Bitkom Start-up Survey 2026: AI Costs Hires Before Jobs

The Bitkom survey of 102 tech start-ups published on 11 August 2026 shows that AI changes employment first through positions that are never filled. 27 per cent of the start-ups held back on new hires over the past twelve months because of AI, while 7 per cent cut jobs for that reason. For the next twelve months, 28 per cent plan to forgo hires because of AI and only 2 per cent plan job cuts. The survey is not representative, yet its pattern is instructive for mid-sized companies, because there the effect of AI arises when vacancies are refilled. Our calculation for a company with 300 employees and 8 per cent annual staff turnover shows that ten office positions can disappear within three years without a single dismissal, provided management reviews every vacancy before it is advertised.

The Bitkom survey shows the effect of AI on start-ups mainly as forgone hires

Anyone looking for the employment effects of AI in tech start-ups finds them mainly in positions that were never created. Over the past twelve months around one in four tech start-ups (27 per cent) held back on new hires because of AI, while 7 per cent cut jobs for that reason, the German digital association Bitkom reported on 11 August 2026. At the same time 16 per cent hired additional staff because of AI. In half of the start-ups AI has no effect on staffing needs. All start-ups surveyed use AI in their business processes. According to the press release, Bitkom President Ralf Wintergerst said that AI makes many office tasks and software development in particular more efficient.

On average the start-ups surveyed employ 12 people and have one open position, after 13 employees a year earlier and 15 in 2024, according to Bitkom. The press release does not say whether this decline is linked to AI, so we do not attribute it to any cause.

Over the next twelve months start-ups plan to forgo hires 14 times as often as to cut jobs

Forgone hiring remains stable in the start-ups' plans while job cuts decline. Over the next twelve months 28 per cent want to forgo new hires because of AI, 2 per cent want to cut jobs for that reason and 19 per cent want to hire additional staff. 47 per cent see no effect, while 7 per cent are unsure or gave no answer.

Looking back, almost four start-ups held back on hiring for every start-up that cut jobs because of AI, while the outlook shows 14. Both questions thus support the thesis that the employment effect of AI shows up first as unfilled positions. They also qualify it, because almost one in five start-ups wants to hire additional staff because of AI. A net effect on the number of jobs cannot be calculated, because the press release neither gives the number of positions affected nor states whether multiple answers were possible. The shares in the outlook add up to 103 per cent.

A sample of 102 start-ups limits the finding to a pattern

The survey shows a direction while a projection to all start-ups is not possible. Bitkom describes the result as not representative but as a meaningful picture of sentiment. One percentage point corresponds to roughly one start-up, so the 7 per cent that cut jobs are around seven companies.

Even with random selection, the statistical margin for a share of 27 per cent with 102 respondents would be around plus or minus nine percentage points (95 per cent interval, own calculation). For a share of 7 per cent it would be around plus or minus five points. The lower bound for forgone hires of around 18 per cent is still above the upper bound for job cuts of around 12 per cent, so the gap between the two answers exceeds random variation.

Size and growth matter for mid-sized companies. A start-up with 12 employees that forgoes a hire grows more slowly. A mid-sized company with 300 employees often grows little and has departures every year that it needs to refill. The pattern still applies, because in both cases the decision is made before a position is advertised. Which tasks AI changes is described in AI and jobs: which tasks change.

In mid-sized companies the effect of AI arises when vacancies are refilled

In mid-sized companies the effect of AI shows up first in the refill rate, because departures free up positions every year without anyone being dismissed. A company that uses AI in accounting or order processing can reduce demand there by not refilling vacancies.

Without a review of each vacancy, however, this approach cuts positions at random. Departures are spread across functions according to employees' life circumstances, whereas the potential of AI is spread according to tasks. A blanket hiring freeze therefore also hits functions whose workload stays the same, so the remaining employees take on the work of the vacant position without relief from AI. How to calculate demand and supply separately for each function is described in Strategic workforce planning for three years.

A company with 300 employees can save ten office positions without dismissals

For a company with 300 employees and 8 per cent turnover, departures and a few internal transfers are enough to remove ten office positions close to AI within three years, provided management reviews every vacancy individually. Any executive team can replace the assumptions with its own figures.

