Progress Software Earnings: ARR Up 1%, Revenue Down 2%, and 8 Peers Compared

Progress Software reported fiscal third quarter 2026 revenue of $246 million, down 2%, and annualised recurring revenue of $873 million, up 1% in constant currency. The company calls itself "an AI infrastructure software leader" in the same release. Progress Software earnings are the cleanest recent example of a pattern across mid-cap software: AI appears in every headline and almost none of the growth.

The short answer

AI is not lifting mid-cap software evenly. Among nine vendors with recent filings, the five whose revenue grew 3% or less all carry a large maintenance or on-premises base that is shrinking. Their cloud and AI lines are growing, but from a base too small to offset the decline. The four growing near 13% sold subscription first.

What Progress Software's earnings actually show

Progress reported its fiscal third quarter, which ended on 31 August 2026, on 30 September. Revenue of $246 million decreased 2% year over year on both an actual and a constant currency basis. ARR, the company's measure of annualised term-based contracts, rose 1% in constant currency to $873 million.

Profit went the other way. Non-GAAP operating margin reached 43%, up from 40% a year earlier, and cash from operations rose 20% to $88.0 million. This is a business that is getting more profitable as it stops growing.

The AI language sits on top of those numbers. The release describes Progress as "an AI infrastructure software leader". CEO Yogesh Gupta tied the quarter to helping organisations "realize the full potential of AI". The release headline also announced that the acquisition of Domo's AI and data platform business had closed.

My reading is blunt. Nothing in this quarter's revenue is AI revenue that the company is willing to size. The 10-Q says the results of Nuclia, the agentic retrieval business Progress bought in 2025, are not disclosed because "those amounts are not significant".

Where the $3.8 million of Progress revenue went

The 10-Q splits revenue into four lines. Laying them side by side shows that the decline is almost entirely one line, and it is not the line that AI is supposed to affect.

Maintenance did the damage

Maintenance revenue was $100.0 million, down from $104.8 million. That is a 4.6% drop on a line that still makes up about 41% of total revenue. Maintenance is the annual support fee customers pay on perpetual licences, so it shrinks when customers retire old deployments or move them elsewhere.

Progress gives a different explanation. Management attributes the revenue decline mainly to the timing of renewals on multiyear subscription contracts. That may well be right for the licence line, which recognises term renewals up front. It does not obviously explain a maintenance line falling for a second consecutive comparison.

SaaS barely moved

SaaS revenue was $72.6 million against $71.5 million, growth of about 1.6%. This is where Progress's cloud products, including its file sharing and retrieval businesses, would show up if AI demand were pulling them. A 1.6% growth rate on the SaaS line is the number that should worry a holder more than the headline 2%.

Domo is not in these numbers yet

The Domo AI and data platform deal closed on 22 September 2026, after the quarter ended. The 10-Q puts the price at about $400 million, funded from the company's revolving credit facility. Progress raised its fiscal 2026 revenue guidance to between $1.044 billion and $1.052 billion, from between $990 million and $1.002 billion. That is a $52 million increase at the midpoint, and the release does not split it between Domo, currency and the base business.

I would treat that guidance raise as bought, not earned, until a later filing proves otherwise. Progress has run an acquire and integrate model for years, and it works on cash flow. It is not evidence of AI demand reaching the existing product set.

Nine mid-cap software filings, one growth table

One company is an anecdote. So I took the latest earnings release from eight other mid-cap infrastructure and application software vendors and put their growth rates beside Progress. Every figure below comes from the company's own release, either on SEC EDGAR or a newswire copy of the same text.

Recurring metrics are each company's own and are not comparable across rows. Teradata's third quarter had not been reported when this was verified on 29 October 2026. Box reports no ARR, so RPO is shown. Dropbox growth includes FormSwift, which the company is winding down.

The simple averages make the gap concrete. The five slowest vendors averaged 0.96% revenue growth. The four fastest averaged 12.75%. Those are unweighted averages of a hand-picked set, so treat them as a description of this table, not of the sector.

