NBIS vs CRWV vs IREN: Which AI Infrastructure Stock Win the Race? | Stock Analysis

AI data centers were supposed to be the safest way to play the AI boom. Then the

neoclouds crashed. Corweave dropped.

Nibbius pulled back. Iron spiked then wobbled. Behind that volatility is a

wobbled. Behind that volatility is a brand new business model that Wall Street still doesn't fully understand.

Are these companies the picks and shovels of the AI gold rush or just overleveraged landlords with fancy GPUs?

Today we'll break down the Neocloud crash and ask one question. Which

Neocloud actually survives the next 5 to 10 years? Coreweave, Nibbius, or Ian?

10 years? Coreweave, Nibbius, or Ian?

Let's get into it. First, what exactly is a Neocloud? It's not a classic hyperscaler like Microsoft Azure, Google Cloud, or Amazon Web Services. Instead,

think of Neoclouds as specialized AI landlords. They secure land and cheap

landlords. They secure land and cheap power. They spend billions on Nvidia

power. They spend billions on Nvidia GPUs and data centers and then they rent that compute to whoever needs AI capacity the most. A lot of them started life as Bitcoin miners. Mining got

tougher, margin shrank, but they were already good at buying power, building data centers, and running intensive workloads. So, they pivoted from mining

workloads. So, they pivoted from mining Bitcoin to hosting AI workloads. Same

infrastructure DNA, completely different customers and contract structures. If

Microsoft, Google, and Amazon are so huge, why do they even need neoclouds?

The answer is simple. AI demand is growing faster than construction timelines. Training and serving big AI

timelines. Training and serving big AI models requires massive GPU clusters, huge amounts of electricity, and billions in upfront capeex. Even mega

caps can't spin up that capacity overnight. Neocloud stepped in as

overnight. Neocloud stepped in as flexible shock absorbers. They take on the power contracts and data center builds and hyperscalers and AI labs sign multi-year capacity deals with them.

That's how companies you barely heard of a few years ago are now announcing 10 plus billion dollar contracts and multi-gawatt power pipelines. On the

surface, the Neocloud business model looks incredible. Step one, lock in

looks incredible. Step one, lock in multi-year contracts with blue chip customers. Step two, show a huge backlog

customers. Step two, show a huge backlog and tripledigit revenue growth. Step

three, pitch yourself as the AI infrastructure toll road. But under the hood, it's very different from software.

They must spend billions upfront on GPUs, power, and buildings. They usually

finance that with a lot of debt. New

sites can take years to go from planned to revenue generating. So, the equity investor sits on top of construction risk, interest rate risk, and customer concentration risk. When everything is

concentration risk. When everything is going perfectly, that leverage amplifies returns. When something slips, a delay

returns. When something slips, a delay project, a weaker contract, or tighter credit, that same leverage amplifies the downside. That's the core reason the

downside. That's the core reason the Neocloud stocks have been so volatile.

Let's hit the bare case head-on.

Skeptics like Jim Chonos argue that Neoclouds are not real tech companies at all. In their view, these firms are

all. In their view, these firms are basically highly levered data center landlords with AI branding and NVIDIA stickers on the front door. They say

neoclouds deserve lower infrastructure style valuations, not high growth multiples. And they worry the industry

multiples. And they worry the industry is overbuilding in response to a temporary AI mania. Hyperscalers have a safety net. If AI demand cools, they can

safety net. If AI demand cools, they can repurpose capacity for storage, search, video, or internal workloads. Neoclouds

don't have that same flexibility. Their

revenue is tied almost entirely to AI demand and a small group of giant customers. The bare case in one line.

customers. The bare case in one line.

You borrowed heavily to build AI only infrastructure. If this boom slows, your

infrastructure. If this boom slows, your equity gets crushed. Now the bull case looks at the same data and reaches the opposite conclusion. Bulls argue that we

opposite conclusion. Bulls argue that we are still in the early innings of AI adoption. Every major company is rolling

adoption. Every major company is rolling out AI features. Training new models and running inference at scale will keep pushing demand higher. They point out that many Neoclouds say their current

GPU clusters are fully booked. An

independent forecast suggests Neocloud revenue could grow at very high double-digit rates for the rest of this decade. In that world, being an AI

decade. In that world, being an AI landlord is actually powerful. Your

mission is to secure land and power before everyone else, and you monetize that scarce capacity through long-term contracts. For the Bulls, this recent

contracts. For the Bulls, this recent sell-off isn't the end. It's a reset.

