Why State Procurement Grew 38% and What It Means

The title of this piece was commissioned as 'Why State Procurement Grew 38% and What It Means'. That is a title-bank sentence. It is not, by itself, a fact. Repeating it in a pitch deck is how a rounding error becomes a market, and how a GeM-platform movement becomes a story about every rupee a state spent on anything. This article exists to stop that slide.

Here is the honest version as of 17 August 2026. We looked for a primary public source. A Press Information Bureau release on GeM, PRID 2249335, dated 6 April 2026, has been widely quoted as saying GeM's cumulative GMV had reached about ₹18.4 lakh crore, that FY 2025-26 contributed a large increment (figures around ₹5 lakh crore have been repeated in secondary reports), and that procurement by States recorded a growth of 38.3 percent during FY 2025-26. Open that page. If the live text differs, the live text wins. If the page is gone, do not keep the percentage alive from memory.

Even if the 38.3 percent sentence is still on the PIB page, it is a GeM sentence. It means: state buyers on GeM spent more through GeM this year than last, on GeM's definition of state procurement. It does not mean every finance department's total procurement grew 38 percent. It does not mean IT spend grew 38 percent. It does not mean agentic-AI spend grew 38 percent. It does not mean your category grew 38 percent. Those are different series. If you need them, build them.

The method before the moral

Write the definition. Whose spend? Which units count as 'State' — departments, ULBs, societies, SPSUs? Which instruments — GeM, state e-procurement, off-platform? Which FY alignment — April-March? Goods only or goods plus services? Inclusive of rate contracts? Inclusive of cancelled orders? If you cannot write the definition, you cannot use the percentage.

Pull at least three primary views. One: GeM statistics and the latest GeM or Commerce / PIB note, with the date you opened them. Two: Department of Expenditure circulars and any union compilation that mentions state adoption of GeM. Three: the state finance or e-procurement dashboard for the states you actually sell to. If those three disagree, the disagreement is the finding. Do not average them into a friendlier number.

Separate migration from expansion. A state that moves existing stationery and medicine onto GeM will show GeM growth. The citizen may not have received 38 percent more anything. For a founder, migration can still be a real go-to-market fact — more of the spend you can see is now on a portal you can bid on. For a CIO, it is not a mandate to inflate next year's AI head by 38 percent.

1. Screenshot or export the GeM statistics page on a dated folder.
2. Open the PIB / Commerce note you intend to cite. File the PRID and the access date.
3. List the ten buyers you actually care about. Pull their state-portal award data if it is public.
4. Build two columns: GeM-reported state GMV, and on-budget IT or goods spend from the state budget documents.
5. Write one paragraph on what you still cannot see: off-platform, societies, district spends below threshold.
6. Only then write a sentence that uses a percentage, with the definition glued to it.

Sentences that are legal in a deck, and sentences that are not.

Sentence	Status on this file	Repair
State procurement grew 38%	Not acceptable as a general fact	Name the platform, the FY, the source, the definition
PIB's GeM note of 6 April 2026 reported 38.3% growth in state procurement on GeM in FY 2025-26; we verified the live page on [date]	Acceptable if the page still says that	Keep the access date
Therefore TAM for state AI is 38% higher	False inference	Build a category series or stay quiet
Therefore states must buy agents this year	False inference	MCC, SDC readiness and human gates decide that
Our category on GeM grew X% for these buyer types	Acceptable if you computed X	Show the query

What a real 38.3 percent on GeM might mean — if the page still says it

If you verify the PIB sentence, several interpretations remain live. More state departments may have been onboarded. Existing departments may have moved more categories onto GeM. Average ticket sizes may have risen. A few large infrastructure or medicine orders may have dominated. Without a breakdown by category and by state, you cannot tell which story you are in. Ask GeM or the state buyer for the cut you need. Do not infer it from the headline.

For a GovTech founder the useful meaning, if verified, is narrower. The portal you already knew how to bid on carried more state activity last year. That can justify a GeM-first motion, better category pages, and patience with state onboarding. It does not justify inventing a national land API or promising auto-FIR. Growth in marketplace GMV is not a change in the legal line.

For a state CIO the useful meaning is even narrower. Your colleagues may be under pressure to show more GeM numbers. That pressure can help you insist that agentic software, if bought, is bought on a reconstructable file rather than as a side WhatsApp pilot. It can also tempt you to rush a bid before MCC. Do not let a percentage outrun a freeze calendar.

Two decks after the same headline

Objections you will hear — and what to do with them

Close enough for a TOFU blog

Not on this site. TOFU is where false percentages breed. If we cannot cite, we hedge. If we can cite, we scope.

Our investors expect a big TAM

Give them two series and a hole. Investors who need a fake 38 will also need a fake churn story later. You are allowed to prefer the ones who can read a definition.

The state wants a success number for the assembly

Give them GeM's own sentence, attributed, or give them your department's own spend. Do not launder a platform migration into a claim that the state grew 38 percent more effective.

FY26 is over, so the number is frozen

Press notes get corrected. Definitions get restated. Freeze your access date, not your curiosity.

A one-week method playbook

1. Day 1: open PIB PRID 2249335 and gem.gov.in/statistics. Save both. Write the exact sentence you are allowed to use.
2. Day 2: pull DoE and your target state's e-procurement or finance dashboard. Note gaps.
3. Day 3: extract your GeM category, if any, for state buyers. This is the only percentage you own.
4. Day 4: write the migration-versus-expansion paragraph. If you cannot, you may not use the headline in a CIO meeting.
5. Day 5: strip any deck sentence that says 'state procurement grew 38%' without platform, FY and source. Replace or delete.

File note you can paste

Subject: Use of '38 percent state procurement growth' in notes and decks.

This department / firm will not state that 'state procurement grew 38 percent' as a general fact. A title-bank phrase is not a source. If we cite FY 2025-26 growth in state procurement on GeM, we will cite the live PIB / GeM text (PRID 2249335 or successor), the access date, and the definition that it is GeM-reported platform activity. We will not infer TAM, AI spend, or a mandate to buy agents from that sentence. Category claims will use our own extract. This note is not investment advice.

Prcept AI would rather lose a TOFU click than teach a false percentage to a secretariat. If your growth story needs this headline unscoped, the story is the product. We are not in that business.

What this does not mean for an AI head

Even a verified 38.3 percent rise in state activity on GeM is not an argument that your AI line should rise 38.3 percent. GeM GMV is dominated, in most years, by goods and by categories that are not agentic software. A medicine order and a model runtime are not cousins. If your finance secretary is being shown the headline as a spending mandate, bring the definition to the meeting and offer the clerk-plus-parent list from the CIO priority note instead.

It is also not an argument that procurement risk fell. More transactions on a portal can coexist with the same invented APIs, the same zero CR envelopes, and the same MCC surprises. Platform growth is not control growth. Score control on the packet, not on GMV.

If you cannot open PRID 2249335 or gem.gov.in/statistics on the day you file, delete the percentage from the file. A missing primary is not a reason to remember a number fondly. Title banks are allowed to be wrong. Cabinet notes are not.

This article is informational field guidance, not legal, investment, statistical or procurement advice. Figures mentioned were checked against secondary reports of PIB PRID 2249335 as of 17 August 2026 and must be re-verified on the live GeM, DoE and PIB pages before use. If those pages differ, those pages win.