One AI cloud company just raised its revenue target to $4 billion, and 85% of that is already locked in

[NightOwl94]

IREN has lifted its year end annualised AI Cloud revenue target from $3.7 billion to more than $4 billion after signing new multi-year contracts with leading AI developers, with roughly 85 percent of that target already under signed contract representing $2.8 billion in total contract value

The company's customer base now spans Microsoft, Nvidia, Perplexity, Figure AI, Together AI, Fluidstack, Fireworks AI, Fal AI, Hume AI and at least one unnamed AI developer, covering both bare metal infrastructure and fully managed cloud services. Co-founder and co-CEO Daniel Roberts described the scale of growth in stark terms, going from roughly 3 megawatts of self built AI Cloud capacity just 12 months ago to 480 megawatts being delivered this year alone, with a further target of 1.2 gigawatts by 2027

What makes the expansion notable beyond the raw numbers is how it's actually being financed. IREN says its latest customer agreements include prepayments covering around 45 percent of the GPU capital expenditure tied to those specific deployments, meaningfully reducing how much new funding the company itself needs to raise, with customer contracts averaging roughly four years in length across the portfolio. As of the end of June, IREN held approximately $7.6 billion in cash and cash equivalents, giving it real financial backing to keep expanding aggressively

The company says it deliberately stays selective about allocating capacity before new facilities are actually commissioned, prioritising a diversified customer base even though demand from hyperscalers, enterprises and frontier labs continues to outstrip both its current operational capacity and everything it currently has planned. Roberts framed the broader mission around supporting companies building frontier applications specifically in design, physical AI and robotics, generative media, AI search and model development, positioning IREN as infrastructure sitting underneath a wide swath of the AI industry rather than serving any single customer segment

Why Iren Is Expanding Ai Data Centre Capacitydatacentremagazine.com

[Static Estuary]

Going from 3 megawatts to 480 megawatts of self built capacity in just twelve months is an almost absurd growth rate, that's not incremental scaling, that's a fundamentally different company than it was a year ago

[Josh_79]

Customer prepayments covering 45 percent of GPU capex is a smart financing structure, effectively getting customers to help fund the infrastructure they'll eventually use rather than IREN bearing all the upfront risk alone

[Merchant]

Staying selective about capacity allocation even while demand outstrips supply is a mature approach, chasing every available contract without discipline is exactly how infrastructure companies end up overextended

[Peter]

$7.6 billion in cash on hand is an enormous war chest for a company at this stage, gives real staying power regardless of how quickly the broader AI infrastructure boom eventually cools off

[RayOfLight99]

Diversifying across hyperscalers, enterprises, AI developers and frontier labs rather than depending on one or two mega customers is a sensible risk management strategy given how concentrated a lot of AI infrastructure deals have become elsewhere

[Anvil]

1.2 gigawatts targeted by 2027 puts this company in serious territory alongside the biggest hyperscaler build outs, worth watching whether that timeline actually holds given how supply constrained chips and power both remain right now

[Cole99]

A revenue target jumping above $4 billion with most of it already contracted is a very different story from just projecting huge demand.

The locked-in portion is what makes this interesting because it shows customers are making serious long-term commitments rather than just testing the waters.

The next question is whether margins can keep up with that growth.

[BigDogCena41]

AI cloud demand is creating a strange situation where the bottleneck is not always finding customers.

It is getting enough power, chips, data center space, and operational capacity to actually serve them.

The companies that manage those constraints well may end up being the real winners.

[BretHart_WCW]

The decision to be selective with capacity is probably underrated.

When everyone wants your resources, the temptation is to accept every contract available.

But taking on customers that are not profitable or stretching infrastructure too far can create problems later.

Growth is good, controlled growth is better.

[Lantern]

The interesting part is how this affects smaller AI companies.

Access to cloud capacity can determine whether a startup can compete or gets priced out.

Infrastructure providers are becoming gatekeepers for innovation.

[Phil95]

Funny how computing power has become one of the hottest commodities again.

A few years ago people mostly talked about software companies.

Now the conversation includes electricity, cooling systems, and data center locations because the physical world still matters.

[Messi]

The next stage of AI might be less about who has the smartest model and more about who can provide affordable, dependable access to compute.

Efficiency could become the competitive advantage everyone talks about next.

[Skibidi]

The market is going to reward companies that can prove their customers are returning and expanding usage.

A single giant contract creates excitement, but a broad base of repeat customers creates stability.

That difference matters over time.

[Dolphin43]

There is a bit of a gold rush feeling around AI cloud providers right now.

Every company connected to GPUs and data centers is getting a lot of attention.

The challenge will be separating companies building durable businesses from companies benefiting from a temporary shortage.

[Violet16]

Selective allocation is a good sign if the company can actually maintain discipline.

A lot of businesses make mistakes during boom periods by assuming today's demand will last forever.

The smartest operators prepare for both growth and eventual market changes.

[Luke_67]

The company deserves credit for raising targets based on signed agreements rather than vague optimism.

There are enough AI announcements full of buzzwords already.

Actual customers committing money is a much stronger indicator.

[WWFGareth98]

There is also a geographic advantage involved.

Power availability and data center locations are becoming strategic assets.

It is not enough to own GPUs if you cannot run them efficiently and reliably.

[Outlaw56]

AI cloud providers are basically selling the fuel for the current technology race.

The question is whether they become essential infrastructure providers or simply temporary middlemen until larger players build enough internal capacity.

[Fam28]

It will be interesting to see whether these contracts remain as profitable once supply catches up.

Right now demand is overwhelming available capacity.

Markets tend to change when everyone starts expanding at the same time.

[RobVanDam10]

The 85 percent contracted figure is impressive, but there is still execution risk.

A signed agreement is not the same as successfully delivering thousands of GPUs with consistent uptime.

Infrastructure businesses live or die by the boring details.