AudioCodes: How Recurring Is the New Services Engine, Really?
AudioCodes already has a real layer of subscriptions, SaaS and managed services, but at the end of 2025 its RPO was almost identical to deferred revenue and only $19.7 million extended beyond 2026. The new engine is becoming more recurring, but it still has not built the kind of deep backlog that would break the company’s dependence on quarterly order flow.
What This Follow-up Is Isolating
The main article argued that AudioCodes’ newer engine still does not deliver clean operating leverage. This follow-up isolates a different question: when AudioCodes talks about services, subscriptions, Live Platform and AudioCodes Live, how much of that layer really behaves like recurring revenue with good visibility, and how much still depends on implementation work, labor hours and orders that need to be renewed quarter after quarter.
The answer is fairly sharp. There is a real recurring engine here, but it is not as deep as a headline like “53.2% services” might imply. The company clearly does have subscriptions, managed SaaS and managed services, but at year-end 2025 deferred revenue and RPO, meaning contracted revenue not yet recognized, were almost identical. And only $19.7 million of that RPO sits beyond 2026. That is a real contractual base. It is not yet the deep backlog of a mature SaaS company.
The more interesting point is the gap between the word “services” and the quality of those services. The company explicitly says that SaaS, maintenance and support revenue is recognized ratably over the contract term, but professional services revenue is generally recognized over time based on labor hours consumed, or ratably only when the transfer pattern is consistent. So a shift toward services does not automatically equal a clean subscription engine.
| Layer | What the company says | Why it matters |
|---|---|---|
| SaaS, maintenance and support | Revenue is recognized ratably over the contract term | This is a genuinely recurring layer |
| Professional services | Revenue is generally recognized over time based on labor hours consumed | This is service revenue, but not always deep subscription revenue |
| Subscription agreements | Typically run 12 to 36 months on a per-user, per-month basis | There is a contractual base, but the duration is still short to medium term |
| Current revenue base | Revenue is generated primarily from on-premises deployments | The cloud transition is real, but it still has not replaced the old revenue base |
What Is Actually Recurring Inside “Services”
This is where the distinction matters most. AudioCodes is no longer relying only on maintenance and support. It is expanding professional services and managed services around AudioCodes Live, Live Platform, Microsoft Teams, Cisco Webex, Zoom and broader UC and contact center environments. That also explains why service revenue held up in 2025 even as the company says product support revenue declined by about 5%.
But in terms of recurrence quality, this is a more mixed picture than it first appears. Maintenance, support and SaaS do sit on a contractual model that is recognized over time. Professional services, by contrast, are often recognized based on labor hours consumed. In other words, part of the services layer is classic recurring revenue, and part of it still depends on delivery, implementation and managed work being performed.
That is exactly why 53.2% services does not automatically make AudioCodes a SaaS company. The company itself says 2025 services growth was driven mainly by professional and managed services while product support declined as customers moved away from legacy telephony and from maintaining on-premises equipment. That strengthens customer attachment and creates follow-on selling opportunities, but it does not necessarily create the same kind of visibility, margin profile and operating leverage as a cleaner software model.
The positive side is still real. AudioCodes Live is consumed on a monthly subscription basis, Live Platform is presented as a multi-tenant managed SaaS offering, and the company explicitly says per-user, per-month subscription agreements are becoming an increasing proportion of the business. So it would be wrong to say there is no recurring transition here. The more accurate claim is narrower: the new engine is becoming more recurring, but it still mixes subscription revenue, managed services and implementation work.
The Contract Layer: Real Base, Limited Depth
The year-end 2025 numbers are striking. Deferred revenue stood at $57.9 million. Total RPO also stood at $57.9 million. That near identity matters.
Analytically, if RPO includes both deferred revenue and non-cancelable contracts not yet recognized, yet ends up almost identical to deferred revenue, that suggests reported visibility is not materially larger than what has already been invoiced and parked in deferred revenue. This looks more like a pool that gets refilled than a reservoir that is deepening.
