Job Cuts at IONOS and 1&1: AI Explains Only Part of Them

AI explains only part of the job cuts at IONOS and 1&1 Versatel, because the IONOS savings figure equals a calculation of saved salaries and an AI effect of the size measured in research only produces the 450 positions if it reaches the entire company. On 10 September 2026 the two United Internet subsidiaries announced around 800 full-time job cuts. 1&1 Versatel justifies this with fewer hierarchy levels, while IONOS names harmonised platforms, AI in internal workflows and an adjusted personnel structure without quantifying any share. Executive teams in mid-sized companies should therefore prove every source of savings separately and net, because the workforce and the works council examine each justification on its own.

1&1 and IONOS are cutting 800 jobs and give different reasons for the cuts

1&1 Versatel bases its cuts on its organisational structure, while IONOS names AI as one of three components. On the evening of 10 September 2026 United Internet announced in an ad hoc disclosure that the business customer subsidiary 1&1 Versatel is to shrink from around 1,350 to about 1,000 full-time employees. The target organisation provides for simpler management structures and fewer hierarchy levels so that productivity rises. For 2026 1&1 expects one-off expenses of around 60 million euros and is therefore lowering its EBITDA forecast from 800 to around 740 million euros, reports Börsen-Zeitung on 10 September 2026, citing dpa-AFX. After full implementation 1&1 expects an annual contribution to earnings of around 25 million euros from the 2028 financial year.

On the same evening IONOS announced in its own ad hoc disclosure that it will reduce its full-time headcount from around 3,800 to about 3,350, roughly half in Germany and half abroad. The company lists the harmonisation of technical platforms and processes, the increased use of artificial intelligence in internal workflows and the adjustment of the personnel structure as components. The reduction is to run mainly through voluntary programmes, so the savings of up to 30 million euros a year from 2027 depend on the take-up rate. Börsen-Zeitung arrives at group savings of 60 million euros on 11 September 2026 because it puts IONOS at 35 million euros, while United Internet names around 55 million euros.

The IONOS savings almost exactly match the personnel expense of 450 employees

The published IONOS savings are a calculation of saved salaries without visible costs for AI systems. Up to 30 million euros divided by around 450 positions gives around 66,700 euros per position and year. According to the 2025 Annual Report, 2025 personnel expenses were 274.0 million euros for an average of 4,101 employees, giving around 66,800 euros per head. Both figures are approximations, because the disclosure counts full-time equivalents while the HGB average counts heads including Sedo, which is up for sale and excluded from personnel expenses. Excluding Sedo the cost per head would be slightly higher.

The comparison matters for the AI rationale, because AI savings arise net. Licences, computing power, integration and maintenance of the systems cost money on an ongoing basis. If the annual savings equal the full personnel expense of the positions that disappear, these costs are booked elsewhere or included in the reinvestment in AI products. The disclosure is silent on this, so the net effect of AI cannot be seen from outside.

At 1&1 Versatel the one-off expense pays for itself after around 2.4 years of full effect, at IONOS after around 14 months, provided the savings reach the ceiling. One plausible explanation is the location of the positions, because Versatel operates in Germany, where partial retirement and severance cost more, while IONOS plans half of the reduction abroad. The disclosures do not confirm this, especially as the one-off expense may also include spending on IT and advisers.

AI explains the IONOS cuts only if it works everywhere as in customer service

An AI effect of the size researchers measured in customer service only produces the 450 positions at IONOS if it applies to the entire workforce of 3,800 full-time employees. Erik Brynjolfsson, Danielle Li and Lindsey Raymond analysed 5,172 customer support agents at a software company. With an AI assistant they resolved 15 percent more requests per hour on average, with less experienced agents becoming faster and better while the most experienced gained little time (Quarterly Journal of Economics, 2025). With the same number of requests, 15 percent more output per hour means around 13 percent less working time, because 1 divided by 1.15 is about 0.87.

How many positions this amounts to at IONOS depends on the size of customer service, which IONOS does not publish. According to Börsen-Zeitung the cuts mainly affect customer service and support locations where AI is already being tested more widely. An upper reference point is the share of personnel expenses that IONOS reports under selling expenses in its 2025 Annual Report, at 121.9 of 274.0 million euros or 44 percent, which also includes sales and marketing.

