A $9 Billion Pentagon Cloud Contract Went Largely Unspent. It Is Being Replaced by a $21.6 Billion One.

Abhishek DashUpdated October 25, 20265 min read
A cavernous data hall at dusk with dust-sheeted server racks on both sides, a single lit cart holding boxes and a clipboard, and open steel doors revealing a larger construction hall beyond under a five-sided skylight.

In short

The Joint Warfighting Cloud Capability has obligated roughly 13 to 19 percent of its 9 billion dollar ceiling, with ordering authority running to 7 December 2027. Combined potential value of issued task orders is about 4.67 billion dollars of a 9.07 billion dollar ceiling. AWS holds an estimated 56 to 72 percent of obligated and potential value despite placing 81 task orders against Microsoft 83, while Microsoft holds 28 percent of obligated value and Oracle and Google hold 11 and 5 percent. The replacement vehicle, JWCC Unified Cloud Marketplace Core, carries a 21.6 billion dollar ceiling over up to ten years, roughly 2.4 times the current contract, and expands scope to anything-as-a-service including SaaS and a third-party marketplace.

Key takeaways

  • Only 13 to 19 percent of the 9 billion dollar JWCC ceiling has been obligated, with ordering authority through December 2027
  • AWS holds an estimated 56 to 72 percent of value on 81 task orders, against 83 for Microsoft, which holds 28 percent of obligated value
  • Order volume rose from 7 in 2022 to about 67 a year in 2024 and 2025, then fell to 16 through April 2026 while average order size grew
  • The replacement UCM Core carries a 21.6 billion dollar ceiling and adds SaaS and a third-party marketplace to hyperscale infrastructure
  • Navy consolidated all department usage per cloud provider onto single task orders in June 2026 to chase volume pricing
  • JWCC Next was described as a replacement rather than a supplement, with longer terms intended to end five-year transition cycles

The obvious reading of a $9 billion contract that is two-thirds unspent is that demand was overstated. The record points somewhere less comfortable: the contract was too narrow, and the Pentagon is now buying a much larger one that fixes the narrowness.

The numbers

JWCC, the Joint Warfighting Cloud Capability, has a ceiling of about 9.07 billion dollars. Somewhere between 13 and 19 percent of it has been obligated. Ordering authority runs to 7 December 2027, which at the time of the data reviewed left roughly 16 months.

Combined potential value of issued task orders is about 4.67 billion dollars. So even counting every order at full potential rather than actual obligation, half the ceiling is untouched.

Order volume tells a story of its own. Seven orders in 2022. Forty-one in 2023. Roughly sixty-seven a year across 2024 and 2025. Then sixteen through April 2026.

That looks like a slowdown until you pair it with the money. FY2026 obligations reached 606.1 million dollars through the data cutoff, against 340.4 million for the whole of FY2025. Fewer orders, more money, so average order size roughly tripled.

Read together, the pattern is agencies consolidating. Instead of many small task orders, a few large ones. That is consistent with what the Navy did in June 2026, when its Neptune programme consolidated all department usage per cloud provider onto a single task order each, explicitly to chase volume pricing and faster acquisition.

Where the money actually is

The concentration is the most striking figure in the record.

AWS placed 81 task orders. Microsoft placed 83. Nearly identical counts. But AWS holds an estimated 56 to 72 percent of obligated and potential value, while Microsoft holds 28 percent of obligated value, with Oracle at 11 percent and Google at 5.

Order counts and dollars have almost no relationship here. That means the difference is order size and workload type, which in a defence cloud context means the large, long-running, compute-heavy programmes landed with one provider and the smaller application workloads spread across the others.

The sources reviewed do not decompose the gap further. Anyone who tells you AWS won on price has not shown it from this record.

Why replace something underused

A contract at 13 to 19 percent utilisation with 16 months left might be judged a failure of forecasting. DISA's response has been to recompete at 2.4 times the size.

The replacement, JWCC Unified Cloud Marketplace Core, carries a 21.6 billion dollar ceiling over up to ten years. It expands scope to what DISA describes as anything-as-a-service, including SaaS and a third-party marketplace. A second vehicle, UCM Premier, adds tiers for non-hyperscale infrastructure and small business innovators.

