Venture Geopolitics Issue 39
17 March 2026
“Londonmaxxing” has been a welcome & wholly enjoyable trend. A playful reminder that San Francisco is just another dense cluster of talent far away & that the things possible there are possible wherever people have the focus & ambition.
In many ways, the meme is Europe emulating a distinctly American attitude. Having lived & worked in the US, the biggest delta is cultural. You are encouraged to sell the cookies, make the pitch, believe you can achieve anything.
The combined pressures of AI competition & geopolitical fragmentation are forcing Europe to rediscover its own voice. ‘Londonmaxxing’ is a signal that confidence & narrative matter. But the uplifting mood was complicated by another headline: Entrepreneurs First, an extraordinary talent incubator born in the UK, raised $200M while now being described as “based in San Francisco”.
This reflects a deeper tension. Europe’s structural advantage has always been talent density — the research institutions, the scientific depth, the intellectual seriousness. EF built its model on proving that many of the best founders in the world come out of our universities. They were right. The difficult question is what happens when that talent is systematically incubated here, but subsequently shipped elsewhere?
The idea that you must relocate to the Bay Area to win feels increasingly outdated & unhelpful. Founders should absolutely absorb the energy of the US — it is often catalytic. But it has never been easier to operate with your feet on both sides of the pond: to access global capital (capital that often goes significantly further when you are not paying San Francisco salaries) & to build great companies from anywhere.
Raising ambition is not about changing location, but changing mindset. And as technology becomes more strategic, there is a growing case for choosing to build at home.
Building here is still hard, not least because markets remain fragmented (for now). Doing things ‘right’ is always harder - building around the complexity of humans, of societies & of cultures, of real regulatory & institutional fabric. But founders who succeed in navigating that complexity often emerge with something more durable. The work itself becomes a moat — not only because of the relationships & distribution built over time, but because the product must be meaningfully better. It has to function across languages, regulatory environments & cultural contexts. Companies forged in that environment are shaped less like growth vehicles & more like institutions. The types of institutions we need at this very moment in time.
Europe has SAP, Siemens, Bosch (America has Facebook & Snapchat). What I mean to say, is Europe builds institutions that outlast their founders, that serve societies, that compound for decades. The strongest European systems tend to encode, protect & evolve culture; the strongest American ones are often optimised to move faster than it. That difference shows up in the kinds of companies each ecosystem produces.
For those of us working in technology, this is a formative moment. The task is not to reject global ambition or capital, but to co-opt it — to channel that energy into building enduring outcomes here. Londonmaxxing may be a meme, but the question it surfaces is serious: where will the next generation of enduring companies be built, & how will that shape the societies we live in?
IPO / Publics
SoftBank’s PayPay (mobile payments platform) priced its IPO below range to raise $880M (here)
MDA Space (satellite & space infrastructure provider) completed $300M expedited US cross-listing (here)
Microsoft publicly backed Anthropic’s lawsuit against the Pentagon in a court filing, urging a judge to block its “supply-chain risk” designation despite holding major federal business (including part of the $9B Joint Warfighting Cloud Capability deal & a $6B+ GSA framework). CEO Satya Nadella’s stance implies the era of “corporate silence” toward federal policy may be fading (here)
AWS will deploy both in-house Trainium chips & Cerebras wafer-scale processors in a new inference offering, making Amazon the first hyperscaler to integrate Cerebras’ experimental ultra-large chips at scale, signalling a diversification of compute architectures as inference economics become central to AI profitability (here)
Meta delayed its next frontier model “Avocado” to at least May amid performance gaps vs Gemini & Claude. Despite BNs spent rebuilding its AI division under Alexandr Wang, internal benchmarks reportedly place the model between Gemini 2.5 & 3.0, raising strategic questions about whether Meta may temporarily license rival models for distribution across Facebook, Instagram & WhatsApp (here)
