Cloud Marketplace Sales Carry About 4% of Software Spend, Not the Reported 20%

Sellers surveyed by Tackle say 20% of their revenue came through cloud marketplaces. The market-level figures do not support that as a description of B2B software. Hyperscaler marketplace sales were about $45 billion in 2025, which is 3.6% of Gartner's software total. We estimate about 4.1% for 2026. Both can be true, and the gap is the useful part.

The short answer

Hyperscaler marketplaces carried about $45 billion in 2025 on Omdia's curve, which is 3.6% of Gartner's $1.25 trillion software total. For 2026 we estimate about $59 billion, or 4.1%. If those sales are multi-year contract values, the annual share is closer to 2%. No cloud publishes an audited total.

AWS Marketplace publishes no sales total, so every share figure is an estimate

AWS does not publish a total for AWS Marketplace sales in the sources we could open. Its December 2025 Marketplace announcement gives product and speed claims, with no customer, seller or dollar total. That absence shapes everything downstream, because any share figure you read has to be built by someone outside the company.

The same holds across the 3 hyperscalers. What each discloses is a growth rate, an adoption claim or a list of named sellers. None offers a dollar figure that you could divide by a software market.

What the 3 clouds actually disclose

Microsoft comes closest to a usage statistic. At Ignite 2025 it claimed 2X+ sales growth, a 75% increase in average spend and about 2X growth in AI solutions purchased. A separate session summary said 85% of Microsoft customers with commitments are buying through Marketplace. Neither claim comes with a base, so a doubling cannot be turned into dollars.

Google Cloud's backlog and revenue are public, and they are covered in the comparison of Microsoft, Google Cloud and AWS backlog. We found no Google Cloud Marketplace sales figure in the earnings coverage or partner posts we checked. Treat that as a failure to find one, not proof that none exists.

Cloud marketplace disclosures in one table: 3 annual figures out of 11

The best public evidence on AWS Marketplace sales is a set of seller milestones, and they measure different things. The table below logs 11 named sellers, using a December 2025 IT Channel Oxygen survey as the backbone and adding primary documents where we could open them.

Sources: IT Channel Oxygen, 1 December 2025, and its 4 March 2025 Okta report; IT Brief for Snowflake; the Dynatrace Q4 FY2026 release; AWS for the Presidio quote. Most rows are secondary reports of company claims, not filings.

Why $1 billion means 3 different things

The honest count is 3 annual figures, 3 lifetime or aggregate figures and 5 with no stated period. Omdia's Alastair Edwards described the shift to IT Channel Oxygen: "It used to be all about cumulative sales (since launch) on AWS Marketplace. Now the place to be is the 'annual' $1bn club, which is a much smaller group of vendors, but growing all the time."

So when you see a vendor in the $1 billion club, ask which club. Dynatrace's release says only that it "Surpassed $1 billion in AWS Marketplace sales", with no period attached. That wording is accurate and it cannot be compared with Snowflake's figure.

Why marketplace sales cannot be divided by revenue

The tempting calculation gives striking numbers. Against company revenue, the annual figures work out at about 43% for Snowflake, using fiscal 2026 revenue of $4.7 billion, about 25% for CrowdStrike, using fiscal 2025 revenue of $3.95 billion, and about 24% for Splunk on the $4.2 billion in the Oxygen survey.

I would not quote those as channel share. Marketplace sales look like bookings and revenue is recognised over time. AWS itself charges its listing fee on pre-tax total contract value, and Okta's milestone is explicitly contract value. A 3-year deal counts in full on the left of that ratio and in thirds on the right.

The filings do not rescue the calculation. Asked on a call whether AWS Marketplace was over 10% of annual contract value, Okta CFO Brett Tighe called it a "greater, growing percentage" and gave no number. The parts of the Snowflake and Dynatrace fiscal 2026 annual reports we reviewed give no marketplace percentage either, though our review did not cover every page. This is the same gap between a headline and an auditable measure examined in the ranking of SaaS AI disclosure metrics by rigour.

Marketplace market estimates: $30 billion in 2024, about $45 billion in 2025

The market-level numbers come from one research line, which is the first thing to know about them. Canalys is now part of Omdia, so the figures quoted across the trade press are mostly one analyst team's model rather than independent confirmation.

