In the race to build the physical infrastructure of artificial intelligence, two companies — Applied Digital and IREN — are staking out divergent paths through the same uncertain terrain. Applied Digital has grown with striking speed, tripling its revenue in a single year by leasing high-performance computing facilities, while IREN is dismantling its crypto-mining past to reinvent itself as an AI cloud provider with contracts from Microsoft and NVIDIA. Both are spending billions before they earn billions, and the choice between them is less a question of which is better and more a question of
IREN vs. Applied Digital: Comparing Two AI Infrastructure Plays
One customer accounted for 59 percent of revenue—lose that contract, everything shifts.
Why does it matter that Applied Digital has one customer representing 59 percent of its HPC revenue? Isn't that just how these contracts work early on?
It matters because if CoreWeave walks away or significantly reduces its footprint, Applied Digital's revenue drops by more than half. That's not a normal business risk—that's existential. The company has built its entire expansion strategy around that relationship.
But we should note that CoreWeave itself is a growing company with major backing. The risk is real, but it's not like Applied Digital is dependent on a customer that's unstable. Still, concentration risk is concentration risk.
Both companies lost money last year. How do we know either one will ever be profitable?
They're both in the investment phase. You spend billions building data centers before you collect revenue from them. The question is whether their revenue growth will eventually outpace their spending. Applied Digital grew revenue 183 percent year-over-year. That's the kind of trajectory that could lead to profitability if it sustains.
Except free cash flow was negative $2.8 billion, and stock-based compensation was inflating the cash flow numbers. The real cash burn is worse than the headline suggests. IREN's situation is similar—negative $2.2 billion in free cash flow. We don't actually know when either company reaches cash flow breakeven.
What's the deal with IREN's $638.8 million impairment charge? Is that a one-time thing or a sign of deeper problems?
It's a one-time charge related to writing down the value of mining hardware as the company transitions away from Bitcoin. It's not an ongoing expense. But it does mean the company made a big bet on mining that didn't pan out the way it expected.
Right, but it also signals that IREN's management made a strategic error. They built a massive mining operation and then had to abandon it. That's not nothing. The impairment makes the loss look worse than it is, but the underlying decision to pivot away from mining suggests the original strategy wasn't working.
IREN owns its own land and power infrastructure. Why is that better than Applied Digital's model?
Because IREN doesn't depend on third parties for critical resources. It controls its destiny. If there's a permitting delay or a utility issue, IREN can solve it directly. Applied Digital has to negotiate with landlords and utility companies. Vertical integration also means IREN captures more of the margin.
But it also means IREN has to manage more complexity and more capital expenditure. Applied Digital can lease land and focus on building data centers. IREN has to do everything. That's not necessarily better—it's just different. And IREN's debt-to-equity of 1.9x is lower than Applied Digital's 2.9x, but both are carrying substantial leverage.
If I had to pick one, which would you choose?
IREN. Lower debt, vertical integration, major contracts with Microsoft and NVIDIA, and 100 percent renewable energy. The transition from mining to AI services is real, and the company has the infrastructure to scale.
I'd want to see more quarters of data before committing. Both companies are in unproven territory. IREN's lower debt is an advantage, but Applied Digital's lower valuation multiple might offer better upside if it can maintain its growth rate. The honest answer is we don't know which one wins.
The Pulse
- Applied Digital nearly tripled its revenue to $611 million in a single fiscal year, but a $250 million net loss and negative $2.8 billion in free cash flow reveal how expensive that growth actually is.
- A single customer accounts for 59% of Applied Digital's high-performance computing revenue — a concentration risk that could unravel the financial story if that relationship changes.
- IREN reported a staggering $702.6 million net loss, but $638.8 million of it came from writing down obsolete Bitcoin mining hardware as the company pivots toward AI services — a painful but deliberate transformation.
- IREN's lower debt-to-equity ratio of 1.9x versus Applied Digital's 2.9x, combined with land and power infrastructure it owns outright, gives it a structural cushion that its rival lacks.
- Both companies are burning cash at scale to build data centers that don't yet generate the revenue needed to justify the debt — the entire bet rests on AI infrastructure demand remaining as intense as it appears today.
In the race to build the physical infrastructure of artificial intelligence, two companies — Applied Digital and IREN — are staking out divergent paths through the same uncertain terrain. Applied Digital has grown with striking speed, tripling its revenue in a single year by leasing high-performance computing facilities, while IREN is dismantling its crypto-mining past to reinvent itself as an AI cloud provider with contracts from Microsoft and NVIDIA. Both are spending billions before they earn billions, and the choice between them is less a question of which is better and more a question of which risks an investor is willing to carry into an unresolved future.
Two companies are racing to build the physical backbone of artificial intelligence, and they're taking fundamentally different paths. Applied Digital and IREN both operate neoclouds — data centers engineered for the raw computational demands of modern AI — but their financial structures and risk profiles diverge sharply enough that choosing between them means understanding what kind of bet you're actually making.
Applied Digital's model is straightforward: build the facility, lease it to clients who need more processing power than traditional data centers can provide, and collect rent. Its fiscal year ending May 2026 produced $611.3 million in revenue, nearly triple the prior year — the kind of acceleration that catches investors' attention. But the simplicity masks real vulnerability. One customer recently accounted for 59 percent of its HPC segment revenue. The company also carries a debt-to-equity ratio of 2.9x, lost $250.3 million for the year, and generated negative $2.8 billion in free cash flow. Stock-based compensation inflated reported cash generation so severely it represented 245 percent of operating cash flow.