Atlassian will cut ~1.6k jobs to self-fund AI investment & enterprise sales expansion. CEO Mike Cannon-Brookes framed the move as redeploying capital toward AI growth, though critics argue productivity narratives are increasingly used to mask structural challenges in mature SaaS (here)
Alibaba plans to ride the country’s OpenClaw/agentic AI trend by releasing its own enterprise-ready AI assistant, built atop its Qwen family of models. The agent gets built-in access to the Alibaba galaxy of platforms (including Taobao & Alipay) (here)
Big Dogs
Anthropic will offer 1M-token context windows on Opus 4.6 & Sonnet 4.6 at no extra charge. Removing prior surcharges on large context suggests competition is shifting toward capability bundling & enterprise platform stickiness vs pure model pricing (here)
Anthropic also exploring AI consulting JV with Blackstone & other PE firms, a move that could formalise the emerging “services wrapper” around frontier models as corporates seek help operationalising agents across workflows (here)
Anthropic’s product velocity through the roof! In ~2 weeks the company shipped Claude Code Security, voice mode, free memory tools, a multi-vertical plug-in marketplace, structured outputs & web search GA, Sonnet 4.6, enterprise data feeds (FactSet, MSCI, S&P Global, LSEG), Google ecosystem integrations, legacy COBOL modernisation tooling & API data-residency controls, alongside acquiring Vercept. Much of the tooling is iteratively generated & refined by Claude itself, compressing development cycles & illustrating how frontier AI firms are entering a phase of compounding product surface expansion (here)
Anthropic commits $100M to Claude Partner Network ecosystem. The AI lab is deepening enterprise distribution by funding integrators & software partners building on Claude APIs, reflecting a platform strategy aimed at embedding models across regulated industries & vertical workflows (here)
Revolut secured a full UK banking licence (after a 4-year wait). This unlocks large-scale lending at home & strengthens case for approvals abroad as the $75B fintech targets deeper penetration across the US (here)
Musk sidelined additional xAI founders amid coding push struggles. Leadership churn highlights mounting execution pressure as firms race to monetise coding copilots & infrastructure bets (here)
Mira Murati’s Thinking Machines secured a multibillion-dollar chip supply deal with Nvidia (here)
Regulation
UK MPs rejected a blanket social-media ban for under-16s in favour of consultation, highlighting a more incremental European regulatory approach vs sharper platform restrictions seen in parts of Asia (here)
UK is consulting on a scaled-back national digital ID framework after earlier proposals faced backlash. Policymakers are attempting to balance identity verification infrastructure benefits with public concerns over surveillance, data centralisation & private-sector access (here)
UK will subject major water utility acquisitions to national security screening – ministers are expanding the Investment Security Unit’s remit to include critical water infrastructure deals amid rising foreign interest in distressed UK utilities. Officials argue water networks now sit alongside energy grids & telecoms as strategic assets vulnerable to hostile leverage or supply disruption. The move signals a broader trend of governments redefining “critical infrastructure” in economic security terms, particularly where private capital ownership intersects with sovereign resilience (here)
US banking regulators plan to ease post-2008 capital rules on lenders (i.e. reduced capital buffer requirements) potentially boosting credit supply & risk appetite at a time when private credit markets are under scrutiny (here)
Per Sifted, the leaked EU “28th regime” proposal falls short. While it would be introduced as a regulation (ensuring consistent implementation across states) it stops short of creating a fully unified company structure. Instead of an EU-wide registry, founders would file via a single portal with data routed to national registers, while legal disputes would remain in national courts. Ecosystem groups are pushing back, warning this undermines the core goal of standardisation (here)
Washington is trying to calm major sovereign wealth funds after proposed US tax changes raised fears of reduced foreign investment. New rules could expand what counts as “commercial activity”, potentially increasing tax liabilities on activities like direct lending or minority stakes. Funds such as Mubadala (>$100B invested in the US) & Qatar Investment Authority ($580B globally) warned investment could fall (here)
Venture Capital / Finance