The yearly split sharpens the point:
| Period | RPO | Read-through |
|---|---|---|
| 2026 | $38.2 million | Most of the visibility sits inside the next 12 months |
| 2027 | $9.8 million | Only a limited layer extends into the following year |
| 2028 and later | $9.9 million | Multi-year visibility is still modest |
| Beyond 2026 combined | $19.7 million | Only about 8.0% of 2025 total revenue and about 15.1% of 2025 service revenue |
Out of the $57.9 million of RPO, only $152 thousand belongs to products. Almost all forward visibility sits in services. That is positive in one sense because it confirms that the recurring layer really is inside the services engine rather than inside delayed hardware shipments. But it also means there is no large product backlog that can cushion a weak quarter. If the services layer does not deepen, there is not much product visibility to offset it.
The deferred revenue roll-forward sends a similar message. The company recognized $42.7 million from the opening balance and added $43.1 million of new deferred revenue during the year. The result was an increase of less than half a million dollars in the ending balance. That is not deterioration, but it is not acceleration either. The deferred revenue pool is staying full at roughly the same depth.
Why This Still Is Not Deep Backlog
This is where the follow-up connects directly to the company’s own warning. AudioCodes says it has a limited order backlog and that any quarter depends substantially on orders received and delivered in that quarter. That is not a minor footnote. It is an explicit admission that the transition toward services still has not created a visibility layer strong enough to smooth out a soft quarter.
The commercial terms point in the same direction. Typical product agreements are renewable 12-month agreements or can be terminated at will on 90 days’ notice, with no inventory commitment and no minimum sales obligation. Subscription agreements, by contrast, typically run for 12 to 36 months, but that is still a relatively short-to-medium duration layer. So there are real contracts here, but a meaningful part of the model still depends on renewal, expansion, delivery and fresh order intake rather than on a deep multi-year commitment base.
There is also an operating layer to this. The company says new-customer sales cycles typically run six to twelve months after a design win, and sometimes longer. Adoption of cloud-based and managed-service delivery has made deployment more agile in some segments, but it has not removed dependence on the commercial machine. If win rates slow, the existing RPO is not deep enough to absorb the hit cleanly.
And then there is the strategic gap the main article already highlighted. AudioCodes says revenue today is still generated primarily from on-premises deployments, and that SaaS pay-per-use models can even hurt short-term revenue recognition. So the long-term economics of the transition may well be better than the current headline numbers suggest, but as of year-end 2025 the market still is not looking at a mature cloud engine with deep visibility. It is looking at a business in transition that has not finished the shift.
Conclusion
This follow-up does not mean AudioCodes’ recurring-revenue narrative is weak or artificial. There is a real subscription layer here, there is managed SaaS, there is almost $58 million of deferred revenue, and services still account for more than half of the top line. That is not cosmetic.
But it is also not yet the kind of visibility layer that deserves to be called deep backlog. Almost all of RPO sits in services, almost all of it overlaps with deferred revenue, and only $19.7 million extends beyond 2026. At the same time, the company still says revenue is generated primarily from on-premises deployments and that its order backlog is limited. That means the new engine is more recurring than the old one, but it still has not freed the company from dependence on quarterly commercial execution.
Current thesis in one line: AudioCodes has already built a real recurring-revenue base, but as of the end of 2025 it is still short-to-medium in duration rather than the deeper visibility stack of a mature SaaS company.
The strongest counter-thesis is that this reading may be too conservative. Monthly subscriptions and pay-per-use models can reduce reported backlog and deferred revenue even when customer value and lifetime economics are actually improving. That is a fair objection. Still, until the layer deepens in the numbers, the market is unlikely to give it full credit.
What changes the read from here is not more AI or UCaaS marketing. The market will need to see RPO and deferred revenue grow faster than service revenue, the portion beyond 12 months become more meaningful, and managed services and subscription offerings start building deeper visibility rather than merely replacing legacy support or implementation work.
Disclosure: Deep TASE analyses are general informational, research, and commentary content only. They do not constitute investment advice, investment marketing, a recommendation, or an offer to buy, sell, or hold any security, and are not tailored to any reader's personal circumstances.
The author, site owner, or related parties may hold, buy, sell, or otherwise trade securities or financial instruments related to the companies discussed, before or after publication, without prior notice and without any obligation to update the analysis. Publication of an analysis should not be read as a statement that any position does or does not exist.
The analysis may contain errors, omissions, or information that changes after publication. Readers should review official filings and primary sources before making decisions.