Dear all,

“German corporations are the most risk-averse in Europe. France is far better.”

A senior EU official told me this last week when we talked about how to drive deep tech collaborations between enterprises and Deep Tech startups. It stuck with me because it’s half-true — and the diagnosis isn’t complete.

Germany’s reputation for risk aversion is backed by data. On Hofstede’s Uncertainty Avoidance Index, Germany scores 65 — notably higher than the UK (35) and Sweden (29), though lower than Italy (75) and Poland (93). It’s a proxy for how strongly societies prefer predictability and rules over ambiguity.

Yet fear of failure doesn’t seem to be the decisive factor. According to the Global Entrepreneurship Monitor (GEM), 45% of Germans cite fear of failure as a barrier to starting a business — lower than the UK (57%) and comparable to France (42%).

One key issue around risk in Germany is rooted in a century of Ordnungspolitik: the belief that good rules drive good outcomes. Create a process or a law and you get more business impact — or a better society. But in a fast-moving environment, more rules don’t eliminate risk. They displace it. By suppressing experiments in the name of process and compliance, we push uncertainty into the future, where it reappears as systemic risk: missed technology waves, fragile supply chains, and the quiet erosion of talent. The ifo Institute estimates that excessive bureaucracy now costs the German economy up to €146 billion annually in lost output. This is the price of buying an illusion of certainty.

What Germany needs now is not less “Ordnung”, but a new mental model: good outcomes don’t come from good rules alone — they come from good feedback loops. Rules should set safety rails, not build roadblocks. That requires at least three principles:

Consider Open Bosch. Bosch runs a high-volume of PoCs annually as a structured operational routine. By creating a standardised purchasing fast lane that bypasses its own bureaucracy, Bosch reduced the typical time-to-contract for startups. It is a machine that ingests external Deep Tech—from quantum sensors to AI logistics.

Then there is the BMW Startup Garage. This unit pioneered the “Venture Client” model: buying product, not equity. The data is unarguable. Since 2015, this systematic pipeline has graduated over 220 startups. BMW transformed the risk of working with young companies into a measurable supply chain advantage.

Finally, a more visual example of leadership: Henrik Andersen, the CEO of Vestas, staged a literal funeral for “Failure Is Not an Option”—a doctrine once core to their employment guide. He replaced it with “Failure Is Always an Option.” This shows how large organisations can go back and rewrite even the oldest mottos of the company.

Wishing you a week of early failures and faster learning,
Martin

➔ Architect Carl Fredrik Svenstedt is planning extra-ordinarily pretty data centres for Evroc, built for example from natural stone - some say, it’s the Le Corbusier of hyperscalers.

➔ Despite being founded only in 2024, Nscale was named one of Europe’s 100 hottest startups, highlighting its rapid rise in the AI infra scene.

➔ Verda operates multiple renewable-powered data centres in Finland and Iceland and has submitted a proposal to build one of Europe’s first “AI gigafactories” with up to 100,000 accelerators.

➔ Since launching its Hyperstack service in 2023, NexGen Cloud has seen 2,272% year-over-year transaction growth on its platform and serves over 10,000 users.

➔ Originally named Hive, the company rebranded to Hivenet in 2024 to reflect its mission of creating a distributed cloud using unused storage and compute from user devices, reducing reliance on traditional data centres.

If you asked almost anyone today whether they use AI in their everyday work, very few would say no. Not least because using AI has become the expected answer. After all, the saying goes: AI will not take your job — but someone who knows how to use it will.

But how true is that in practice? 

And do enterprises fully embrace AI and are integrating it as they could?

OpenAI’s “The State of Enterprise AI — 2025 Report” is based on de-identified aggregated usage data from over 1 million business customers and a survey of 9,000 workers from across ~100 enterprises. It shows how AI is being used in organisations today.

The results are clear. AI is not just talk — people are genuinely using it, and usage is set to grow even further.

Here are five essential takeaways from it:

The SPARTA Defence Innovation Summit, held on the sidelines of the Munich Security Conference, is shaping up to be the most mission-critical gathering of leaders who define, fund, and deliver Europe’s defence capabilities—directly impacting European resilience, competitiveness, and sovereignty.

With McKinsey & Company, Destinus, and Rolls-Royce as strategic partners helping to shape the summit’s direction, SPARTA brings together generals, senior procurement leaders, and political decision-makers from key NATO nations, alongside defence primes BAE Systems, Rohde & Schwarz, HENSOLDT, MBDA, KNDS, TKMS, and RENK Group as well as scaleups Helsing, Quantum Systems, Morpheus Space, The Exploration Company, and Dronamics.

High-level political and military leaders of NATO, who cannot be named for security reasons, are expected to deliver keynote interventions, setting the strategic frame for Europe’s defence and industrial priorities.

Beyond the stage, attendees will engage in highly curated, pre-scheduled, double opt-in 1:1 meetings purpose-matched to accelerate commercial partnerships..

The focus is on Europe’s most pressing defence technologies: precision and strike drones; aerial and drone defence; ISR (drones, satellites, sensors, analytics); space systems and counter-space; autonomous ground and logistics robotics; maritime and underwater defence, and more.

Three invite-only roundtables, shaped with Lt Gen Sir Tom Copinger-Symes and McKinsey, will convene senior leaders from key NATO nations, Tier-1 primes, and top defence scaleups to tackle the most critical challenges Europe is facing today.

➔ SPARTA is invite-only. Request an invitation.

What does it really take to build an intelligent machine?
How does a machine actually learn to think?
And why is everyone suddenly talking about it?

In The Thinking Machine, the FT Book of the Year, journalist and technology expert Stephen Witt tells the inside story of how modern AI came to be, through the rise of deep learning, data, and the people behind it.

This is not another “AI will change everything” manifesto. It’s a smart, entertaining deep dive into how AI works, what makes it powerful, and what that means for anyone building, investing in, or competing with technology today.

Stephen Witt is known for turning complex technological shifts into compelling narratives. He is an award-winning journalist and author who has written for The New Yorker, The Financial Times, and The New York Times, specialising in technology, power, and the people behind transformative ideas.

➔ We’re fully aligned with our investment partner Apex Ventures on this book. As Dr. Golnaz Borghei put it, understanding NVIDIA’s journey is essential for European players looking to build sovereign capabilities in chips and AI.

➔ Want to win one copy? Reply “AI” to enter the draw!