Dragos Novac argues Europe’s venture market is structurally barbelled. “The reality check is a snapshot of capital under some sort of a cognitive dissonance - we’re in a bearish market, but with a specific bull flavour”. ~1K deals closed in the first 11 weeks of the year, but capital increasingly concentrated in AI mega-rounds & defence seed clusters while the scale-up layer continues to erode. $10M–$50M rounds are down ~44% YoY, signalling fragile graduation pathways from seed → growth. AI accounts for ~41% of deployed capital, which Novac frames as herding rather than pure conviction, particularly in an ecosystem that may already have 3–4x more VCs than its LP base can sustainably support, implying consolidation inevitable (here)
Per Dealroom, European unicorns have erased ~€123B in paper value as 60 of 199 fall below $1B. Fintech alone accounts for ~€64.8B in markdowns (53% of value destruction), underscoring how 2021 pricing assumptions rather than operational weakness are driving ecosystem recalibration (here)
Deutsche Bank flagged ~$30B exposure to private credit while Cliffwater’s $33B fund saw redemptions hit ~14%. Highlighting liquidity mismatches emerging in semi-liquid credit vehicles as institutional allocators reassess risk (here & here)
Carlyle paused fundraising to overhaul its team after struggles in its latest vehicle (here)
Sequoia argues the next $1T company may be a software company masquerading as a services provider - as enterprises shift from buying tools to buying completed work, the biggest winners will be vertically integrated AI businesses that look like services firms on the surface but operate with software-like margins. Historically, for every $1 spent on software around $6 has been spent on services; AI is now beginning to convert much of that services spend into scalable, outcome-driven software (here)
Entrepreneurs First raised $200M for its talent-first venture builder model, while continuing to encourage European founders to spend time in US markets to access customers & capital networks. Businesswire reported the story, declaring EF “a global talent investor & company builder based in San Francisco”. EF was founded by Matt Clifford & Alice Bentinck in the UK in 2011 (here)
New funds:
Spark Capital (first VC to back Anthropic) is in the process of raising ~$3B in new funds, 50% more than the size of funds it raised 2 years ago (here)
Elaia closed €134M for its third deep-tech fund focused on B2B pre-seed & seed (here)
Scout Ventures closed its fifth fund with $125M in commitments. The Austin-based VC will invest in companies building tech tied to national security & defence (here)
Emerald Technology Ventures reached €100M for Global Water Fund II – highlighting growing investor focus on resource infra innovation (here)
Samaipata launched a €110M early-stage AI fund (here)
UVC Partners secured €77M first close for a €150M growth fund – signalling cautious re-entry into later-stage deployment as exit visibility remains limited (here)
Venture Geopolitics
The Strait of Hormuz remains one of the most consequential chokepoints in the global economic system. ~34% of seaborne oil, ~22% of minerals & ~16% of fertiliser shipments transit the corridor, alongside aluminium, chemicals & jet fuel, underscoring how regional conflict can ripple through technology supply chains (here)
Qatar helium shutdown has triggered chip supply chain alarm. This removed ~30% of global helium supply, forcing semiconductor manufacturers like SK hynix to accelerate diversification efforts as geopolitical shocks increasingly threaten critical materials for advanced computing (here)
Trump warned NATO faces a “very bad future” if allies fail to back US operations linked to Iran. The comments signal a more transactional American security posture where alliance solidarity may increasingly depend on participation in expeditionary conflicts, intensifying European debates around sovereignty (here)
Germany rejected Trump pressure to support Iran operations militarily. Defence officials emphasised constitutional constraints & domestic opposition, exposing fractures within NATO over burden-sharing & intervention scope during high-intensity crises (here)
US temporarily lifted sanctions on Russian oil already at sea as prices spiked toward $100 (here), while Zelenskyy urged Europe to stop importing Russian oil funding Moscow’s war effort. He highlighted the contradiction between Europe’s geopolitical stance & persistent energy dependence, particularly as global oil volatility complicates transition timelines (here)
UK announced a £45M AI supercomputer (“Sunrise”) to support fusion energy development. With massive computing power, it aims to enable digital simulations of reactors, accelerating research, reducing costs & strengthening supply chains around AI, materials & superconductors (here)
