"We have a lot of the world top universities you know the talent is clearly there... we just need to nurture it to give it the right space to build something." - Lorenzo [00:03:05]
"Deep tech is like almost at all-time high like just 4% from this peak and also growing 20% from last year... definitely there's some resilience and momentum in the deep tech market specifically." - Lorenzo [00:04:21]
Disclaimer: Orignal content owned by or sourced from third parties. It does not represent the views of 'Nuggets' platform or it's team. AI is used extensively across this platform including for summaries. Accuracy is not guaranteed, there can be mistakes. Any info or content on this platform is not a financial, legal, or investment advice. Do your own research. Refer for complete disclosures:- Terms of Use · Full Disclaimer
"If you look at the overall deep tech funding it's about 140 billion in the US across all categories compared to 20 billion in Europe... a 7x difference." - Nicola [00:12:50]
"We build these great companies in Europe, we create employment, we create IP, we create great products and then when we have to exit, three quarters of that value goes to the US. This I think really raises a sovereignty question." - Nicola [00:16:27]
"The scaling laws in LLMs... we're starting to see that tail off... there are going to be some applications where you can't just hand everything over to the models they're still giving you hallucinations." - Simon King [00:21:23]
"People are talking about well over the next couple of years, compute is going to take up 10% of the world's energy... we're now talking about putting data centers in space." - Simon King [00:30:50]
"When you've got a $10,000 drone that's taking out 50 to $100 million worth of military kit in planes and tanks... that's an asymmetric war that you need to adapt to and frankly the primes are not able to move quickly enough." - Simon King [00:49:16]
Speakers & Credentials
Lorenzo: Head of Research at Dealroom. Expert in ecosystem data mapping, market analysis, and reporting on startup funding trends in Europe.
Nicola: Partner at Walden Catalyst Ventures. A US-based fund investing globally in early-stage deep tech startups, operating out of Paris and covering the European ecosystem.
Simon King: Partner at Lakesar (European-focused generalist VC firm). Possesses 15 years of experience investing in deep tech, operating out of major European hubs including London, Berlin, Munich, and Zurich.
1. Executive Summary
The European deep tech ecosystem is experiencing unprecedented resilience, with funding nearly at an all-time high (only 4% off peak) and growing 20% year-over-year, starkly outperforming regular tech, which remains down 53% from 2021 levels.
Europe possesses a profound talent advantage, producing 1.5 million STEM graduates annually (double the US) and holding 20% of the world's top 10% cited academic papers, yet struggles to retain the resulting economic value due to a massive late-stage funding gap.
The US outpaces Europe in deep tech funding by 7x ($140B vs. $20B) and in AI funding specifically by 12x ($120B vs. $10B), predominantly driven by mega-rounds for foundational models and infrastructure (e.g., Nvidia chips).
Consequently, European founders are forced to look outward for late-stage capital; non-European investors accounted for 70% of funding in rounds over $250M in 2025 (up from 50% in 2024), and 75% of European startup exit value is ultimately captured by US acquirers.
Rather than attempting to match the US in capital-intensive foundational AI models, Europe's strategic path forward lies in application-specific AI, hardware efficiency (photonics, alternative interconnects), novel robotics tightly integrated with local industrial manufacturing, and next-generation defense tech driven by agile startups.
Regulatory momentum (such as the recent push for EU Inc. legislation) and evolving government procurement strategies are essential tailwinds needed to help European deep tech scale domestically and establish true technological sovereignty.
Definition & Momentum: Deep tech focuses on applying innovative scientific or engineering breakthroughs to commercial applications [00:01:47]. The current wave of generative AI has experienced unprecedented adoption speed, reaching 1 billion users in 3 months [00:02:40].
Diverging Trajectories: "Regular tech" (fintech, SaaS, marketplaces) suffered a massive correction and remains 53% down compared to its 2021 peak [00:04:08].
Deep Tech Ascendancy: In stark contrast, deep tech funding is extremely resilient, hovering just 4% below its all-time high and experiencing a 20% growth compared to the previous year [00:04:21].
Capital Allocation Shift: Currently, almost one-third of all startup capital in Europe flows into deep tech startups, which is double the allocation share seen 10 years ago [00:04:43].
Geographical Hubs: Paris overtook London for the first time as the top deep tech hub, driven by momentum in AI (e.g., Mistral), future compute (e.g., Alice & Bob), and robotics [00:05:27]. Munich leads in space and defense tech, while Zurich is the powerhouse for novel robotics [00:06:27].
European Talent vs. The Late-Stage Funding Crisis [00:06:48]
The Talent Pipeline: Europe is an undisputed academic powerhouse. It hosts a third of the world's top technical universities, and roughly one-third of all deep tech startups spin out directly from research institutions [00:07:36].
STEM Dominance: European universities produce 1.5 million STEM graduates annually, roughly twice the output of the US [00:08:04]. Furthermore, European research yields 20% of the world's top 10% most-cited academic papers [00:08:16].
The Growth Pipeline: The ecosystem now boasts nearly 8,000 deep tech companies that have raised over $100k, adding 900 new companies each year [00:10:04]. The total combined value of European deep tech is currently estimated at $690 billion [00:09:17].