That combination is the tell. If demand had simply been lower than forecast, the answer would be a smaller ceiling. A larger ceiling plus a broader catalogue implies the original vehicle could not sell what agencies wanted to buy. Hyperscaler-only infrastructure, with no marketplace and no SaaS, forced everything outside that shape to be procured elsewhere.

That is an inference from the structure rather than a stated rationale, and it should be held that way. But it is the only reading that makes the 2.4x increase coherent rather than reckless.

The stated reason was transition cost

DISA gave a different justification: contract length. It said JWCC Next would supersede JWCC with longer terms, specifically to avoid the five-year transition cycles that come with five-year vehicles.

That is a real problem. Recompeting a defence cloud contract is expensive and disruptive, and a ten-year term trades flexibility for stability. Whether that trade is right depends on how fast the underlying market moves, and cloud infrastructure has been moving fast enough that a decade is a long commitment.

Both explanations can be true. Longer terms reduce churn, and a broader catalogue captures spend that was leaking. Neither requires the other.

The AI question

AI appears in JWCC task orders, but it is not the driver of the spend.

Recent orders reference AI and machine learning tooling in Google landing zones, AI solutions for cybersecurity, and cloud for the F-35 programme. AWS's GovCloud announcements highlight FedRAMP High authorisations for foundation models, which is the model-level authorisation pattern spreading across government procurement.

But the bulk of obligated value is general cloud infrastructure. The framing is that JWCC is an infrastructure vehicle that happens to carry AI workloads, not an AI procurement programme. Treating it as the latter would misread where the money is.

Where this could be wrong

The utilisation figure is given as a range of 13 to 19 percent, which is a wide band and reflects differences between obligated value and potential value across sources rather than a single measurement. The precise number matters for the argument and is not settled here.

The concentration range for AWS, 56 to 72 percent, is similarly wide, and the gap between those bounds is large enough that the strongest version of the claim, that AWS holds nearly three quarters of the value, is not established.

The claim that scope rather than demand was the constraint is an inference. DISA has not said it in those terms in the sources reviewed. It is consistent with the recompete structure and with the Navy consolidation, and it remains an inference.

And whether UCM Core is a new acquisition or JWCC Next rebranded is unresolved. The August 2025 plan anticipated a solicitation in the first quarter of calendar 2026 and an award in early 2027. The September 2026 release, with bids due 6 October 2026, may be that acquisition under a new name or a different one entirely.

The useful point

A contract that is two-thirds unspent is usually read as evidence that the forecast was wrong. Here it reads as evidence that the catalogue was wrong, and the remedy is a vehicle more than twice the size with a marketplace bolted on.

The concentration figure is the one to watch. If the marketplace works as advertised, AWS share should fall as smaller providers and SaaS vendors get a route in. If it holds near 56 to 72 percent through the next vehicle, then the concentration was never about the contract structure, and the recompete will have bought a bigger ceiling without changing who gets the money.

On this page

Sources

  1. Joint Warfighting Cloud Capability contract recordsGovTribe, 2026

Frequently asked questions

Why replace a contract that has not been fully used?

The record suggests scope rather than demand was the binding constraint. JWCC was hyperscaler-only and had no third-party marketplace, so agencies wanting SaaS or smaller providers had to buy outside it. The replacement adds both. That is an inference from the ordering pattern, and DISA has not stated it in those terms in the sources reviewed.

Is AWS winning because it is cheaper?

The order counts do not explain the value gap. AWS placed 81 orders and Microsoft 83, yet AWS holds an estimated 56 to 72 percent of value against 28 percent for Microsoft. That points to order size and workload type, not order frequency, and the sources reviewed do not decompose it further.

What does the marketplace change?

It opens the vehicle to third-party offerings and SaaS rather than hyperscale infrastructure alone. Whether it changes concentration is unresolved. DISA talks about increased competition and equitable participation for small business, and a marketplace can either broaden access or entrench the incumbent storefronts.

Is JWCC Next a new contract or a rebrand?

Unresolved. DISA said in August 2025 that JWCC Next would supersede JWCC, with a solicitation planned for the first quarter of 2026 and an award in early 2027. The September 2026 release is titled JWCC Unified Cloud Marketplace Core with bids due 6 October 2026. The sources reviewed do not establish whether that is the same acquisition renamed.