomes straightforward.

A buyer like Google does not need to acquire a full NeoCloud platform to solve its bottlenecks. What it needs is:

• power-secured sites,

• predictable delivery timelines,

• and teams that can execute without reinventing processes internally.

CIFR fits that profile. And when viewed alongside Terawulf, the picture becomes even cleaner.

CIFR + WULF

Taken together, CIFR + WULF form a coherent infrastructure stack: power, land, build-out, and delivery — without the need to acquire a fully integrated NeoCloud operator upfront. This is exactly the kind of modular consolidation we are starting to see across the sector.

There is also an existing structural link that matters. Google already holds 14% stake in WULF, acquired through a strategic investment. The stake does not imply control, but it is large enough to formally align interests and establish a long-term commercial relationship.

That alone changes how WULF should be viewed. This is not a greenfield relationship, but one where capital, power access, and strategic coordination are already in place.

This makes WULF a natural bridge. It also explains why looking at CIFR and WULF together is more logical than treating them as standalone M&A cases. One brings large-scale power access tied to a hyperscaler relationship, the other brings site development and end-to-end delivery capability.

At current market prices, CIFR and WULF together imply roughly ~$10B of equity value, while controlling about 4.6GW of power-backed capacity, including land, permitted sites, and operating teams.

In that setup, the buyer is effectively paying for power, sites, and build-out capability. Team size and corporate layers are secondary variables and can be rationalized post-transaction, while procurement can be optimized across equipment and construction.

From there, the sequencing becomes clearer.
Once power, land, and execution are either controlled or tightly aligned, the next logical layer is compute at scale. That is where NBIS enters the discussion — not as a starting point, but as a downstream step, once infrastructure constraints have been addressed.From there, the next step becomes obvious.

Wedbush has repeatedly pointed to NBIS as a potential acquisition candidate, not because of near-term profitability, but because of its GPU density, customer exposure, and role as a compute reseller at scale.

In that sequence:

1. power and sites are secured,

2. delivery and execution are internalized,

3. and compute platforms can be layered on top or acquired selectively.

Nebius: compute scale comes last, not first

Nebius sits on a different layer of the stack. This is not a power story, not a land story, and not a development platform. Nebius is fundamentally about compute aggregation and utilization.

As of today, Nebius is one of the largest independent AI compute platforms outside hyperscalers. That scale is exactly why Wedbush and others flag NBIS as a potential acquisition candidate — but it’s also why timing matters much more here than for infrastructure-heavy peers.

From an M&A perspective, NBIS is not a starting point.

At a roughly $24B valuation, this is not a casual deal. The buyer is not paying for scarce land or hard-to-replicate power access. The buyer is paying upfront for:

• utilization,

• customer demand,

• execution capability,

• and future margins.

That immediately narrows the buyer universe.

Buying compute without controlling power, sites, and build-out pipelines means inheriting:

• high operating leverage,

• utilization risk,

• and dependency on third-party infrastructure.

That’s why NBIS only makes sense after infrastructure bottlenecks are addressed.

This is where the discussion becomes interesting.

In theory, buyers like META or NVIDIA get mentioned. META has a long history of paying up for teams when speed and talent density matter more than near-term economics. In that context, NBIS would not be about cheap capacity or immediate synergies — it would be a team-and-execution acquisition, with compute as the vehicle rather than the objective.

NVIDIA is another thought experiment. NVDA doesn’t need NBIS to sell GPUs, but owning a large-scale compute platform could, in theory, give tighter control over deployment, utilization, and downstream monetization — especially after restructuring its cloud-facing organization.

This is not a base case. It’s a strategic thought experiment.

There is also a more ironic angle that keeps coming up in discussions: NBIS arguably makes the most sense for buyers who already control land and power — in other words, miners.

For companies sitting on large energy footprints and permitted sites, NBIS would not be expansion into something new. It would be vertical completion. Power and land are already there. What’s missing is large-scale, monetizable compute and a team that knows how to run it at utilization.

Viewed this way, NBIS is less a hyperscaler target and more a natural fit for infrastructure-first players moving up the stack. Owning compute on top of owned power flips the economics: GPU-heavy operations stop being a variable cost and start becoming an internal layer.

This doesn’t mean such a deal is imminent or even likely. But conceptually, it explains why NBIS keeps resurfacing in M&A conversations.

NBIS is interesting not as a standalone acquisition, but as the final piece in a vertically integrated AI infrastructure stack.

This publication is for educational and informational purposes only and does not constitute financial, investment, or trading advice. Readers are solely responsible for their own investment decisions. The author may hold positions in the securities mentioned.