For context, a separate tracker of nine larger, higher-growth SaaS names found median growth near 22%, which is covered in the revenue growth tracker across four quarters of SaaS filings. The mid-cap infrastructure tier sits well below that, and the lower half of it is barely growing at all.

The legacy software line is shrinking everywhere it is disclosed

The obvious story is that AI demand skipped the slow five. The filings support a less exciting story. Each of the slow vendors that splits its revenue shows a growing cloud line and a shrinking legacy line, and the legacy line is still the bigger one.

Teradata: cloud up 8%, everything else down

Teradata reported public cloud ARR of $686 million, up 8%, inside total ARR of $1.509 billion, up 1%. Subtract one from the other and non-cloud ARR fell from $855 million to $823 million, a decline of 3.7%. That calculation is mine, not the company's.

The outlook is weaker than the quarter. Teradata guided third quarter total revenue to between minus 6% and minus 4% year over year, and recurring revenue to between minus 4% and minus 2%. The CEO quote in the same release highlights the launch of the Autonomous Knowledge Platform "to help enterprises deploy agentic AI".

OpenText: 22 quarters of cloud growth, flat ARR

OpenText reported cloud revenue of $503 million, up 6%, its 22nd consecutive quarter of organic cloud growth. Its ARR is defined as cloud revenue plus customer support revenue, and that total rose only 0.2% to $1.057 billion, or fell 1.8% in constant currency. Working backwards from the rounded figures implies support revenue fell by roughly 4.5%.

The CEO quote says "OpenText is the secure data foundation in the AI stack". The company also announced a €105 million investment in Ireland to expand agentic AI and sovereign cloud. Neither statement is attached to a revenue number.

Pegasystems: the fastest cloud line, still not enough

Pegasystems is the sharpest case. Pega Cloud ACV grew 20% to $976 million and now makes up 59% of total ACV. Maintenance ACV fell 12% and subscription licence ACV fell 6%. Total ACV grew 6%, and revenue grew 3%.

COO and CFO Ken Stillwell said ACV growth through the first three quarters "did not reflect the opportunity we see in the market". Founder and CEO Alan Trefler said "The AI market is still sorting itself out". That is the most candid AI sentence in any of the nine releases, and I give Pega credit for it.

What each mid-cap software vendor says about AI, and what it quantifies

The second half of the brief was to compare what these filings claim about AI. The comparison is easy to summarise. Almost every vendor names AI. None of them sizes it in the release.

Five9 is the only vendor here that gave investors a number. On the August earnings call, as reported by CX Foundation, the company said AI revenue grew 78% year over year and made up 15% of subscription revenue, up from 9% a year earlier. That figure comes from secondary coverage of the call, not the release, and should be upgraded to the transcript when it is available.

The pattern matters because AI revenue disclosure is a choice. Vendors with a growing AI line tend to break it out, because it supports the multiple. The comparison of how vendors define those numbers is covered in the ranking of AI revenue disclosure metrics. Vendors that keep AI in the adjectives and out of the tables are telling you something by omission.

There is also a regulatory edge to this. The SEC has brought cases over AI claims that a company could not support, which is logged in the AI washing enforcement record. None of the language above approaches that line, as far as I can tell. It is positioning language, and positioning is legal. It is just not evidence.

Why four mid-cap software stocks are still growing

The fast four do not share an AI story either. Appian has no AI statement in its release at all. What they share is a business mix that was already subscription or cloud before AI arrived, so there is no large legacy base dragging the total down.

Appian grew cloud subscriptions 23% to $131.7 million and reported cloud net ARR expansion of 115%. Commvault grew subscription ARR 20% to $1.104 billion and has stopped disclosing total ARR, which included maintenance on perpetual licences. That second detail is telling. Commvault removed the legacy line from its headline metric, which makes its growth look cleaner and its comparability worse.