The Funding Chasm: Despite leading in talent and early-stage creation (Seed/Series A), Europe faces a massive deficit in late-stage capital. Overall deep tech funding in the US is $140 billion vs. $20 billion in Europe (a 7x gap) [00:12:50].
Foreign Dependency: For mega-rounds ($250 million+), a staggering 70% of the capital is provided by non-European investors (a deterioration from 50% in 2024) [00:14:57].
The Sovereignty Drain: Because domestic late-stage capital is absent, exits are offshored. Roughly three-quarters (75%) of European exit value is ultimately captured by US acquirers (financial or corporate), meaning European IP and talent generate value that is extracted across the Atlantic [00:16:16].
The AI Race: Infrastructure vs. Application [00:19:19]
The Infrastructure Disparity: In AI specifically, the US invested $120 billion compared to Europe's $10 billion (a 12x gap) [00:13:14]. The vast majority of US funding (e.g., the $81B raised by OpenAI, Anthropic, and xAI combined) goes into brute-force infrastructure (Nvidia chips, data centers, energy) [00:13:39].
Strategic Pivot for Europe: Competing on capital-intensive foundational LLMs is deemed "impossible" [00:20:37]. However, the scaling laws of LLMs are beginning to tail off in both model size and loop-time processing limits [00:21:23].
The Trust Deficit: Broad LLMs cannot be trusted with mission-critical tasks due to persistent hallucinations. Simon King notes he runs offline, isolated models (e.g., on a Raspberry Pi) specifically to avoid data leakage and hallucination risks [00:22:08].
Europe's Elegance Advantage: Europe is pivoting towards "elegant, efficient, and trustworthy" AI, drawing a parallel to the European vs. American automotive industry [00:24:01]. Mega-seed rounds in Europe (e.g., AMI Labs in Paris, Ineffable Intelligence in London) reflect a focus on applied, efficiency-driven AI and specialized models rather than generalized brute force [00:23:32].
Talent Migration Parity: Over the last decade, AI talent heavily migrated from Europe to the US. Recently, this brain drain has halted, with net migration of AI talent reaching near-zero over the past two years as researchers return to Europe [00:27:36].
Future of Compute: The Energy Bottleneck & Photonics [00:28:05]
The Power Crisis: Hyperscaler compute demands are outstripping supply. Compute infrastructure is projected to consume 10% of the world's energy in the next couple of years, leading to radical proposals like positioning data centers in orbit to access solar power and bypass grid delays [00:30:50].
Inefficiency of Silicon: Current Nvidia chips, while powerful, are highly inefficient due to latency and heat dissipation issues tied to traditional copper interconnects and idle processing waiting for data delivery [00:29:27].
The Photonics Solution: Europe is aggressively backing photonics to solve the energy/interconnect bottleneck. Major recent rounds include $220 million for Olex (photonics compute) and a $250 million round for Axelera [00:28:18]. Transitioning from copper to optical pathways significantly mitigates thermal loss and data latency [00:30:30].
Quantum Leadership: Europe continues to secure massive funding for quantum compute hardware (e.g., EQM in Germany), though the timeline for practical market adoption remains long-term compared to immediate photonics integration [00:31:34].
Robotics: The Advantage of Industrial Proximity [00:38:48]
The Hardware/Software Paradigm Shift: Traditional robotics required months of rigid programming by specialized engineers. The new wave involves dropping an intelligent platform into a factory where it achieves 90-95% efficiency in hours and self-trains to 99.9% over two weeks [00:45:17].
Europe's Unfair Advantage: While China leads hardware manufacturing scale, Europe holds an "unfair advantage" in training data. Unlike foundational software AI, robotics requires physical data gathered inside factories. The extreme density of the European industrial base (specifically the DACH region) allows robotics startups to deploy directly to major manufacturers located minutes away, securing vital proprietary data [00:44:29].
The Humanoid Bear Case: Both investors expressed extreme bearishness regarding bipedal humanoids due to persistent mechanical failures (e.g., breaking hips). Instead, they forecast the rise of wheel-based, multi-armed autonomous platforms specifically tailored for flat factory floors and logistics centers [00:46:03].
The M&A Trap: The recent acquisition of Zurich-based robotics spin-out River (which raised a $22M seed round) by Amazon highlights a persistent failure: promising European startups are often bought out early by US mega-caps rather than scaling independently or being acquired by European industrial giants [00:38:48].
The Defense Boom: Defense and security tech claimed 43% of total deep tech funding, with pure defense startups capturing $1.8 billion (more than doubling year-over-year) [00:47:33].
The Asymmetric Reality: The nature of warfare has shifted to favor agile innovation over legacy defense contractors ("primes"). A $10,000 drone can consistently destroy $50-$100 million of legacy military assets (tanks, planes), creating an asymmetric dynamic that legacy contractors cannot counter rapidly [00:49:16].
The Procurement Mandate: Historically, the US DoD catalyzed its tech sector by purchasing 90% of domestic semiconductor output in the 1960s, whereas European governments bought only 5-10% [00:48:44]. European governments (notably Germany) are beginning to realize they must transition from providing grants to acting as primary customers for startup-driven defense tech.
