Upside

01:35–10:24 — UK data centres: blocked… but a way through
A £1bn London-area data-centre project gets halted after a planning/EIA mess-up → delays of 9–12+ months, potentially worse.
The twist: data centres are being upgraded to “nationally significant infrastructure,” enabling central-government fast-track (DCO route) instead of local NIMBY planning.
Still: EIAs + judicial reviews can keep slowing everything down. Core complaint: the system keeps “deciding whether to build” after we’ve already decided we must.
10:24–20:01 — Davos: in a fractured world, can middle powers go solo?
Big takeaway: the old global order isn’t coming back. Middle powers doing bilateral deals risk being picked off one-by-one. The only viable strategy is bigger blocs + coordination.
On AI/robotics: Europe shouldn’t obsess over winning foundation models; the opportunity is physical AI (robotics, manufacturing, automation) layered onto Europe’s engineering base—if politics and fragmentation don’t smother it.
20:01–22:22 — AI vs jobs: don’t overkill the headline
Entry-level postings are down since early 2023, but the consensus here is: macro + rates explain most of it. AI will reshuffle work (especially junior/clerical tasks), but mass unemployment isn’t the base case.
22:22–27:46 — EU Inc / “28th regime”: real momentum and real resistance
EU Inc aims to make a pan-EU startup entity that’s fast/cheap to set up (48 hours, no minimum capital), plus simpler ESOPs (ideally tax deferred until liquidity).
The fight now: Regulation vs Directive
Regulation = uniform + immediate, but needs unanimity
Directive = easier, but invites delay + fragmentation
Expect pushback framed around labour standards and “race-to-the-bottom” fears.
27:46–34:51 — Has China already won AI?
Reframe “winning”: it’s not god-like AI dominance; it’s economic + military power with AI as a lever. Models converge fast, advantages erode, and the “months not years” gap matters.
Europe’s real risk is strategic irrelevance unless it scales power: capital markets, energy, defence capacity, and political cohesion (with a nod to the UK needing to be onside).
34:51–39:30 — US science funding “collapse”: brutal in pockets, not total
Big cuts and cancellations are real—especially in politically sensitive areas—but most US R&D is private sector, and defence-linked R&D keeps growing. Europe is trying to attract researchers, but this doesn’t yet look like a permanent talent migration.
39:30–43:13 — SaaS: dead? no? trapped?
SaaS faces a fork:
Mature into a cash machine (cut bloat, optimise margins), or
Become a “system of context” by embedding AI/agents deeply
Why the pain: ZIRP-era bloat + expensive orgs + incentive traps.
 Bright spot: incumbents with distribution are already monetising AI add-ons at meaningful ARR.
43:13–45:47 — Defence IPO era: the Overton window moved
A blockbuster European defence IPO becomes the poster child for a broader trend: defence re-rated, ESG lines shifting, and a growing pipeline as European defence budgets rise for the next decade.
45:47–47:02 — Quick hits
Billion-scale European fund raise gets a shout-out. More big AI deals bubbling. New fund launch focused on robotics/manufacturing, positioned as aligned with what Europe needs next.

BBC yeets kids shows onto YouTube, EU does a mega trade deal, Open Cosmos tries to be European-ish Starlink, Meta buys nuclear vibes, UK bins digital IDs (again), Anthropic’s “CoWork”, Grok vs governments, plus Deals of the Week.
Basically: geopolitics, space, energy, AI, and British admin disintegration.
(00:34) BBC on YouTube: “iPlayer walked so Netflix could run”
Dan: YouTube is enormous; BBC is adapting for younger audiences (and maybe… survival mode).
Mads: iPlayer was genuinely visionary; regulators stopped BBC going too commercial back then.
Andrew: Stop geo-blocking. Just take my money. (“Not in your region” = crime.)
(02:51) EU–Mercosur trade deal: 25 years, 700M people, farmers furious
Biggest-ever EU trade deal: Brazil, Argentina, Paraguay, Uruguay. (03:30)
Why now: EU wants options beyond US gridlock + China dependence.
(06:07) Open Cosmos: “EU Starlink?” …Not quite, but it’s something
Takeaway: not a consumer Starlink clone; more “secure, sovereign comms” for governments. (07:56–08:32)
(08:32) Meta’s 20-year nuclear power deals: AI runs on electricity (and contracts)
Hyperscalers locking in long-term nuclear PPAs in the US; Europe stuck with slower buildout, planning pain, and NIMBY boss fights.
(10:12) UK drops digital IDs: “25 incompatible IDs is the national strategy”
Mads: Digital ID is foundational (identity + access). Most OECD countries have it; UK is the holdout. (10:45–12:14)
UK backlash feels emotional; we already have loads of IDs that don’t talk to each other.
Cost debate: rollout ~£2bn-ish vs long-term fraud reduction. (13:33–13:48)
(15:35) Anthropic CoWork: “built in 10 days… by Claude Code”
CoWork: “Claude Code for knowledge workers” — chat UI + sandboxed VM. (19:31)
It’s early/buggy, but the meta-point is wild: AI building AI products at startup speed. (18:17–18:34)
(20:30) Yann LeCun “world models”: why LLMs aren’t the whole story
LLMs can talk; they don’t understand physics.
Four buckets: video prediction, interactive simulators (e.g. “move left/right”), physics engines, and latent world models (JEPA). (22:41–23:27)
Robot “pick and place” success rate cited as a compelling signal for JEPA. (23:27–24:15)
(28:13) Grok: sexualised imagery = policy grenade
Group consensus: when it’s minors, “platform self-policing” isn’t cutting it.
(32:19) JPM Health conference: biotech meets the LLM invasion
Nvidia + Eli Lilly: $1B partnership for AI drug discovery. (33:28)
Big pharma facing $200–$300B revenue going off-patent → likely acquisitions spree. (33:28–33:53)
Torch acquisition: “unified medical memory” pulling from records/wearables/visits. (36:54–37:10)
Cool paper alert: stroke triage via CT platform cuts transfer time by 64 minutes; “2M brain cells die per minute.” (38:57–39:25)
(42:19) Deals of the Week — “capital markets therapy session”
Quantinuum files confidentially for IPO; quantum + encryption randomness today, “R&D roadmap” tomorrow. (42:32–43:52)
Aikido Security (Ghent) hits unicorn: $60M Series B led by DST. (46:41–47:05)
Parloa (Germany, call centre automation): $350M Series D @ $3B, six months after Series C. (46:41–47:05)
Equal1 (Ireland, UCD spinout): $60M for quantum servers in data centers. (47:20)
Harmattan AI (France, drones): €200M Series B @ €1.4B, Dassault invested; supplying drones incl. UK Army contract mentions. (47:36–48:12)

