Trial & Error Folge 55 – „Klein und brutal selektiv“: Sind Accelerator noch zeitgemäß?

Accelerators are a bad business, two German operators conclude

Two German startup operators ask whether accelerators still make sense and end up arguing they are a poor business and a public good. Y Combinator shows how far the model can scale, but its brand is hard to copy and the German landscape is thin after years of failures. The value accelerators create is educational, so they belong at least partly in state hands, and neither of them would build one.

Takeaways

• Accelerator economics are poor, the value is educational. Fredrik Harkort says an accelerator is not a good business model and argues founder education should be at least partly state financed. Jan Thomas calculates 20 teams a year at 65,000 euros each, and Harkort says one in ten would have to become a unicorn.
• The ideal accelerator is small and brutally selective. Harkort would take at most 8 to 10 teams once a year instead of Y Combinator's 200 per batch, so the cohort stays small enough for group dynamics and group pressure. Around 80 percent of first-time founders would benefit from a good one.
• Y Combinator is the worldwide lighthouse. It has run since 2005, invests 500,000 dollars for 7 percent and more at a fixed valuation in the next round, and takes about 200 companies per batch. Its alumni include Stripe, Airbnb at 100bn dollars, DoorDash, Coinbase, Reddit, Deel and Rippling.
• Germany's accelerator vacuum is reopening. Many German models failed, and Jan Thomas sees little competition and room for new entrants. He calls EWOR, run by Daniel Dippold with a 50m euro fund, the European counterpart to Y Combinator.
• Corporate accelerators are a grave, Jan Thomas disagrees. Harkort advises founders against corporate programmes, while Jan Thomas argues they give access to infrastructure a startup cannot buy, such as labs at BASF or Bayer and programmes like Startup Autobahn with Plug and Play and BMW's Startup Garage.
• Coaches must have built and sold companies. Harkort says 90 percent of founders lack the same thing and that mentors should have founded and exited themselves. His own early mentor at BodyChange was Jochen from FlixBus, only a few years ahead of him.

The German accelerator landscape

• Super Public takes four or five startups, gives no money and takes no equity.
• Schub gives 25,000 euros for 2 percent to three to five startups.
• Antler and Entrepreneur First back talent before there is an idea.
• GoStudent Builders brings teams of one to three people to Vienna and requires them to ship AI-first. Jan Thomas questions the model because the company built there can be bought out at a fixed price.