UK LibDebs called for a “genuinely independent” nuclear deterrent review. Proposals to reassess reliance on US Trident systems reflect growing anxiety about the durability of transatlantic guarantees in a more unpredictable security environment (here)
UK defence insiders warn Palantir’s data aggregation power creates sovereign risk. Officials argue that even if raw datasets remain state-owned, the company can derive strategic insights by combining metadata across departments, building a detailed intelligence picture of population & critical infrastructure (here)
Russia tests nationwide internet shutdown architecture with Moscow blackouts. Temporary mobile internet disruptions suggest the Kremlin is trialling scalable censorship & traffic-control systems, signalling a move toward Chinese-style sovereign internet governance models amid wartime information control (here)
Russia launched far-right “Paladins” network targeting Europe. Intelligence officials warn Moscow is backing extremist online cells designed to incite violence, destabilise democratic institutions & exploit migration tensions, reflecting a broader shift toward hybrid warfare using ideology, cyber influence & grassroots mobilisation (here)
TikTok investors may pay a ~$10B fee to the Trump administration as part of deal negotiations – a direct insertion of executive power into corporate transactions (here)
Likewise, Intel shareholders alleged the company offered strategic concessions to US authorities to avoid political pressure campaigns (here)
“Londonmaxxing” reflects The bull case for London - reasons to be cheerful about the city! (here)
Per the FT, debates around “sovereign AI” increasingly centre on the “economics of anti-scale.” Smaller nations are exploring domestically controlled compute & models despite efficiency disadvantages, prioritising resilience & political autonomy over hyperscale optimisation. “Deglobalisation — which is in effect what this is, however rational — is expensive for individual countries, but a windfall for their suppliers” (here)
Peter Thiel (founder of Palantir) commenced a series of closed door lectures in Rome depicting those who lobby for tech regulation as harbingers of the antichrist. Thiel’s views stand in stark contrast to those of Pope Leo, who has warned of the dangers of AI & called for stronger regulation to minimise its risks (here)
The UK is pledging £1B to accelerate domestic quantum computing capability. The funding package aims to support hardware scale-up, talent pipelines & sovereign supply chains as Britain seeks to remain competitive with US & Chinese quantum investments. Officials frame quantum not only as a scientific race but as a strategic computing capability tied to encryption resilience, advanced materials & defence modelling (here)
Pentagon expands oversight of Stars & Stripes media outlet. Proposed controls on editorial scope reflect growing tension between military comms strategy & journalistic independence during conflict periods, with implications for information warfare narratives (here)
Norway (via Norwegian Consumer Council) is pushing back against platform “enshittification”, advocating for open digital ecosystems & public-interest platforms to counter concentration among US tech giants, aligning with broader European digital sovereignty debates. This has been done via an absurdist video (see here) from the publicly funded Norwegian council which joined up with <70 groups/individuals across Europe & the US, including trade unions human rights orgs to urge policymakers to resist deliberate deterioration of platforms & devices (here)
Office.eu launches as sovereign European productivity suite. The Hague-based initiative aims to provide 100% European owned alternatives to Microsoft 365 & Google Workspace, reflecting mounting political demand for local data control & cloud independence (here)
Hua Hong (semiconductor manufacturer) has developed 7nm chipmaking capability for AI chips marking a significant step in China’s domestic semiconductor push as US export controls bite. Until now only SMIC could produce at this level at scale, but collaboration with Huawei & prototype testing for blacklisted designer Biren suggest China is steadily localising advanced AI compute supply chains despite sanctions (here)
Strategic Sectors
AI