EU Inc. Legislation: Recent policy movements (the "28th regime" or "EU Inc") are pushing for standardized digital incorporation across Europe within 48 hours. While imperfect, it signifies a major cultural shift where startup ecosystem friction points are actively elevated to the top of the EU Parliament's agenda [00:50:46].
The Reference Vault
4. Data & Figures
Data Point
Value
Context
Timestamp
Generative AI Adoption Speed
1 Billion users
Number of users achieved globally within just 3 months.
The Venture Ecosystem Flywheel: The conceptual model explaining how startup ecosystems mature. Early capital breeds small successes; founders and angels recycle that capital and experience back into the ecosystem, raising the ambition and fund size incrementally for the next cohort. Europe's flywheel is broken at the late stage because the massive exits are offshored, meaning the liquidity and expertise aren't recycled domestically. [00:37:28]
The "Bullshit Index" (LLM Hallucination Vulnerability): A mental model for testing the trust-boundary of LLMs. By providing a model with a true statement tweaked to be slightly false and asking it to formulate an argument, models will eagerly hallucinate to please the user. This proves that raw LLMs cannot yet be universally trusted for highly regulated or mission-critical enterprise deployments. [00:22:34]
The Proximity Data Moat (Hardware vs. Software AI): In foundational software AI, data is generally scraped universally from the web. In advanced robotics, physical training data must be acquired in situ. Europe's dense concentration of manufacturing facilities (especially in the DACH region) located minutes away from research hubs provides an unreplicable, geographical data moat for training industrial robotics that US or Chinese firms cannot easily access. [00:44:29]
Asymmetric Warfare Economics: A structural framework recognizing that legacy military spending is fundamentally misaligned with modern combat mechanics. Large defense primes build $100M, multi-year platforms, whereas startups iterate daily to produce $10K autonomous drones that can neutralize those legacy platforms. This shifts the geopolitical advantage from legacy manufacturing scale toward rapid, software-driven iteration. [00:49:16]
6. Anecdotes
The River M&A Paradox: A prime illustration of Europe's exit problem occurred precisely during the reporting week. River, a highly promising autonomous delivery robotics startup spun out of ETH Zurich, secured a $22M seed round. Instead of scaling into an independent European champion, it was acquired entirely by Amazon (whose family office had led the seed round) shortly after. This validated the quality of the tech but underscored the ongoing trend of US conglomerates strip-mining European talent and IP. [00:38:48]
The Quarantined LLM: To demonstrate the lack of enterprise trust in modern AI models, Simon King recounted running open-source models (like Open-Claw) strictly on an isolated network using a Raspberry Pi. Because these models are eager to please and prone to hallucination, he physically air-gaps them to prevent them from accessing or leaking his private network data, illustrating the danger of unconstrained enterprise deployment. [00:22:08]
The 1960s DoD Procurement Catalyst: Highlighting how governments can create industries, Simon pointed out that in the 1960s, the US Department of Defense explicitly purchased 90% of all domestic semiconductor output. This massive, guaranteed procurement—rather than mere grants—was the primary engine that built Intel and the entire US silicon ecosystem, contrasting sharply with European governments who bought only 5-10% and failed to incubate their own tech sector. [00:48:44]
7. References & Recommendations
Organizations, Companies & Startups Mentioned:
Dealroom (Research and data platform hosting the Deep Tech Report)
Walden Catalyst Ventures (Deep tech VC fund)
Lakesar (European generalist/deep tech VC fund)
OpenAI, Anthropic, xAI (US foundational model leaders raising $81B combined)
Mistral (Paris-based AI powerhouse)
Alice & Bob (Paris-based Quantum/Future compute startup)
AMI Labs (Paris-based applied AI startup receiving mega-seed funding)
Ineffable Intelligence (London-based applied AI startup receiving mega-seed funding)
Olex ($220M round for photonics/optical compute)
Axelera ($250M round for future compute)
EQM (German quantum computing leader proceeding with IPO)
River (Zurich-based robotics startup acquired by Amazon)
Anybotics (Leading quadruped robotics company)
Exotec (Iconic French warehouse robotics company)
Academic Hubs Referenced:
ETH Zurich (Switzerland - Powerhouse for robotics)
TUM (Technical University of Munich) (Germany - Defense and space)
Cambridge, Oxford, Imperial College (UK hubs)
Concepts/Policies:
EU Inc (The 28th Regime): Proposed regulatory framework designed to allow digital incorporation across all EU states within 48 hours to unify the fragmented market.
Jul 16, 2026
Secrets of building The Whole Truth | Shashank Mehta, Founder and CEO | Unstarted | 16 Jul 2026 | Z47 Moments
"I fundamentally cannot live with the gap between my do and my say i find hypocrisy very very putting off" Shashank Mehta 07:04 https://youtu.be/HA7kNZgkcT8?si=CyHcafj8CzT5cQBu&t=7m4s "if you craft your life around your weaknesses you will…
Deep Tech Funding Resilience
-4%
Deep tech funding is only 4% down from all-time highs and grew 20% YoY.