The calculation is generous because it converts every hour gained into a position cut, yet even so an effect in service explains at most around half of the reduction, so the rest has to come from harmonised platforms, the adjusted personnel structure or AI in development, marketing and administration. According to Börsen-Zeitung, group CEO Ralph Dommermuth said on the analyst call in early August that the potential of AI runs through the entire company, from development via marketing to the call centres. The disclosures do not prove such an effect. On 11 September 2026 CIO.de still ran the dpa report under the headline “Kahlschlag für KI” (slash and burn for AI), although the text mainly mentions AI as an area into which the savings are to flow.

Savings from AI and from fewer management levels need separate evidence

Savings from AI and from fewer management levels rest on different assumptions, so they are proven differently. For AI the assumption is that a system takes over tasks, which can be measured by the handling time per case before and after the introduction. Because the effect in the study was large for new agents and small for experienced ones, the same assistant produces different savings depending on the composition of a team. What this means for occupations is described in AI and jobs, the effect in service in AI in customer service.

For fewer management levels the assumption is that leadership tasks can be spread across fewer people without decisions slowing down. This is tested on the span of control and the number of levels between customer and decision, as described in Spans and layers and Span of control.

Works councils and labour courts examine each justification for job cuts separately

In Germany a job reduction of this size is usually an operational change under Section 111 of the Works Constitution Act (BetrVG), for which the works council has to learn the reasons. Management informs it in good time, consults on a reconciliation of interests and negotiates a social plan, which in a pure headcount reduction becomes enforceable above the thresholds of Section 112a BetrVG. The works council can only propose alternatives if it knows the cause. If AI is introduced, a co-determination right is added as soon as the system can monitor behaviour or performance (Section 87(1) no. 6 BetrVG), as described in Works council and AI introduction.

In the labour court a general reference to AI is not sufficient either. If a company removes a hierarchy level, it must explain concretely according to the case law of the Federal Labour Court how the tasks of the positions that disappear will be done in future without overloading the remaining employees. The same applies to a justification based on AI, because the loss of the need for employment must already have taken concrete shape when the notice is received. The process is described in Operational change in reorganisations. This assessment does not replace legal advice. How to prepare the messages and sequence of a reduction is described in Communicating layoffs.

Executive teams should separate savings by cause and calculate them net

Executive teams can justify each source of savings separately and verifiably:

1. Break the savings down by cause: in the four weeks before the decision, management and controlling assign each full-time position to be cut to exactly one cause. The step succeeds when the causes add up to the reduction target without a remainder.
2. Measure the AI share in a pilot: the head of a pilot area measures cases per hour with and without the AI system against a comparison group over eight to twelve weeks. Only the measured difference enters the workforce plan, separately for experienced and new employees.
3. Calculate net: in the business case, controlling deducts licences, operation, integration and onboarding from the personnel savings. If the savings per position equal the full personnel expense, costs are missing from the calculation.
4. Redistribute leadership tasks: before informing the works council, HR and the heads of department determine who takes over decisions and approvals of the management positions that disappear. The metrics are the span of control and the duration of typical approvals before and after the restructuring.
5. Inform the works council with the same breakdown: management and HR present the breakdown by cause before the decision, so that the reconciliation of interests and the social plan rest on the same figures as the planning.
6. Review the effect separately: after six and twelve months controlling reports for each cause whether the savings have been achieved. For the AI share the measures are handling times, backlogs and overtime, for the management structure the duration of decisions.

How to calculate the benefits and costs of AI reliably in advance is described in AI business case.

Niehues Consulting combines organisation design and AI management

Niehues Consulting supports executive teams and heads of department in building organisations that deliver reliably. Organisation design creates structures and decision paths, while in delivery management a delivery manager or product owner steers the programme until the change has arrived in operations. Where AI enters the workflows, a business case and a prototype are added. The approach follows the Niehues Delivery Model with its phases Understand, Build and Introduce. Founder Marin Niehues, a management consultant from the delivery side, builds organisations that deliver, in projects ranging from venture-backed scale-ups to DAX 40 groups in over 20 countries on six continents. In the free 30-minute initial consultation you assess together which tasks AI demonstrably takes over in your organisation and where the structure itself causes costs: book a consultation.