Expectations come down, valuations cool off, and long-term investors finally get a chance to start positions without paying peak prices. Now, let's talk companies. Starting with Core Weeave.

companies. Starting with Core Weeave.

Coreweave is the scale leader in the Neocloud space, a massive revenue backlog tied to customers like OpenAI and Meta, gigawatts of contracted power across multiple sites, and a deep

partnership with Nvidia, including early access to new GPUs, and even a deal where Nvidia can buy and use capacity for years. That's the bull story. If

for years. That's the bull story. If

Nvidia remains dominant and AI demand stays strong, the biggest, best connected Neocloud could be extremely valuable. But Cororeweave also comes

valuable. But Cororeweave also comes with serious weight. It carries very large debt running into the tens of billions. Interest costs are high and

billions. Interest costs are high and current returns on capital look modest next to its cost of capital. When a

third party developer delayed a key site, Corweave had to cut guidance and the stock was punished instantly. So

Cororeweave is the high-scale high lever Nvidia proxy. If everything goes right,

Nvidia proxy. If everything goes right, the upside is huge. If the buildout or financing stumbles, equity feels it first. Next, we have Nibbius. Nibbius is

first. Next, we have Nibbius. Nibbius is

based in Europe and has built its Neocloud business a bit differently. It

started by serving smaller AI customers and startups, then landed large anchor contracts with companies like Microsoft and Meta. Financially, it looks like

and Meta. Financially, it looks like this. It has a smaller revenue base than

this. It has a smaller revenue base than Coreweave, but it's growing extremely fast. It has less debt and at least for

fast. It has less debt and at least for now a more conservative balance sheet.

Because of that, the market values it at a higher multiple of sales, but it isn't risk-free. It also relies on partners to

risk-free. It also relies on partners to build data centers, and there are already schedule pressures on some big projects. NBIS is effectively the

projects. NBIS is effectively the cleaner balance sheet, higher valuation option. You're trading some financial

option. You're trading some financial safety for a richer price tag and very similar execution risks. Finally, let's

look at IEN. Iron started as an Australian Bitcoin miner, but today its strategy is all about renewable powered data centers. It builds and runs many of

data centers. It builds and runs many of its own sites, and it has secured large amounts of power capacity in attractive locations. The thesis here is that power

locations. The thesis here is that power and land may become the real bottlenecks in AI. GPUs are critical, but if you

in AI. GPUs are critical, but if you don't have power, you can't use them.

Iron has signed a multi-billion dollar AI cloud contract with Microsoft and is targeting multi-billion annualized AI cloud revenue over the next couple of

years. But remember, today AI cloud is

years. But remember, today AI cloud is still a small slice of its total business. It has to execute a huge pivot

business. It has to execute a huge pivot from minor to AI landlord at high speed.

Iron is the power first optionality play. If energy and land are the scarce

play. If energy and land are the scarce resources and they execute well, the upside is big. If they miss timelines or economics are weaker than hoped, the

downside shows up quickly. So after the Neocloud crash, who actually survives and thrives? Here's a simple framework.

and thrives? Here's a simple framework.

This is not financial advice, just a way to think about it. Poore is the high-scale hidea Nvidia proxy. Biggest

upside if everything goes right, biggest balance sheet risk if it doesn't. Nius

is the diversified customer, cleaner balance sheet option. You pay more per dollar of sales but get a structure that looks a bit less aggressive. Iron is the power and land bet. If energy is the

long-term bottleneck and they manage the pivot, it could be very interesting. All

three are high beta. High execution risk plays on the AI cycle. If you believe AI demand will keep outrunning capacity for years, the Neocloud crash might be the

opening chapter, not the ending. But

whether you look at coreweave, nebius or IN remember don't just chase the contracts. Always understand the debt,

contracts. Always understand the debt, the buildout risk and how the story looks if the cycle slows down. If this

breakdown helped, consider subscribing so you don't miss the next deep dive into AI infrastructure and semiconductor investing.