Hosts: Dan Bowyer, Lomax Ward, Mads Jensen
Episode summary
This week we cover Trump’s Greenland rhetoric and the broader implications for NATO, European defence spending, and sovereignty. Then we dive into the tech and markets that (quietly) moved: Meta’s acquisition of Manus, Octopus/“Kraken” spinning out as a major energy-software player, Discord’s IPO momentum, Revolut’s Turkey expansion via license acquisition, and why UK pensions are underperforming (plus what reform could actually fix). We finish with a big CES-driven AI corner—Nvidia’s autonomous-driving stack “Alpamayo,” the China compute/memory bottleneck, AMD vs Nvidia software moats, Grok’s enormous fundraise, and why Claude Code is showing up even inside Google.
Topics
00:37 OpenAI “Health ChatGPT” launches (not in Europe) — what’s the deal?
02:26 FDA signals faster AI regulation: “move at the speed of Silicon Valley”
08:10 Trump + Greenland: history, Thule, and where’s the grift?
11:29 Pax Americana cracks
16:10 France: ban social media under-15s
18:21 EU regulation vs US deregulation: wedge widens
21:06 Meta buying Manus
22:27 Kraken (Octopus) spinout: big UK tech hiding in plain sight
25:02 IPO watch: Discord momentum + how it monetises
26:15 Revolut → Turkey via FUPS: buying access and licenses globally
29:19 UK pensions: underperformance + pushback on forced private allocations
31:57 Why pensions underperform
39:09 German dentists pension lawsuit + “shrimp farm VC” cautionary tale
40:27 Markets: FTSE hits 10,000 + quietly outperforms S&P
43:15 Biotech rebound + psychedelics
47:02 CES 2026... 
48:05 Nvidia’s “Alpamayo”: full autonomy stack + open models/data
50:34 Nvidia vertical apps vs chips: execution risk outside core semis
52:07 China + H200s: memory bottlenecks and geopolitics of compute supply
53:20 AMD vs Nvidia: the CUDA/software moat
54:35 Grok raises: $20B at $230B
56:44 Claude Code: adoption even inside Google
57:25 Deals of the week: Faculty exit to Accenture
59:47 Biographica raises to build next-gen crop tech
Notable moments / quotables
“The only people using cash or violence to negotiate are mobsters.” (12:32)
“It’s super easy to get autonomy to 99%… the last 1% takes 10 years.” (48:52, referenced)
“Much better to pay 2% to deliver 15% than 0.1% to deliver 1%.” (30:25)
Deal of the week
Lomax: Faculty reportedly sold to Accenture (UK AI/enterprise decision intelligence)
Dan: Kraken spun out of Octopus Energy (energy metering + grid/flows platform)
Mads: Biographica raises to build technology for next-gen, more resilient crops

