Applied Digital vs Core Scientific Stock: AI Contract Test

Demand for artificial intelligence workloads has made power-connected data centers as valuable as the chips inside them. For investors comparing AI data center stocks, the key question isn't which company has the loudest backlog headline. It's whether signed contracts can turn into operating cash flow without crushing the balance sheet.

Applied Digital and Core Scientific both began in crypto infrastructure, yet their AI stories now look different. APLD has disclosed a larger contract book, while Core Scientific's case rests more on operating sites and AI-hosting delivery. Start with the contracts, then test the costs behind them.

Key Takeaways

• Applied Digital offers clearer contract visibility, with about 1.41 GW of leased capacity and a disclosed $36.2 billion base-term lease backlog as of May 31, 2026.

• Core Scientific's AI-hosting case depends more on converting existing crypto-mining infrastructure into reliable, high-density computing facilities.

• Long-term contract values are not the same as current revenue, cash flow, or profit; investors must assess commencement dates, power delivery, customer concentration, and contract terms.

• Both companies face substantial capital requirements, including cooling, power systems, construction, debt, and potential shareholder dilution.

• APLD may offer greater development and backlog visibility, while CORZ may offer a stronger operating starting point; neither investment case is risk-free.

AI data center stocks: contracts favor APLD's visibility

As of May 31, 2026, Applied Digital said it had executed long-term leases for about 1.41 gigawatts of critical IT capacity across five campuses. Those leases represented a disclosed revenue backlog of roughly $36.2 billion in future lease value over their initial 15-year base terms.

That is a large number, but the word "over" matters. It is not $36.2 billion of current revenue, cash, or profit. APLD reported fiscal Q4 2026 revenue of $258.7 million, a figure covered in this APLD results review. The gap shows how early the company remains in its build-out.

Contract question	APLD	Core Scientific (CORZ)
Headline disclosed lease backlog	$36.2 billion in future lease value over initial 15-year terms, as of May 31, 2026	No directly comparable company-wide backlog figure
Disclosed capacity	About 1.41 GW across five campuses	AI hosting contracts tied largely to its existing data center footprint
Main investment case	A large signed lease pipeline that must be built and energized	Converting operational mining infrastructure into reliable HPC hosting
Main pressure point	Construction funding, debt, power delivery, tenant concentration	Contract economics, customer reliance, site conversion, execution

By August 2026, third-party coverage had described Core Scientific's AI contracts as exceeding $10 billion, largely tied to major hosting arrangements. However, its public contract presentation is less uniform than APLD's lease backlog. This comparative APLD analysis is useful context, but investors should still read primary filings and individual contract announcements.

  A 15-year contract total measures future revenue potential. It does not tell you when each building opens, how much debt funds it, or what profit remains after power and construction costs.  

APLD's figures create clearer visibility. Core Scientific may have a stronger operational starting point at certain sites. That difference makes a straight backlog-versus-backlog comparison misleading.

Read the contract language before valuing the stock

A signed lease, a nonbinding memorandum, and an unbuilt capacity pipeline are very different things. They often get grouped together in stock market discussions, especially when AI enthusiasm is high.

A binding lease should identify the customer, MW capacity, lease term, commencement conditions, and pricing structure. It gives investors a real claim on future payments, although the company still must meet construction and power milestones.

Contracted revenue represents lease payments across the full term. APLD's $36.2 billion figure has value because it comes from executed leases. Yet timing still matters. A facility beginning service in 2028 contributes nothing to 2026 revenue.

Pipeline opportunities sit lower on the confidence scale. A land option, letter of intent, or negotiation with hyperscale cloud providers can become a signed contract, but it isn't contracted revenue. Prospective discussions aren't signed commitments. Follow-on announcements can also use words such as "expected" or "prospective." Keep those figures separate from signed base-term commitments.

APLD has also said about 70% of its contracted revenue is supported by U.S.-based investment-grade customers. That improves credit quality. At the same time, it highlights customer concentration. Losing one large tenant or delaying one deployment could affect several years of forecasts.

Core Scientific's operating base versus APLD's build-out

Former crypto miners have an advantage over blank-field developers: access to power, land, buildings, substations, and teams experienced in running energy-heavy sites around the clock. Those assets can provide a faster route into cloud computing and high-performance computing hosting.

Still, a Bitcoin mining hall isn't automatically ready for AI. Modern graphics processing units run far hotter and cost far more than mining rigs. Customers need redundant power, secure network links, hardened uptime standards, and layouts that support dense GPU clusters.

That makes thermal management a central requirement. Air cooling can work at lower rack density, but high-density AI hardware often needs liquid-cooling solutions. Cooling equipment, piping, pumps, and power distribution add capital expenditure before a site produces hosting revenue. GPUs, cooling systems, transformers, and networking hardware can also face supply chain bottlenecks, delaying site commissioning.

