NFON’s expansion into artificial intelligence is beginning to generate recurring revenue, but the newer products are not yet large enough to compensate for pressure in the cloud telephony business on which the company was built. NFON reported lower sales and earnings for the first half of 2026 and has reduced its full-year expectations after weaker demand in Business Telephony, including particularly soft performance in the UK.

Consolidated revenue fell 3.8% to €42.5 million from €44.2 million a year earlier, while adjusted EBITDA declined 22.9% to €4.4 million. The number of seats across the group dropped 4.2% to 629,869, although blended average revenue per user remained broadly stable at €9.98.

Recurring revenue from Intelligent Assistant and Customer Engagement moved in the opposite direction and recorded what NFON described as strong growth. Those products form the centre of its attempt to expand beyond cloud telephony into AI-assisted conversations, contact-centre applications, and automated workflows, but their smaller starting base means they cannot yet offset contraction in the older division.

The imbalance prompted the company to reduce its 2026 guidance. NFON now expects revenue between €84.5 million and €86 million and adjusted EBITDA between €9.5 million and €10.5 million, replacing earlier expectations for low to mid single-digit revenue growth and EBITDA above €12 million.

AI growth has to overcome the legacy base

The figures provide a more useful measure of enterprise AI adoption than a product announcement because they expose what happens when new technology enters an existing recurring-revenue business. NFON already has customers, partners, billing systems, and communications products through which AI services can be sold, avoiding the distribution problem facing a startup entering the market from scratch.

However, the same installed base creates a difficult arithmetic problem. Business Telephony remains materially larger than NFON’s newer services, so rapid percentage growth in Intelligent Assistant can still be overwhelmed by a modest decline across thousands of conventional communications seats.

Total recurring revenue fell 3% to €40 million even as its share of group revenue increased to 94.1%, illustrating the tension clearly. The stability of blended ARPU suggests the immediate pressure is not simply aggressive price discounting but the number and mix of services being sold.

NFON has been simplifying products and systems accumulated over time while integrating technology from botario and reshaping its partner and sales programmes. Those changes are intended to create a platform on which AI-oriented services can be sold more consistently alongside the core communications products.

Enterprise buying cycles remain stubborn

The company attributed part of its weaker first-half performance to cautious investment and longer decision-making cycles among larger customers, which complicates the common assumption that intense interest in AI translates immediately into recognised software revenue. Contact-centre and voice automation often touch customer-facing processes, data, compliance, and existing communications infrastructure, leaving deployments requiring more integration and operational approval than a standalone productivity tool.

An AI assistant handling calls, for example, needs to connect with customer records, routing systems, escalation processes, and potentially regulated information. Errors are also visible directly to customers, which raises the threshold for testing and governance before the system can move into production.

The UK provides a particularly useful test because NFON reported weaker performance there in Business Telephony and has realigned its sales structure in response. British organisations have a mature cloud communications market but also face a crowded vendor landscape in which Microsoft, Cisco, Zoom, Google, and other large platforms increasingly bundle communications and AI into broader subscriptions.

A specialist provider therefore has to demonstrate that its narrower focus produces enough additional value to justify another commercial relationship. AI features embedded deeply into voice and contact-centre workflows could provide that differentiation, but revenue has to become large enough to show that customers are buying the functionality rather than merely testing it.

The numbers will make the transition visible

NFON’s recurring-revenue structure should make the next stages relatively easy to assess. If Intelligent Assistant and Customer Engagement move from fast-growing smaller products into material businesses, the change should begin to appear in group revenue, seat economics, margins, and a return to overall growth rather than remaining confined to management commentary.

The opposite is equally true because continued contraction in telephony raises the amount of new AI revenue required merely to keep the group level. NFON has already acknowledged that the positive contribution from the strategic growth areas will not fully compensate during 2026 for weaker Business Telephony.

Many established European software vendors face a similar transition as they add AI to mature portfolios. Existing customers and distribution provide an advantage, but the technology has to become economically meaningful before weakness in the legacy product erases the gain.

NFON’s first-half results therefore show a transition that has started without yet changing the direction of the company. Its AI services are generating the type of recurring growth management wants, while the larger business around them is shrinking, leaving the next reporting periods to show whether the new products can become large enough to alter the group’s economics rather than simply improve its product story.