Another article on AI deepfakes blurring reality in both warfare & financial markets as scams scale globally. Per the FT, real images of toxic smoke over Tehran now circulate alongside fabricated scenes of US soldiers captured by Iranian forces, illustrating how generative media is overwhelming verification capacity. A Columbia policy brief estimates AI-driven fraud losses rose from ~$12.3B in 2023 toward ~$40B projected by 2025, with criminals using synthetic celebrity endorsements, cloned executives & impersonated relatives to trigger urgent payments or crypto transfers. Regulators are attempting piecemeal responses – from SIM card limits in Malaysia & Mexico to bank data-sharing regimes in Singapore & new deepfake identity protections in Denmark & Ireland – yet enforcement remains fragmented as scams operate across borders. Scholars argue platforms distributing billions of fake adverts daily are the “cheapest cost avoiders”, but liability frameworks remain contested. The result is a widening trust deficit: consumers cannot realistically detect high-quality deepfakes themselves, turning misinformation resilience into a systemic governance & infrastructure challenge rather than a media literacy problem (here)
Andrej Karpathy’s autoresearch experiments cut GPT-2-level training time ~11% in 48 hours – demonstrating how automated research loops could accelerate model improvement cycles (here)
Nvidia is expanding beyond GPUs into full-stack AI strategy. A $2B investment in neocloud provider Nebius Group — helping finance multi-gigawatt data-centre build-outs & large hyperscaler deals — highlights efforts to secure downstream demand even as analysts warn about circular financing risks. At the same time, planned ~$26B spending on open-weight models signals ambitions to shape the application layer, while new inference-focused chips & forecasts of up to $1TN in cumulative AI chip revenue underscore confidence in deployment-driven growth (here, here & here)
GenAI traffic remains highly concentrated with ChatGPT & Gemini dominating global usage while Claude has recently accelerated share gains. SimilarWeb estimates show ChatGPT declining from ~75.7% share 12 mo ago to ~62%, while Gemini rose from ~5.7% to ~24%, benefiting from distribution via Android & Workspace. Claude climbed from ~1.7% to ~3.3% within months, but this share is likely to be materially higher now following Opus 4.5, Claude Code & Opus 4.6 launches which drove corporate adoption & doubled ARR since the start of the year (here)
Notable deal: Mirendil (neolab building specialised models for scientific R&D) is reportedly in talks to raise $175M at $1B co-led by Andreessen Horowitz & Kleiner Perkins. Mirendil was founded by former researchers from Anthropic, Google DeepMind, xAI & OpenAI (here)
Cybersecurity
Google closed its $32B acquisition of Wiz in the largest deal in its history! The deal delivered >$8B combined to founders & multibillion-dollar returns to major VC backers as cloud security consolidates (here)
The White House published its national cyber strategy (just 7 pages!) framing it as an instrument of American power - innovation over regulation & offense is no longer optional (here)
OpenAI acquired Promptfoo, an enterprise AI security & evaluation platform, reinforcing the strategic importance of testing, red-teaming & reliability tooling as agent deployments scale across regulated industries (here)
Iran-linked hackers disrupted Albania’s parliament systems in a data-wiping attack (here) & Stryker (Medical tech platform) suffered a global outage after a wiper malware attack claiming ~50TB of data exfiltration – both stories underscore how geopolitical cyber operations increasingly target commercial critical infrastructure (here)
North Korean “fake workers” are using AI-generated identities to infiltrate European companies & secure remote jobs. Intelligence agencies warn operatives are deploying deepfake video, synthetic CVs & automated coding outputs to pass hiring processes, then funnelling salaries or data back to state-linked networks. The campaign illustrates how genAI is reshaping economic espionage from network intrusion toward labour-market infiltration, turning HR pipelines into an unexpected security frontier (here)
The UK’s Companies House security lapse exposing UK director data (here) & Sweden’s e-government source code leak highlight the fragility of centralised digital state systems increasingly relied upon for welfare, tax & identity services (here)
Notable deals: Kai launched after raising $125M to scale its AI-driven autonomous security platform (here); Qevlar AI raised a $30M for its AI SOC (here); Escape raised €15.4M for its AI offensive security platform (here) & Augur raised $15M Seed to deploy sovereign AI security monitoring across critical infrastructure (here)
Defence
Poland plans to deploy the EU’s first integrated anti-drone shield combining radar, EW & interceptor systems reflecting how cheap unmanned aerial threats are driving rapid innovation in layered air defence across Europe (here)
European & Asian militaries are ‘scrambling’ to build sovereign Starlink-like satellite constellations to reduce dependency on the US. Starlink uses nearly 10k satellites to provide connectivity almost anywhere on Earth. But to avoid reliance on a privately, US controlled service, countries are developing alternatives such as the EU’s IRIS (planned ~300 satellites by 2030), China’s Guowang & Qianfan constellations (targeting up to 13k satellites) & Russia’s delayed Sfera network. These systems aim to ensure secure troop coordination, drone control & real-time intelligence links during high-intensity operations (here & here)
