HomeProcurementFBI's Enterprise AI Compute Hardware IDIQ Would Set Gemini Terms Before GSA's...

FBI’s Enterprise AI Compute Hardware IDIQ Would Set Gemini Terms Before GSA’s Deal Lapses

FBI's own Chief AI Officer has called AI compute pricing steep, and this still-unawarded vehicle would set Gemini licensing terms on FBI's own schedule, not GSA's. For systems integrators, the eventual award matters more than exactly when the window closed.

FBI’s Enterprise AI Compute Hardware IDIQ Plans Multiple Winners, Not One

FBI is buying enterprise AI compute hardware through a live, multiple-award opportunity, not a single sole-source deal. Solicitation 15F06726R0000307 plans two to four separate winners. That leaves room for a mid-market AI-hardware reseller or systems integrator, not only the largest federal systems houses. The bar is the right OEM relationships. FBI amended the solicitation, “Enterprise GPU Servers for Artificial Intelligence,” in August 2026. The amendment released the government’s answers to offeror questions and revised solicitation documents. The aggregate ceiling is $88 million, not obligated. Each awarded contract carries a $2,000 minimum guarantee. The planned contract start is September 1, 2026. No award has been made yet.

The proposal deadline remains uncertain, with August 17 and August 25, 2026 reported as possible closing dates. Until the deadline is resolved, firms should treat the opportunity as time-sensitive. That uncertainty does not change the immediate capture question: which agencies are developing similar AI-hardware requirements, and can your firm position for that work before competitors do?

Field Answer
Signal FBI’s amended $88M multiple-award IDIQ for enterprise AI compute hardware, with a proposal deadline of either Aug. 17 or Aug. 25, 2026, and 2-4 winners planned
BD implication New-business capture through direct bid (if the window is still open) for AI-hardware resellers/systems integrators; a single documented instance, not yet an established pattern, worth watching at other agencies
Customer FBI’s enterprise AI infrastructure requirement; no program office or division identified beyond Chief AI Officer Katie Noyes’s public statements
Buyer Federal Bureau of Investigation, Department of Justice
Funding $88M ceiling, not obligated; $2,000 minimum guarantee per awarded IDIQ once contracts are signed
Vehicle / path Solicitation 15F06726R0000307, multiple-award (2-4), firm-fixed-price IDIQ
Incumbents / ecosystem None publicly disclosed; no predecessor contract or incumbent vendor has been named
Access strategy Direct bid, if the window remains open, or pre-positioning OEM-channel authorizations now for teaming once winners are named
Timing Proposal deadline remains unsettled between Aug. 17 and Aug. 25, 2026; the planned Sept. 1, 2026 contract start and award announcement are the next key events
Confidence Moderate
Pursuit posture WATCH
Upgrade triggers Resolution of the proposal deadline; the award announcement naming winners; a comparable bespoke AI-hardware IDIQ at another agency
Downgrade triggers Evidence this solicitation is canceled or folded back into a GSA-managed vehicle

The Solicitation’s Scope Covers the Full AI Hardware Stack, Plus Gemini Licensing

This solicitation isn’t just servers. It covers the full stack an enterprise AI deployment needs: compute hardware, networking, software licensing, and support. That breadth is why the eligibility bar matters as much as the dollar figure. To compete for the full scope, a bidder needs OEM relationships across more than one product category. The multiple-award structure softens that, since a firm can win on a single category. The scope:

  • Enterprise AI compute servers and expansion hardware
  • Rack-scale and pod-scale AI computing systems
  • AI inference server systems
  • High-speed networking equipment, including network interface cards and data processing units
  • OEM software and licensing, including Google Gemini on-premises licensing under a capacity-based allocation model
  • OEM technical-support and warranty coverage

Three eligibility rules cut the bidder pool:

  • Every product must be new, genuine, and sourced through an authorized OEM channel. No refurbished, gray-market, or used equipment.
  • Every award carries a minimum three-year OEM warranty, with the government registered as the end user.
  • Support entitlements must survive contract closeout, not expire with it.

None of that is unusual on its own. New-material and authorized-channel terms are standard practice across DoD and civilian IT acquisitions. The Gemini licensing requirement is what narrows the field here. A bidder needs an authorized AI-hardware OEM relationship and the ability to deliver Google’s on-premises licensing terms. Set-aside status for this solicitation is not publicly disclosed. Treat it as unresolved, not as full-and-open by default.