In the 25 positions close to AI, 6 employees leave over three years (8 per cent times 25 times three). If the company does not refill these positions, six of the ten disappear by themselves. In the other 75 office positions 18 employees leave over the same period. If the company fills four of these vacancies with trained employees from positions close to AI, the target of ten positions is reached. It fills the remaining 14 vacancies externally. The office functions shrink from 100 to 90 positions and the workforce by 3.3 per cent without a single dismissal.

If the company instead leaves every third office vacancy unfilled across the board, eight positions disappear over three years. Because departures are spread evenly, only two of them are in functions close to AI and six are in functions whose workload stays the same. The company saves fewer positions and creates overload where AI provides no relief.

A position only counts as gone when the function handles its workload with one person fewer three months after the departure, because forgoing a hire without measurement easily leads to overtime and backlogs. For retirements every departure requires a handover, as described in Knowledge management: retaining knowledge when people leave.

Entry-level roles and apprenticeships need a separate decision in the headcount plan

When positions close to AI disappear, the company often loses the roles in which newcomers got to know customers, products and processes. Without a separate decision it will lack successors in a few years for the experienced roles whose holders retire. The company therefore derives the number of apprenticeships and entry-level positions from known retirements over the next three to five years. The entry-level role is redesigned: newcomers check and correct the preparatory work of the AI application. In doing so they learn the same processes as before. Which roles should be filled from within is covered in Talent management: identifying and filling key roles, induction in Onboarding as a productivity lever.

The works council must be informed about the refill plan itself

A refill rule that lets positions disappear because of AI is part of workforce planning. Under section 92 BetrVG of the Works Constitution Act the employer must inform the works council about it in good time and comprehensively on the basis of documents, including future staffing needs and vocational training measures. Transfers require the works council's consent in companies with, as a rule, more than twenty employees entitled to vote (section 99 BetrVG). If employees' skills are no longer sufficient for changed tasks, it has co-determination rights over in-company training measures (section 97(2) BetrVG). Co-determination over the AI application itself is covered in Works council and AI introduction. This overview does not constitute legal advice.

Executive teams in mid-sized companies should review every vacancy before advertising it

Executive teams in mid-sized companies can derive an approach from the Bitkom survey that manages the effect of AI through refilling decisions:

1. Introduce a vacancy review: from September 2026, management and the head of department decide on every office vacancy before it is advertised. The metrics are the share of vacancies reviewed and their split across the three decisions.
2. Identify positions close to AI: by the end of October 2026, heads of department and HR record for each function the tasks that AI can take over. Success means a target headcount up to 2029 exists for every office function.
3. Draw up the 2027 headcount plan with a refill rate: by the budget round in December 2026, controlling and HR set for each function the share of departures to be refilled. The metric is the quarterly deviation.
4. Measure workload after every position removed: three months after a departure that was not refilled, the head of department checks lead times, backlogs and overtime. If they rise, the position is filled after all.
5. Set entry-level roles and apprenticeships: before apprenticeships for 2027 are advertised, the training manager and HR development derive the number of entry-level positions from retirements. The metric is the share of experienced roles filled internally.
6. Involve the works council early: management presents the vacancy review and headcount plan in autumn 2026 and agrees rules for training and transfers by the end of March 2027, before the first position disappears under the new review.

Niehues Consulting combines organisation design with AI-aware workforce planning

Niehues Consulting supports executive teams and heads of department in building organisations that deliver reliably. Organisation design creates structures, roles and decision paths. 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, followed by introduction in the business units.

The approach follows the Niehues Delivery Model with its phases Understand, Build and Introduce. In Understand it clarifies which tasks AI takes over in each function and where employees will leave. In Build the vacancy review and redesigned roles are created, while Introduce involves managers and the works council. Founder Marin Niehues, a management consultant from the delivery side, builds organisations that deliver. His projects range from venture-backed scale-ups to DAX 40 groups, with assignments in more than 20 countries on six continents. In the free 30-minute initial consultation you assess together which positions in your company are close to AI and how your 2027 headcount plan should handle refilling: book a consultation.