*By Daniel Szabo, GP & Co-Founder, Generation Tech Partners — Hamburg, 24 April 2026*
Yesterday, at PEC 26 in Berlin — the annual gathering of the most senior GPs and LPs in DACH software and tech private equity — I was invited to deliver three hot takes in the closing AI Rapid Fire panel.
The brief: each thesis should split 150 of the sharpest PE minds in the room 50/50.
I delivered. The paddles lit up. The room split.
What follows are the three theses, why I stand behind each one, and how they translate directly into how we invest at Generation Tech Partners.
Thesis 1: "AI won't replace the frontline. It will erase middle management."
The standard narrative is that AI will hit the frontline first. Call center agents. Cashiers. Drivers. Anyone doing repetitive physical or transactional work.
The narrative is wrong.
Middle management exists for one reason: to aggregate, filter, and route information between layers. That is the exact task AI now performs better, faster, and around the clock. Layers of team leads, regional managers, and VPs-of-something are a legacy of a pre-digital information economy.
The frontline, by contrast, is where human judgment still compounds. The closer reading the customer across the table. The service technician diagnosing the edge case no playbook covers. The operator deciding in the moment when the process breaks.
What it means for PE. Portfolio companies carrying four to six management layers are structurally overpriced. The EBITDA expansion from removing two or three of those layers and pushing AI tooling to the people actually doing the work is not a cost exercise. It is a business model redesign.
What we do at GTP. Within the first 100 days, we map every information-routing role and move decision-making as close to the customer as possible. AI tooling deploys first at the frontline — sales, service, operations — not at HQ. Flat beats fancy.
Thesis 2: "Bewertungsfaktor arbitrage is dead. AI is the new financial engineering."
For the past 20 years, private equity returns in the Mittelstand came from one dominant source: buy at one multiple, grow the business at market pace, sell at a higher one. Multiple arbitrage.
That playbook is finished.
Three forces are breaking it at the same time.
First, entry multiples have compressed. Auction processes, debt scarcity and limited target supply have pushed pricing for quality DACH B2B service assets up — leaving little headroom.
Second, exit multiples are capped. Strategic buyers are more selective. Secondary buyers are modeling lower IRRs. Nobody is paying a premium anymore for a generic services roll-up.
Third, the operational bar is rising. Investors increasingly expect earnings growth to come from operational change driven by AI, not from cost-cutting.
What was "financial engineering" in the 2000s — debt structuring, tax optimization, earn-outs — is now table stakes. The actual engineering problem is operational: deploying AI into workflows in a way that expands EBITDA at double-digit percentage points per year.
What we do at GTP. We underwrite no deal on multiple expansion. Every thesis stands or falls on the AI-driven operational case. If we cannot see a credible path to material earnings growth through transformation — not optimization — we pass.
Thesis 3: "The knowledge worker is dead. You are either a Deployer or an order-taker."
The knowledge worker — the mid-level professional whose value was built on domain expertise, judgment, and the ability to write a clean email — is no longer a coherent category.
Two roles remain.
The Deployer orchestrates AI to do the work of five, ten, or fifty people. Deployers don't write reports. They build systems that produce reports. They don't run processes. They design and supervise the agents that run processes.
The Order-Taker executes what AI tells them. They follow the script, process the ticket, complete the task. Their economic position is defined by what AI decides to assign them.
The middle — the senior analyst, the associate director, the manager-of-managers — is collapsing fast.
What it means for PE. The CEO profile that worked for the last cycle — the "seasoned operator with 20 years in the industry" — is not the right profile for a 2026 buyout. We need leaders who are Deployers by instinct.
What we do at GTP. In every portfolio company, we back a digital-native operator-CEO. Not a consultant. Not a career manager. Someone who has actually built, automated, and scaled with AI in the last three years. The difference in value creation is not incremental. It is structural.
The Pattern Behind the Three
These three theses are not separate arguments. They are three faces of the same underlying shift.
Private equity value creation has moved from the cap table to the operating model.
The firms that will generate top-quartile returns in the next cycle are the ones that treat AI as the new core competency — not as a portfolio initiative or an "AI center of excellence" slide.
That is why we built Generation Tech Partners the way we did. German Mittelstand B2B services. €0.5–5M EBITDA. Owner-managed succession situations. We buy, we transform, we build long-term value.
We buy companies. We don't advise.
We don't optimize. We transform.
---
*Generation Tech Partners is a Hamburg-based small cap PE fund acquiring owner-managed B2B service businesses in the German Mittelstand. Fund I: €60M, SFDR Art. 8. Target companies: €0.5–5M EBITDA (size criterion only — no indication of valuation, purchase price or target return). To discuss a succession situation or an LP commitment: daniel@generationtech.partners*