Omdia's October 2025 release puts enterprise software sales through hyperscaler marketplaces at $30 billion in 2024, rising to $163 billion by 2030 at a 29.1% compound rate from 2025. The release text we read gives no 2025 total, so we derived one.

The working behind the $45 billion

Dividing $163 billion by 1.291 to the fifth power gives $45.5 billion for 2025. That agrees with the $45 billion Canalys expected for 2025, as reported in March 2025. A third check sits in the same Omdia release: 3 categories worth $28.7 billion are said to be 63% of spending, which implies $45.6 billion if that split refers to 2025. The release does not say which year it means.

Rolling forward one year at the same rate gives $58.7 billion for 2026. That is our interpolation of a forecast, not a measured result and not a figure Omdia published.

The forecast record is mixed

Marketplace forecasts deserve some suspicion, and the record cuts both ways. In November 2023 Tackle projected revenue through the marketplaces would go from $15 billion in 2023 to more than $100 billion in 2026. The Omdia path implies $58.7 billion for 2026, about 41% short of that call. Analysts erred the other way: the older Canalys forecast of $85 billion for 2028 now sits about 13% below the $97.8 billion the Omdia path implies.

The cloud marketplace share of software spend: about 4%, with the working shown

Our central estimate is that marketplaces carried 3.6% of worldwide software spending in 2025 and will carry about 4.1% in 2026. The share is rising because marketplace sales are compounding at about twice the rate of software spending, at 29.1% against Gartner's 15.1%.

Software from the Gartner April 2026 forecast, read via a reposted release. SaaS from the Gartner November 2024 forecast, also reposted. The 2-year term is our illustrative assumption, not data.

Why the two denominators bracket the answer

Neither denominator is right on its own, which is why we use both. Gartner's software total includes spending that rarely passes through a marketplace, so the 3.6% understates the share of software that could realistically go that way. SaaS alone is too narrow, because marketplaces also sell other deployment types and infrastructure software, so 15.2% overstates it.

The contract-value row is a stress test. No source tells you the average term inside the Omdia figure, or even whether it counts contract value at all. If the figure works like AWS fee accounting and the average deal runs 2 years, the annual flow halves. My bounded view for 2025 is 2% to 4% of all software, and 8% to 15% of SaaS.

Why cloud marketplace sellers report 20% and the market shows 4%

Both figures can be accurate because they describe different populations. Tackle's 2025 survey summary reports that respondents averaged 20% of revenue through marketplaces and expected 32% within 12 months. Tackle sells marketplace software, and its respondents are unusually likely to be marketplace sellers already.

The arithmetic rules out the broad reading

Apply 20% to Gartner's 2025 software total and you get about $251 billion of marketplace sales, or 5.5 times the implied $45.5 billion. So the 20% cannot describe the software market. The largest software companies also sit inside Gartner's denominator, and their channel mix is mostly undisclosed.

The Clazar and Partner Insight report from April 2025, based on inputs from 100+ SaaS companies, sits lower: 22% of companies report more than 20% of revenue through marketplaces. Clazar is also a vendor in this market. The summaries we could open do not make the two samples comparable, and a 20% mean sits awkwardly beside a report where most companies fall below 20%.

I would stop putting the 20% on a board slide as a market benchmark. It is a useful signal of what committed participants experience, and that is a narrower claim. The expected 32% deserves more caution still, because it is a seller's hope about its own channel. Budget pressure on the buyer side, covered in the data on CFO vetoes of software purchases, does not move on a seller's expectations.

Cloud commit burn-down: what each hyperscaler lets a buyer retire

Commit burn-down is why this channel exists at its current size. A buyer signs a multi-year spending commitment with a cloud provider, then buys third-party software through that cloud's marketplace so the purchase counts against the commitment. In Tackle's survey, 75% of respondents named access to committed cloud spend as a top advantage.

The pool is large. Omdia's estimate of close to $470 billion in commitments compares with $419 billion from Partner Insight six months earlier. On Omdia's numbers, 2025 marketplace sales equal about 9.7% of that stock. The comparison is rough, because the stock spans several years and the sales are one year.

Sources: AWS listing fees, CIO Dive, February 2025, Microsoft fees, Microsoft MACC enrolment, Microsoft session summary, Google, May 2025, Google drawdown rules. Retrieved 5 November 2026.