FBI’s Own Cost Complaint Lands Weeks Before GSA’s Gemini Discount Expires

FBI’s Chief AI Officer, Katie Noyes, has described current AI compute pricing as “a little steep” for the Bureau. The Bureau has more than 100 approved AI use cases, with adoption doubling year over year through 2024-2026. That adoption curve provides a plausible business context for the vehicle’s timing, although the FBI has not publicly stated that it is the reason for the acquisition.

GSA’s OneGov program negotiates government-wide AI pricing deals. Most cover software and model access, not hardware. ChatGPT, Claude, and Google’s Gemini for Government were available under OneGov at promotional rates as low as $0.47 to $1 per agency for a year. That promotional pricing expires September 30, 2026. Outgoing Federal CIO Greg Barbaccia has said the next rate won’t be the same: “I don’t think it’s going to be $1. I think it’ll be transparent at least.” Read that as a signal the near-free rate ends. Barbaccia didn’t promise strict usage-based billing; don’t read it that way.

FBI’s own IDIQ separately puts Google Gemini on-premises licensing in scope, on FBI’s own capacity-based terms. That holds only if the vehicle is awarded. Performance is planned to start September 1, 2026, a month before GSA’s Gemini pricing lapses. No award has been made. That start date is a plan, not a certainty. Set against a possible August 25 deadline, evaluation and award would need to move fast to hit it.

GSA’s OneGov deals were never a path to buying GPU servers or rack-scale AI hardware. Whether OneGov’s Gemini deal even covers on-premises, capacity-based licensing isn’t established either. So the comparison is about timing, not substitution. FBI’s own Gemini terms would take effect on FBI’s own schedule, right as the centrally negotiated Gemini rate expires. Deliberate positioning ahead of a price increase is one reading. Coincidental timing tied to FBI’s own infrastructure needs is another. Neither explanation is established. The timing is nevertheless a useful signal of how agencies may increasingly pursue AI model licensing and infrastructure through agency-specific acquisition vehicles rather than relying solely on government-wide arrangements.

FBI’s amended IDIQ caps this multiple-award vehicle at $88 million across two to four contracts, with only a $2,000 minimum guarantee per award. (Solicitation 15F06726R0000307)
“All products shall be new, genuine, commercially available OEM items sourced through authorized channels.” (FBI solicitation 15F06726R0000307, eligibility terms)
Path What it covers What FBI chose for this buy
GSA OneGov Centralized AI software/model pricing (ChatGPT, Claude, Gemini for Government); no GPU or AI-server hardware Covers AI software/model access pricing; whether FBI holds or uses OneGov’s Gemini access isn’t established
Agency-specific acquisition vehicle A bespoke hardware-and-licensing vehicle one agency builds and controls directly This solicitation (15F06726R0000307)

No Bidder Is Named, but the Vendor Categories Eligible to Win Are Identifiable

No bidder or awardee has been named publicly. FBI has announced no award. Eligibility rules out most of the market on its own. A winner needs an authorized-channel relationship with an enterprise AI-hardware original equipment manufacturer, or with an AI product firm building on top of one. It has to deliver Google’s Gemini on-premises licensing terms. It must also back a three-year OEM warranty with the government as registered end user.

NVIDIA, Dell, HPE, and Supermicro are among the enterprise AI hardware OEMs whose GPU and accelerator server lines plausibly fit this solicitation’s scope. Naming a category isn’t the same as naming a bidder. Which of these manufacturers submitted a proposal is not publicly disclosed. The same goes for their authorized federal resellers and systems integrators.

A multiple-award structure changes the competitive picture in three ways:

  • Two to four winners means a bidder does not have to beat the whole field to take a slot.
  • Once task orders start, a winner may still need a subcontractor for a product category outside its core strength.
  • No single company controls the full $88 million ceiling, so task-order activity would spread across several winners rather than one.

For this requirement, no incumbent contract or vendor is publicly disclosed. FBI’s public contract history runs to routine IT hardware and maintenance awards at a much smaller scale. One example is a $5.8 million F5 hardware maintenance award to V3GATE, LLC, with $1.3 million obligated at award. The NGAGE-scale AI-compute requirement represents a materially larger and more specialized acquisition than the routine federal IT hardware and maintenance work previously associated with the FBI. That fits a new requirement category, not a recompete of prior work.

What a Systems Integrator or AI-Hardware Reseller Should Do Now

Your next move depends on two things. Is the proposal window still open, and did you already have a bid in?

