Why a VA Software Inventory Is Not a License Position
The VA planned about $985 million for software in fiscal 2025, including commercial licenses. GAO, in a 25 March 2026 statement for the House Veterans’ Affairs Committee, found the department identified its five most widely used vendors by licenses installed (and still could not tell whether it was buying too many or too few). A VA software inventory that is not compared to purchase records cannot answer that. Initial functionality for a centralized inventory was due in late March 2026. GAO called that, if successful, “a critical first step.” A first step is not a position.
That’s the problem.
This is not a federal-only story, and it is not last year’s hearing. The May 2025 House Committee on Veterans’ Affairs oversight session is a different event. This is the March 2026 GAO statement: classic SAM against a nine-figure spend. The federal CIO software-license inventory mandate is the same work, written as policy rather than a finding. You cannot manage that estate if you cannot reconcile what is in use to what you bought. Inventory without purchase records is a library. Purchase records without inventory is a budget line.
What GAO actually found
GAO-26-109060 restates, then updates, two 2024 reports. In January 2024, GAO said that while VA identified its five most widely used software vendors with the highest quantity of licenses installed, “it faced challenges in determining whether it was purchasing too many or too few of these software licenses.” Specifically, VA was not tracking the appropriate number of licenses for each item of software currently in use, and it did not compare in-use inventories to purchase records on a regular basis. Both key activities: not met.
VA agreed with the 2024 recommendations and took preliminary actions. In early March 2026 it told GAO it planned initial functionality for a centralized inventory later that month. Implementation, GAO said, “would allow VA to identify opportunities to reduce costs on duplicate or unnecessary licenses.” A first step is not a comparison.
The five vendors, as of 31 July 2022, are in a footnote of the statement: Microsoft (identified twice), HCL Technologies, 1E, and Raytheon Technologies. Quantity installed is not a license position. VA could name the vendors. That is a discovery output. It is not an answer to how many you should hold.
The second finding is the cloud. In November 2024, GAO tracked restrictive software licensing practices: “any software licensing agreements or vendor processes that limit, impede, or prevent agency efforts to use software in cloud computing.” Those practices either increased the cost of cloud software or services or limited the department’s options when selecting a cloud service provider. VA had not established guidance for managing those impacts, and it had not determined who is responsible for them.
If the challenges continue, GAO warned, the department “may jeopardize its ability to effectively manage its software licenses that provide critical services to veterans.” VA expected further actions on the restrictive-practices working group by 30 September 2026. As of March 2026 it had not provided an update. GAO did not publish a waste figure for the whole estate. VA’s own projection ($136 million in cost avoidance from fiscal 2025 through 2029, on one vendor) is not a recovery percentage you can copy onto a slide.
A software license inventory has to answer both
Look at your own file. Most SAM teams still have a row that says “installed” and a purchase ledger that says “bought.” Those two words describe different things. Until you compare them on a regular basis, you cannot say whether you are over-licensed or exposed. That comparison is audit defense and cost optimization on the same page: too few is a true-up, too many is waste. Too many and too few are not two projects. They are the same missing reconciliation.
Picture the room. Your CIO wants to know, this week, whether you are over or under on the top five. Finance wants a number against a spend class that, at VA scale, is about $985 million. The vendor will walk in with a quantity close to what is already deployed and treat that as demand. If SAM takes three weeks to recast in-use inventory against purchase records, the vendor’s number becomes the file.
This is not a criticism of the teams building those reports. Traditional SAM tools collect comprehensively and answer slowly. They prove completeness. They do not tell you, this week, whether the thing you installed is the thing you should still be paying for. The gap is interpretation work.
If that feels familiar, the problem is not your team’s maturity. A commercial question is being answered with an inventory project. As of March 2026 VA had assessed its top 15 licenses and begun moving them to enterprise agreements. Policy and procurement guidance were drafted, not approved. Procedures to compare in-use licenses to purchased licenses were still missing. A centralized inventory without that comparison is still a library.
Restrictive cloud licensing is already in the contract
The cloud finding is not a migration story. It is a contract story. Restrictive practices raise the cost of cloud software or limit which provider you may use: extra fees on another CSP, a conversion or repurchase to take on-prem licenses to the cloud, lock-in via terms. The cost of the move is not only compute. It is the license term you already signed.
VA had not written guidance and had not named an owner. If BYOL is more expensive in one CSP than another, that is a commercial term, not an architecture decision. A September 2026 deadline is not a position. A named owner and a read of the clauses. Those are a position.
What to reconcile before the next conversation
Do this in the next thirty days, as a decision pack rather than a project.
| What to reconcile | Why it has to be live |
|---|---|
| In-use inventory vs purchase records | Until you compare the two regularly, you cannot say too many or too few. |
| Quantity installed vs licenses you need | VA named its top five by licenses installed. Installed is discovery. Need is a position. |
| Cloud terms that raise cost or limit CSP choice | Restrictive practices limit, impede, or prevent using the software in the cloud. Cost them before you pick a region. |
| A named owner for those impacts | VA had not established guidance or determined who is responsible. |
| Duplicate or unused licenses you would stop buying | GAO’s opportunity: reduce costs on duplicate or unnecessary licenses. No invented savings. Use VA’s $136 million one-vendor projection as a reminder, not as your number. |
Then say, in one page, whether the next conversation is a cleanup, a true-up, or the same buy again.
Can a non-specialist walk the CIO through that pack today? If the answer is no, you do not have a software inventory problem only. You have a library and a nine-figure reading list.
The decision layer, not another inventory
You already have discovery data, purchase records, and (if you are where VA was in March) a centralized inventory on the way. The gap is not another inventory. The gap is turning that estate into a decision: what is in use, what you bought, which cloud terms raise the cost of the move, and what should happen before anyone picks up the phone. LICENSEWARE sits on the inventory and ITSM tools you already run. It is not a rip-and-replace SAM suite. It is a decision layer: what matters, why it matters now, what should happen next. Software Inventory Manager is the pack that turns installed into in-use.
If you are heading into a renewal or a cloud move this year and your current tools still need three weeks to turn “installed” into too many or too few, book a Software Intelligence Review. You can also start on the free plan and run the analysis on your own data.
The question to walk in with
Do not let the next meeting open as “we need a better inventory project.” The inventory is the input. The right question is: what are we actually using, what did we buy, and which contract terms raise the cost of the cloud we just chose or limit who can run it. That is a data question, not a sales question.
$985 million is a budget line. It is not a license position. Too many and too few live in the same gap. Restrictive cloud terms live in the same file. The vendor will walk in with a number. The only question is whether you have yours first: current, defensible, and tied to the contract in front of you.