The AWS cap is widely quoted and thinly sourced

Many advisers quote a 25% limit on how much AWS Marketplace spend can retire a commitment. We could not trace that figure to an AWS document. Trade reporting from July 2026 describes such caps as often falling between one-fifth and one-quarter of the commitment, which is directional at best. The number that binds you is the one in your own contract.

The hosting rule has a real strategic consequence. A multi-cloud SaaS vendor may be listed on all 3 marketplaces and still fail to deliver full commit value on any one of them. Buyers should ask where the workload actually runs before assuming the purchase retires anything.

Where this marketplace share estimate is weakest

This estimate stacks forecasts from 2 firms, and that is its main flaw. Omdia's $30 billion for 2024 is the only anchor presented as a base year, and the release gives no counting method beyond the phrase "enterprise software sales". If Omdia counts data products or services, the numerator is inflated relative to Gartner's software category.

Four things that would move the number

Mismatched vintages. Omdia's model dates from October 2025, Gartner's software total from April 2026 and Gartner's SaaS figure from November 2024. Gartner has revised software spending several times since, and the SaaS outlook is the stalest input, so the 15.2% ceiling may be somewhat too high.

Possible undercount. Omdia's figure covers hyperscaler marketplaces. Some software a buyer draws against a commitment may reach them through paths that never appear in that total. Snowflake's chief executive cited over $3.9 billion booked through its AWS collaboration over 4 quarters, far above its $2 billion marketplace figure, so co-sell can be much larger than transacted marketplace sales.

The wrong denominator for you. A 4% global average says little about a security or infrastructure vendor selling to US enterprises. Named security sellers dominate the ledger above, and some of the annual figures run to about a quarter of company revenue on a bookings basis. Your segment may be well above the average, and small vendors may be well below it.

The research base. Most company figures here are press claims repeated in trade coverage rather than audited filings. The two seller surveys come from vendors with a commercial interest. If AWS published one audited annual total, most of this section would be obsolete, and I would welcome that.

What the 4% figure means for your co-sell budget

The 4% does not argue against investing in a marketplace. It argues against treating 20% as the default outcome. Reported growth is fast: CrowdStrike reported 91% in 2024, Snowflake doubled, and Okta's marketplace revenue rose over 80% in fiscal 2025. But those are the sellers who chose to announce, and nobody issues a press release for a flat channel.

The cost side is easy to model. On a $500,000 AWS private offer, the fee is $15,000 at 3%, or $17,500 if it runs through a partner-led private offer at 3.5%. A $2 million deal pays $40,000 at 2%. A $500,000 renewal pays $7,500 at 1.5%. These are small numbers beside the sales and marketing share of revenue seen in the 12 IPO filings we compared, so the fee is rarely the deciding cost.

The real cost is operational. Only 58% of Clazar's respondents have trained account executives to route deals through the marketplace, and fewer than half have automated usage tracking, reporting and CRM integration. On co-sell, Tackle's respondents said it influenced about 22% of net-new deals, with 83% co-selling with AWS, 51% with Microsoft and 35% with Google Cloud.

My recommendation is to sequence by where your buyers hold commitments. Start with one cloud, and prove that the deal would have been slower or smaller without the commit. A marketplace that only rebills deals you would have closed anyway is a 3% tax. It is also worth keeping in mind for buyers, since a purchase routed to burn down a commitment can still add to the app sprawl your finance team is trying to cut.

Frequently asked questions

Where to start with cloud marketplace numbers this week

Pull your last 20 closed deals and tag each one with the buyer's cloud provider and whether a commitment existed. If your account teams cannot answer that, you have found the first gap, and it is cheaper to fix than any listing. The result tells you which marketplace to prioritise, and whether you have a commit story at all.

Then rewrite any marketplace figure in your planning deck with its period and basis beside it: annual or lifetime, sales or contract value, seller survey or market estimate. Drop the ones you cannot label. If you sit on the buying side, ask your cloud account manager for the drawdown rule in writing, along with the list of your vendors that qualify.

The same discipline applies to distribution claims in the next wave of agent stores, which is the question behind what distribution through agent marketplaces is actually worth.

Related reading