Guests: Dan Bowyer, Mads Jensen, Lomax Ward and Dan Gray
Part 1 — 2025 predictions: what hit, what missed?
02:59 — OpenAI “dethroned”: debate, but mostly “yes”
04:47 — China/Taiwan “decisive year”: mostly no
05:46 — Defence tech surge: yes
06:46 — Europe/UK tech sovereignty: yes-ish but hard to measure
07:25 — Space enabling tech keeps attracting capital: yes
07:30 — US recession H2 2025: nope
09:07 — Stock correction early 2025: correct-ish
09:45 — M&A rebounds: yes
10:24 — IPO window reopens: yes-ish
---
Part 2 — 2026 predictions (with yardsticks)
11:29 — (1) Dan Gray: “European Re-industrialisation”
Big industrial families + family offices start allocating more directly into innovation (seen around Munich/TUM).
13:11 — (2) Dan Bowyer: “Apple wins personal AI”
Dan bets 2026 is Apple’s “Siri actually works” year—especially via on-device models + partnerships (Google mentioned).
16:35 — (3) Mads: “3 major tech IPOs”
From this list: SpaceX, Databricks, Canva, Anduril, Anthropic — pick 3/5.
Mads goes big on Anthropic growth + enterprise leadership.
Dan Gray adds: IPOs cluster; post-IPO M&A often spikes; real test is 6–9 months later.
19:08 — (4) Lomax: “Biotech comes back from the dead”
Biotech rally underway (XBI up hard off lows).
24:39 — (5) Dan Gray: “Politics kills the 28th regime”
He doubts a single EU incorporation regime survives politics, but suggests a workaround:
28:51 — (6) Mads: “Chinese open-source AI hits 60% of downloads”
Notes China open-source share ~44% by end of 2025 (per the conversation), and cites growing adoption of open-source models in startups.
30:11 — (7) Lomax: “Longevity clinics go from boutique → (somewhat) mainstream”
Thesis: preventative, subscription-style health scales (Neko, Function Health examples).
33:31 — (8) Mads: “EU expands tariffs on Chinese goods beyond EVs”
Europe stops pretending tariffs are morally evil, and starts protecting industrial base more aggressively (supply chain breadth and/or higher rates).
34:50 — (9) Mads: “No AGI in 2026”
AGI definition mess continues.
40:19 — (10) Lomax: “Cyber becomes a clear and present threat”
AI lowers cost of recon, phishing, persistence; cyber as statecraft sits below “war thresholds.”
43:14 — Bonus (Mads): Robotaxis
Waymo keeps lead in the West (incl. London expansion/testing), Tesla makes progress but stays behind.

With European VCs Lomax, Mads, Andrew and Dan
2025 year-in-review for European tech: capital, unicorns, geopolitics, AI, health, defence, and space space spaaaace.
01:25 – European tech in numbers
$45bn into European startups (flat vs 2023–24, still ~½ of 2021 peak).
28 new unicorns in 2025 (up from 14 in 2024).
US dominates: ~$250bn private tech funding.
VC fundraising still tight (~$10bn into European funds).
Exits are back: Wiz–Google, Klarna IPO, SMG IPO.
04:16 – 2025 in one sentence
Dan: “Extreme volatility.”
Lomax: “The final nail in globalisation.”
Mads: “China became a true peer to the US.”
Andrew: “Shipping fast beat gold-plated tech.”
05:49 – Most exciting tech moments
Mads: Claude Code, Chinese open-source AI, ASML–Mistral deal.
Lomax: Wiz’s $32bn cash exit.
Andrew: 6G moves from lab to real specs.
08:39 – Darwin Awards (biggest screw-ups)
Dan: Meta / Zuckerberg.
Lomax: The entire EU institutional stack.
Mads: Europe exporting founders to the US (incentive failure).
Andrew: Rachel Reeves & UK growth policy.
11:59 – What smart people got wrong
Mads: GPU bans didn’t stop China—backfired.
Lomax: AI bubble didn’t burst.
Dan: Investors piling blindly into defence.
Andrew: LLMs are “dial-up,” not the endgame.
16:42 – Why 2025 was a good year
Europe got a massive wake-up call (Dan).
4 new European decacorns (Lomax).
Founders kept building despite chaos (Mads).
VC rediscovered deep tech & hardware (Andrew).
21:23 – Geopolitics: a less naive world
Globalisation fragmenting into blocs.
Trust replaced by “trust but verify.”
Sovereignty = opportunity for European founders (AI, defence, energy).
34:30 – AI Corner: the year AI got real
DeepSeek shock wipes ~$600bn off Nvidia (Jan).
Claude Code: $1bn ARR in ~6 months.
Google Gemini comeback beats rivals on benchmarks.
China dominates open-source AI.
Rise of VLAs (Vision-Language-Action models) → physical AI, robotics.
Big question for 2026: “Are we the horses?”
47:44 – Health & bio highlights
GLP-1s everywhere: diabetes → cardio, kidney, neuro.
Sales ~$62bn, heading toward $120bn+.
Preventative health clinics scale (Function, Neko).
Biotech rebounds; AI-designed drugs hit Phase 2.
Psychedelics back: AbbVie deal, mental health momentum.
56:51 – Defence & space
Modern warfare now rewards fast shipping founders.
Global launch cadence: every ~1.5 days.
Space shifts from experimentation → permanent infrastructure.
Blue Origin finally launches New Glenn; SpaceX eyes Mars (again).