IREN is attempting something more complicated. The company is actively winding down its Bitcoin mining business to pivot toward AI cloud services, and the transition has been costly — a $702.6 million net loss for fiscal year 2026, with $638.8 million of that tied to non-cash impairment charges on mining hardware being written off. Beneath those charges, the strategic picture looks more promising: a five-year data center deal with Microsoft, a cloud services agreement with NVIDIA, a debt-to-equity ratio of just 1.9x, and full ownership of its land, electrical substations, and hardware. The company runs entirely on renewable energy, a feature that matters to hyperscalers.
The core tradeoff is legible. Applied Digital is cheaper on a valuation basis but more leveraged and more dependent on a handful of customers. IREN carries less debt and owns more of its own infrastructure, but it's mid-transformation, and its financial statements are clouded by one-time charges. Both companies are burning cash — Applied Digital's negative $2.8 billion and IREN's negative $2.2 billion in free cash flow reflect the brutal capital intensity of building data centers before revenue arrives to pay for them.
Neither company is profitable yet. The question for investors isn't which one will succeed — it's which path to profitability feels more credible, and which risks feel more manageable. Applied Digital offers a simpler model with higher leverage and concentration risk. IREN offers lower debt and vertical integration, but more moving parts and an unfinished reinvention. The answer depends less on the companies themselves than on what an investor is willing to hold through the uncertainty ahead.
Two companies are racing to build the physical backbone of artificial intelligence, and they're taking fundamentally different paths to get there. Applied Digital and IREN both operate data centers designed for the computational intensity of modern AI workloads—what the industry calls neoclouds—but their financial structures, operational models, and risk profiles diverge sharply enough that choosing between them requires understanding what kind of bet you're actually making.
Applied Digital has grown explosively. In its fiscal year ending May 2026, the company pulled in $611.3 million in revenue, nearly triple what it earned the year before. That kind of acceleration is what catches investors' attention. The company designs and operates infrastructure specifically built for high-performance computing and AI clients who need far more raw processing power than traditional data centers can provide. Its main play is simple: build the facility, lease it to customers like CoreWeave, and collect rent. But that simplicity masks a serious vulnerability. One customer accounted for 59 percent of the company's HPC segment revenue recently. Lose that contract, and the financial picture shifts dramatically. The company is also carrying substantial debt—a debt-to-equity ratio of 2.9x means $2.90 in total debt for every dollar of equity. Despite the revenue growth, Applied Digital lost $250.3 million in the fiscal year. Free cash flow was negative $2.8 billion, and stock-based compensation inflated reported cash generation so severely that it represented 245.5 percent of operating cash flow. The company is spending heavily to build capacity across North America, and most of that spending is being financed through debt.
IREN is attempting something more ambitious and more complicated. The company is actively dismantling its legacy Bitcoin mining business to pivot toward AI cloud services. For the fiscal year ending June 30, 2026, IREN generated $707 million in revenue, up 41 percent year-over-year. But the company reported a net loss of $702.6 million, with $638.8 million of that loss coming from non-cash asset impairment charges tied to writing down its mining hardware as it transitions away from crypto. The company has secured high-profile contracts—a five-year deal with Microsoft for data center facilities and a cloud services agreement with NVIDIA—that signal confidence from major technology players. IREN's debt-to-equity ratio stands at 1.9x, meaningfully lower than Applied Digital's. The company also owns its land, electrical substations, and hardware directly, creating what's called vertical integration. It operates on 100 percent renewable energy, a feature that appeals to hyperscalers like Microsoft.
The financial picture reveals the core tradeoff. Applied Digital is cheaper on a valuation basis—its sales multiple is lower—but it's also more leveraged and more dependent on a handful of customers. IREN carries less debt relative to its equity base, but it's in the midst of a costly transformation, and the massive impairment charges make its recent financial statements harder to parse. Both companies are burning cash as they build out new infrastructure. Applied Digital's negative free cash flow of $2.8 billion and IREN's negative $2.2 billion reflect the capital intensity of the business: you have to spend enormous sums upfront to build data centers before you can collect revenue from them.
The risks cut differently for each company. Applied Digital's concentration in North Dakota makes it vulnerable to regional regulatory changes and utility permitting delays. A disruption in the supply chain for cooling or power equipment could derail project timelines. IREN faces exposure to grid constraints in Texas and Australia, and any regulatory intervention targeting large-load energy users could constrain its ability to operate at full capacity. Both companies are also betting that the demand for AI infrastructure will remain robust enough to justify the debt they're taking on and the capital they're deploying.
For investors trying to choose, the decision hinges on which operational model you trust more. Applied Digital offers simplicity and predictability—it's a landlord collecting lease payments—but that comes with higher leverage and customer concentration risk. IREN offers vertical integration and lower debt, but it's still in transition, and the financial statements are clouded by one-time impairment charges. The company that owns its own land, power infrastructure, and hardware may have fewer external dependencies, but it also has to manage more moving parts. Applied Digital's higher debt-to-equity ratio adds risk in an uncertain interest rate environment, while IREN's lower leverage provides a cushion. Neither company is profitable yet, and both are spending billions to build capacity. The question isn't which one will succeed—it's which one's path to profitability is more credible, and which risks you're more comfortable carrying.
Notable Quotes
Applied Digital's lease agreements give it long-term revenue predictability, while tenants bear the brunt of outfitting data centers with AI hardware.— Investment analysis
IREN's vertical integration strength, lower debt-to-equity ratio, and smaller forward P/E ratio provide advantages over Applied Digital.— Investment analysis