Core Scientific's appeal is its existing large-scale facilities, but AI contracts still test conversion and execution. An operating building doesn't guarantee AI-hosting revenue. APLD's story is more development-led, with official filings describing a business focused on HPC and cloud infrastructure. Future revenue still depends on bringing new campuses online on schedule, as outlined in this company filing.

NVIDIA remains the leading supplier of high-performance AI GPUs, while AMD is building a credible alternative. Neither CORZ nor APLD needs to beat those chip designers. Their job is to provide reliable, affordable places for expensive systems to run.

Where these companies fit in the AI infrastructure chain

AI data center stocks cover several different businesses. Holding one does not give you exposure to every part of the technology spend.

• Data center real estate investment trusts, including Digital Realty Trust and Equinix, own mature facilities and collect rent from many enterprise and cloud customers.

• Infrastructure and networking providers such as Vertiv Holding sell power, cooling, and networking equipment needed to keep dense server halls online.

• Chip companies and upstream suppliers including NVIDIA and AMD sell the processors that make AI workloads possible. The wider chain also includes semiconductor capital equipment and the high-bandwidth memory stack.

• Neocloud and hosting firms such as CoreWeave, APLD, and CORZ focus on GPU capacity, data center space, or both. Iris Energy is a related power-infrastructure peer with crypto-mining roots.

Digital Realty Trust and Equinix have more diversified rent-collection models than APLD and CORZ. The latter sit closer to the power-and-property end of the chain. Their value depends on electricity availability, financing, construction discipline, and long-term lease agreements.

Quantum computing is another interesting tech theme, but it has little direct bearing on today's GPU-hosting economics. A company can benefit from AI spending without having meaningful quantum computing exposure.

Capital expenditure is the opportunity and the danger

Large capital expenditure can create a moat. Power access, utility relationships, specialist equipment, and construction expertise can form a barrier to entry. It can also punish shareholders if costs rise before lease revenue starts.

APLD's reported debt of roughly $2.7 billion is therefore not a footnote. Investors should track debt maturities, interest costs, restricted cash, equity issuance, dilution, and whether construction relies on project finance or the corporate balance sheet.

Use these checks before treating a contract announcement as a reason to buy:

What to check	Healthier sign	Warning sign
Power delivery	A documented interconnection and clear service date	Land is available, but utility power remains uncertain
Contract quality	Named customer, MW amount, lease term, and binding status	A large "pipeline" number with few details
Funding	Debt is matched to project cash flows and construction milestones	Repeated share sales and rising borrowing costs
Customer mix	Several creditworthy tenants	One customer drives most projected revenue
Valuation	Market capitalization matches realistic future cash flow	A stock price assumes every planned site opens perfectly

Forward P/E can be a poor shortcut for companies still building facilities, because earnings may not arrive for years. Price-to-sales can also distort the picture when revenue ramps in steps as sites open. Equinix offers a mature data-center benchmark, but its model isn't directly comparable, and signed leases don't guarantee attractive operating margins. Look at operating cash flow, free cash flow, debt per energized megawatt, expected lease start dates, and the share count.

How to invest without treating a contract as a guarantee

If you're learning how to invest in tech infrastructure, position size matters more than finding the most exciting ticker. Demand from artificial intelligence chatbots is only one potential driver of compute demand. Project delays, customer changes, and financing pressure can produce substantial stock volatility.

Before deciding how to buy stocks, check which shares your broker offers, foreign-exchange costs, dealing fees, and whether fractional shares fit your plan. eToro and Trading 212 are common retail platforms in Europe, although account terms and available securities differ. You can Buy now on etoro, but the platform should not determine whether a stock suits your plan.

A small recurring contribution can reduce the temptation to chase a spike. Use a compound interest calculator to test what steady contributions might become over time. For an example of how a concentrated technology allocation can be presented openly, see this current portfolio breakdown.

Frequently Asked Questions

Which company has the larger AI data center backlog?

Applied Digital has disclosed roughly $36.2 billion of future lease value across about 1.41 GW of contracted capacity. Core Scientific has reported AI contracts exceeding $10 billion, but its contract disclosures are less directly comparable.

Is Applied Digital or Core Scientific the better AI data center stock?

Applied Digital has clearer contract visibility, while Core Scientific has an operating-base advantage at some existing facilities. The better choice depends on whether an investor prioritizes development growth or a more established infrastructure platform.

Does a signed AI data center contract guarantee future profits?

No. A contract does not guarantee that a facility will open on time, receive power, or generate attractive margins after construction, financing, and operating costs.

What should investors check before buying either stock?

Investors should review power-delivery dates, lease terms, customer concentration, debt maturities, construction funding, free cash flow, and dilution risk. They should also compare the market capitalization with realistic cash flows from energized and operational capacity.

Final thoughts

APLD currently provides the clearer contract-visibility story, led by its disclosed $36.2 billion base-term lease backlog. CORZ offers a more operational AI-hosting case, but investors need to inspect each agreement and site conversion more closely.

Unlike a mature operator such as Equinix, both stocks still need power to arrive, data centers to open, and customers to keep paying. A signed contract is the start of the investment case, not the finish.