  1. Resolve the live deadline for solicitation 15F06726R0000307 and review the latest amendment and Q&A responses before deciding whether a direct bid remains possible.
  2. If the window is closed and you didn’t bid, position for teaming now. Confirm your authorized-channel status with the AI-hardware OEMs named above, and confirm you can deliver Gemini on-premises licensing terms, so you are ready when the winners start looking for delivery partners.
  3. Monitor for the award announcement under solicitation 15F06726R0000307. Because the vehicle represents a potentially repeatable agency-specific AI-hardware acquisition model, also track comparable AI-compute IDIQs at other agencies using terms such as “enterprise AI compute hardware” and “AI compute servers.”
  • Your firm holds an authorized-channel relationship with at least one of the AI-hardware OEMs named above (NVIDIA, Dell, HPE, Supermicro, or a comparable enterprise AI-server manufacturer).
  • You can back a three-year minimum OEM warranty with the government registered as end user, and can deliver Google Gemini on-premises licensing terms.
  • Your firm has past performance or a comparable track record at the scale this IDIQ implies, not just at the scale of routine federal IT hardware/maintenance work.
Resolve solicitation 15F06726R0000307’s live deadline today. Whether or not you can still bid, pull your OEM-channel authorizations against this solicitation’s hardware and Gemini-licensing scope now. Then you can approach the winners as soon as the award is public. Teaming conversations are easiest before task orders start, not after.

FAQ

Which FBI office runs this acquisition?

Beyond Chief AI Officer Katie Noyes’s public statements, no program office, division, or program officer has been named publicly. The solicitation’s designated point of contact should be treated as the appropriate channel for bidder communications. Do not assume the Chief AI Officer’s office handles bidder communications.

Is this a full-and-open competition, or does it carry a set-aside?

The set-aside status remains unresolved. Bidders should not assume either full-and-open or small-business status until the solicitation’s applicable acquisition status is established.

How restrictive is this solicitation compared to a typical federal hardware buy?

It is less restrictive than it first appears. New-material and authorized-channel sourcing, plus a multi-year OEM warranty, are standard terms across DoD and civilian IT hardware acquisitions. They are not unusual restrictions unique to this IDIQ. What narrows the field is the added Google Gemini on-premises licensing requirement layered on top of them.

Is there a real subcontracting opportunity here for a smaller AI-hardware reseller?

Possibly, but that rests on the winners’ own sourcing decisions once awarded. No subcontracting plan is publicly disclosed for this solicitation. A firm with authorized-channel ties to the OEMs above has the strongest basis for outreach once winners are named.

What would confirm or break the Gemini-timing read of this contract?

A GSA or FBI statement directly linking this IDIQ’s Gemini licensing to OneGov’s pricing expiration would confirm it. A second agency licensing an AI model through its own hardware vehicle near the same expiration window would strengthen the pattern. Neither exists today. All that supports the read is the timing itself and FBI’s own cost complaint.

What’s the next event to watch for?

The live proposal deadline is the immediate decision point. After that, the award announcement naming FBI’s 2-4 winning contractors becomes the key event. The award is expected around or after the planned September 1, 2026 contract start date.

BD classification: WATCH. Confidence: Moderate. Evidence that would upgrade this: resolution of the proposal deadline, the award announcement naming winners, or a second agency running a similar AI-hardware IDIQ that licenses an AI model outside GSA OneGov. Evidence that would downgrade this: cancellation of the solicitation or its scope folding back into a GSA-managed vehicle.

The Contract Opportunity Atlas

Two issues a week.. Free.

Two issues a week. Data-driven intelligence for small tech firms selling to the federal government. Free.

Subscribe to Contract Opportunity Atlas

Get federal technology, AI, procurement, and GovCon insights delivered to your inbox.

Shahid Shah
Shahid Shah
Shahid specializes in bringing world-class CTO, CISO, and EiR expertise to startups, business units and companies on a part-time (fractional) basis. With a rich background in regulated, safety-critical industries like Med Devices, Digital Health, and Gov 2.0, he possess a unique understanding of complex, high-demand products and services. He is a C-suite native that can easily blend in with technical and engineering teams that need to deliver revenue-generating solutions to the marketplace. He has served as an Entrepreneur in Residence when a market seems lucrative but it's unclear how to build and launch products and services for such opportunities. Shahid has years of leadership experience as a co-founding startup CTO for multiple venture-backed companies, business unit CTO and EiR, and public company CTO helping transform product teams from marginal to high performance. His software/hardware engineering and cybersecurity body of knowledge is up to date because he rolls up his sleeves to create code when appropriate & dive into system architecture and design when required. He also conduct technology due diligence exercises for corporate acquisition or product integration requirements.
RELATED ARTICLES

Most Popular

CATEGORIES