Microsoft Locks $9.7B IREN Deal for Nvidia GB300 Cloud Power

What happened

Microsoft locks a $9.7B IREN deal for Nvidia GB300 cloud power, agreeing a five-year cloud services contract that gives it access to Nvidia’s upcoming GB300 processors through IREN’s data center platform. IREN said the arrangement includes a 20 percent prepayment and staged deployments through 2026 at its 750 megawatt campus in Childress, Texas. The announcement sent IREN shares sharply higher in early trading. Reuters reported that IREN also entered a separate agreement with Dell Technologies to procure chips and ancillary gear worth about $5.8 billion.

Deal terms and structure

IREN describes the pact as a cloud services contract, not a simple hardware purchase. Under the five-year term, Microsoft secures capacity on systems built around Nvidia GB300 accelerators that IREN plans to install in phases. The 20 percent prepayment helps fund procurement and buildout, lowering financing risk as equipment is ordered and delivered. Reuters adds that Dell will handle a large portion of the acquisition and integration of servers and related infrastructure for IREN, an arrangement that can shorten lead times and simplify logistics.

Why it matters for Microsoft

The tie-up supports Microsoft’s effort to meet surging demand for AI training and inference on Azure. Access to Nvidia GB300 capacity through IREN supplements Microsoft’s own builds and helps bridge timing gaps as new campuses and power connections come online. It also diversifies sourcing by using a specialist operator with an existing site and procurement pipeline. Microsoft has been active in securing compute and real estate across partners and consortia, including participation in a $40 billion investor group to acquire a large data center operator, a sign of how critical guaranteed capacity has become.

Why it matters for IREN

For IREN, the contract validates its campus strategy and underwrites a significant portion of its growth plan. Long-dated commitments from a top tier cloud buyer can lower the cost of capital, improve supplier terms, and support phase-by-phase construction. The market reaction echoed that view, with Reuters and financial wires noting a double-digit premarket jump in IREN shares after the deal was disclosed.

Capacity, site, and timeline

The Childress site is slated for up to 750 MW, a scale that suits dense Nvidia deployments and the power and cooling footprints that follow. IREN said GB300 systems are scheduled to roll out in phases through 2026, which aligns with broader industry patterns as utilities, grid operators, and vendors coordinate delivery and interconnection. A phased schedule allows earlier revenue recognition on initial halls, then ramps as power, networking, and racks fill out subsequent buildings.

Chips and the competitive backdrop

Nvidia remains the reference platform for large model training and premium inference. The GB300 generation is expected to push performance per watt and cluster-level efficiency, which can lift useful throughput for tenants and cloud providers. Microsoft has its own silicon roadmap, yet near-term demand exceeds internal supply, so securing Nvidia-based capacity remains central. Recent moves by Nvidia to allocate large volumes to national and enterprise programs highlight the scramble for top chips across regions.

Financing and procurement mechanics

Dell’s role, as reported by Reuters, is to purchase chips and ancillary equipment for about $5.8 billion, then supply IREN. This method spreads working capital needs, taps an integrator’s volume discounts, and concentrates configuration know-how with a large OEM. For Microsoft, dealing with a single operator that has lined up procurement and integration can compress timelines from purchase order to usable compute.

Strategic context in Europe and the U.S.

The Microsoft $9.7B IREN deal for Nvidia GB300 cloud capacity sits within a wider buildout of AI infrastructure. Microsoft has outlined large regional investments, for example in Italy, and has backed efforts to expand data center footprints globally. In parallel, Italy and other European markets have seen a wave of campus announcements and policy attention on power, land, and permitting, showing how geopolitics and energy policy now shape cloud expansion.

Risks and sensitivities

Execution risk includes long-lead electrical gear, grid interconnection, and utility timelines that can slip. Supply risk remains present until Nvidia and its partners deliver GB300 volumes at scale. Pricing and allocation dynamics could shift if demand outpaces supply for longer than expected, or if alternative accelerators gain traction. Regulatory risk includes evolving rules for power sourcing, data center siting, and export controls on high end chips and networking components, which can affect deployment schedules and locations.

Investor read-through

For Microsoft, contracted access to state-of-the-art accelerators supports Azure growth and AI product attach, which underpin revenue visibility for fiscal 2026 and beyond. For IREN, a marquee tenant provides cash flow clarity, which can support additional financing and expansions. For Nvidia and integrators such as Dell, the agreement is another indicator that the next wave of demand is being locked in ahead of delivery. Markets will watch for milestones on prepayments, first-rack deliveries, and power energization at Childress.

Bottom line

Microsoft locks a $9.7B IREN deal for Nvidia GB300 cloud power to secure near-term AI compute, while IREN gains a long-dated anchor tenant to fund and scale a 750 MW Texas campus. The agreement reinforces how capacity certainty, not just chip design, is the currency of this phase of the AI buildout.