2 AI Data Center Stocks to Buy Now

The artificial intelligence (AI) boom has created a land grab for data center capacity, but the real scarcity isn't chips -- it's electricity and the grid connections to deliver it at scale. Training frontier models requires clusters consuming hundreds of megawatts, more power than most cities allocate to their entire industrial base.

Hyperscalers have capital but face years-long queues for utility interconnections. Independent operators that secured power rights early or converted existing sites are capturing massive long-term contracts by solving the constraint that everyone else is still figuring out.

Image source: Getty Images.

Two stocks stand out for contracted megawatt capacity rather than speculative buildouts: a purpose-built campus operator with 15-year leases and fixed minimum payments from hyperscalers, and an independent hosting platform that rejected a $9 billion merger to preserve flexibility across multiple customers.

Both benefit from the same dynamic -- enterprises moving AI workloads from proof of concept to production need reliable infrastructure more than they need the newest chip generation. Here's why these two AI data center stocks are worth buying right now.

The long-dated, contracted megawatt landlord

Applied Digital (NASDAQ: APLD) purpose-builds AI campuses and signs roughly 15-year leases with fixed minimum payments, creating revenue visibility that most AI infrastructure plays lack. In October 2025, for instance, Applied signed an approximately $5 billion, 15-year lease for 200 megawatts with a U.S. investment-grade hyperscaler at its Polaris Forge 2 campus, with a right of first refusal for another 800 megawatts at the site.

Earlier in 2025, Applied expanded the footprint for CoreWeave to a fully leased 400 megawatts at Polaris Forge 1, with the first 50-megawatt phase now ready for service. Total leased capacity stands at around 600 megawatts across both campuses, with contracted commitments that de-risk financing and scale-up. For context, a typical AI data center needs 50 to 100 megawatts to run thousands of high-end graphics processing units (GPUs).

What's the takeaway? Applied Digital gets paid for 15 years, whether customers use the capacity or not. By contrast, GPU cloud developers will eventually face share pricing pressures when chip supply catches up with demand. Applied Digital, on the other hand, simply collects rent in a highly visible manner over a long time frame.

The independent, power-heavy AI host

Core Scientific (NASDAQ: CORZ) operates one of the largest U.S. data center platforms with substantial AI hosting under contract. The company has roughly 200 megawatts of CoreWeave hosting on multiyear terms and has outlined plans to deliver around 900 megawatts of gross AI hosting capacity supported by roughly 1.3 gigawatts of contracted power, while maintaining around 400 megawatts for Bitcoin operations.

On Oct. 30, 2025, shareholders rejected CoreWeave's roughly $9 billion all-stock acquisition offer, and Core terminated the merger -- remaining independent and free to pursue new AI contracts and financing on its own terms. This independence matters because Core can now negotiate with multiple hyperscalers and AI companies rather than being locked into a single customer relationship.

The bottom line? Core rejected a $9 billion buyout to stay independent. Now it can play multiple hyperscalers against each other for scarce power capacity instead of being locked to a single owner. The hybrid model -- keeping profitable Bitcoin mining while adding AI hosting -- funds expansion without constant capital raises.

The contracted capacity advantage

Applied Digital's long-term leases with fixed minimum payments provide maximum revenue visibility but concentrate risk among a handful of large customers. Meanwhile, Core Scientific faces execution uncertainty in converting Bitcoin sites and signing new AI tenants. Still, both companies benefit from the same fundamental constraint: Power capacity matters more than software features when enterprises need infrastructure that scales beyond prototype deployments.

The physical infrastructure buildout determines how fast AI scales, and these two stocks provide direct exposure to contracted megawatts rather than speculative capacity additions. While everyone debates which model architecture wins, Applied Digital and Core Scientific are building the power-heavy facilities that all AI applications need, regardless of which training technique becomes dominant.

George Budwell has positions in Bitcoin. The Motley Fool has positions in and recommends Bitcoin. The Motley Fool has a disclosure policy.