Box sells the governed layer under the agents

Box grew revenue 9%, or 11% in constant currency, with RPO up 15% to $1.7 billion and net retention at 106%. CFO Dylan Smith credited the Enterprise Advanced suite, which bundles AI features with governance and workflow. Box cannot separate the AI part of that suite from the rest, and it does not try.

Aaron Levie's framing is that agents need "secure, well-governed unstructured data". If that is right, Box is being pulled by AI as a data custodian, not as a model vendor. It is a plausible mechanism, and it is still a hypothesis until Box sizes it.

Five9 is the only one where AI is the growth

Five9 grew revenue 10% and subscription revenue 14%, its third quarter of accelerating subscription growth. If AI is 15% of subscription revenue and growing 78%, AI is doing a large share of the acceleration. Contact centre software is also where AI agents replace human labour most directly, which is why pricing there is shifting fastest.

So the honest answer to "which ones is AI passing over" is narrower than the headline. AI is visibly lifting one of nine. It may be lifting Box and Appian indirectly. For the other six, the filings show no AI effect that anyone has chosen to measure.

Where this argument is weakest

The table mixes quarters that end between 30 June and 30 September 2026. Three months is a long time in this market, and the June-quarter reporters may already have moved. Teradata in particular reports its third quarter in early November and guided down.

The sample is mine. I picked nine vendors I judged to be mid-cap infrastructure and application software, and I excluded BlackLine because I could not open its primary release during research. A different nine could produce a different split. The averages are unweighted, so Appian's 19% counts as much as OpenText's 2.9% despite a revenue base many times smaller.

The recurring metrics do not compare. Progress ARR counts term-based contracts, OpenText ARR counts cloud plus support, Pegasystems uses ACV and Box reports none. I have used each company's own definition and computed only within a company, never across companies.

The strongest counter-case is that low growth is the strategy. Progress makes a 43% non-GAAP operating margin by buying mature software and running it for cash. Judging it on growth is judging a different business model by the wrong standard. If Domo and Nuclia take two years to show up in SaaS revenue, the current quarter says nothing about them yet. I find that argument fair on cash flow. I do not find it persuasive on the AI claim, because the company chose to lead its release with that claim.

Finally, I cannot separate AI from cloud migration. A customer moving to Pega Cloud may be doing so to get AI features. If so, the blue bars in the chart above are partly AI demand, and my framing undercounts it.

What the software growth slowdown means for your contracts

You are probably a customer of at least one company in the slow five. Their numbers tell you how they will treat you at renewal.

A vendor whose maintenance line is shrinking has two levers. It can raise maintenance prices on the customers who stay, or it can push you to its cloud product, where the revenue is recurring and easier to grow. Expect both. The price side is covered in the analysis of grandfathered pricing and renewal increases, and the pattern there applies directly.

When the cloud migration pitch arrives with AI features attached, ask for the AI part priced separately. If the vendor cannot tell its own investors what AI earns, it cannot credibly tell you what AI is worth inside your bundle. That is a negotiating point, not a criticism.

Retention data is worth asking for too. Progress says its net retention has generally ranged between 99% and 100%, which means the average customer spends about the same each year. Your account may be one that is being nudged upward to keep that average flat. The spread across vendors is in the net revenue retention comparison across 12 Q2 filings.

If you hold these names rather than buy from them, the read is different. The market sold software broadly on AI fears, as covered in the explanation of the SaaSpocalypse selloff. These filings suggest the damage at the slow end is older and slower than AI. Legacy runoff was happening before agents, and it would be happening without them.

Frequently asked questions

Where to start this week

Pull your own contracts with any of the slow five and find the maintenance line. Note what it cost three years ago and what it costs now. If the increase beats inflation, you are funding the vendor's flat ARR, and you have the number to open a renewal conversation with.

Then put one question to your account manager in writing: what share of your company's revenue comes from the AI features you are asking me to pay for? If the answer is a percentage, you have learned something. If it is a paragraph, you have learned more.

Last verified 29 October 2026 against each company's latest earnings release. Re-verify after Teradata, Progress and the June-quarter reporters publish their next quarter.

Related reading