German Air Force is eyeing operational unmanned combat aircraft by 2029. Airbus proposals signal Europe’s move toward indigenous autonomous strike capability, driven by Ukraine lessons & pressure to reduce reliance on US defence platforms (here)
Open-source intelligence is being constrained as satellite imagery of conflict zones is delayed or withheld. Commercial providers such as Planet extended publication delays on Middle East imagery from 4 days to 2 weeks amid wartime pressure, reducing transparency just as demand for verification rises (here)
Per the FT, AI is compressing the tempo of modern warfare. In Iran, the US reportedly struck >2K targets in just 4 days (matching the number hit in the first 6mo of the anti-ISIS campaign) as frontier models moved from summarisation to operational reasoning. Systems such as Palantir’s Maven, now used by 20-50K personnel across dozens of military entities & adopted by NATO, act as a real-time “software brain” across the full kill chain: identifying targets, selecting weapons & assessing damage. Where approval cycles once took hours or days, AI-enabled workflows are shrinking decisions into seconds, allowing commanders to generate & vet vastly larger target sets. The result is unprecedented scale & speed — alongside growing questions about oversight, traceability & benefits vs risks of machine targeting (here)
A watchdog report claims the US Department of War spent millions on luxury food including $6.9M on lobster tail, $2M on Alaskan king crab & $15.1M on ribeye steak in just September 2025 (here)
US Army awarded Anduril a contract worth $20B (here)
Notable deal: Cambridge Aerospace (drone & missile interception) is speaking with investors about raising about $200M at <$1B (here)
Crypto
Mastercard is buying stablecoin startup BVNK for $1.8B (here)
EVs / AVs
Uber is rolling out Motional autonomous rides in Las Vegas with safety drivers initially present, reflecting gradual commercialisation of robotaxi networks ahead of wider regulatory clearance (here)
Zoox will integrate robotaxi rides into Uber’s Las Vegas app if federal approval is secured. The purpose-built autonomous vehicle lacks steering wheel or pedals & currently operates only under demonstration exemptions (here)
Wayve signed agreements with Nissan & Uber to launch robotaxis in Japan, reinforcing Asia as an early deployment arena for autonomy (here)
Honda scrapped several EV models after forecasting its first annual loss since 1957, while Rivian & Lucid push ~$50k vehicles, signalling a reset in legacy automaker electrification strategies (here)
Travis Kalanick is pivoting CloudKitchens into robotics startup Atoms – targeting automation across food, mining & logistics as part of his long-standing thesis on digitising the physical economy. Uber reportedly investing ~ $100M in its self-driving arm. The company reunites Kalanick with several former Uber operators, including Anthony Levandowski & ex-ATG head Eric Meyhofer, making Atoms both a comeback story & a revival of one of Silicon Valley’s most controversial autonomous-driving circles (here)
Robotics
China is accelerating humanoid robotics development through state-backed AI labs focused on industrial deployment. New facilities are training large multimodal models using real-world factory footage, motion-capture datasets & reinforcement learning environments to build robots capable of logistics, assembly & service tasks. The strategy mirrors Beijing’s EV playbook: scale first, iterate quickly & compress commercialisation timelines through subsidies, local procurement mandates & coordinated supply chains. Western robotics firms still lead in software stack sophistication, but China’s manufacturing density & data access will likely narrow the gap rapidly (here)
Notable deals: Rivian spinout Mind Robotics raised $500M to develop AI-powered industrial robots designed to perform physical manufacturing tasks such as picking parts, assembling components & manipulating wiring harnesses in factory environments. Led by Accel & Andreessen Horowitz, with participation from Rivian (here). Sunday (home robotics) reached ~$1.15B valuation (here)
Space
China & Russia are expanding influence across the global space sector through subsidised launches & diplomatic partnerships. Emerging economies are increasingly opting into lower-cost satellite & launch ecosystems tied to political alignment, raising the risk of a fragmented orbital infrastructure landscape (here)
AgTech / Bio
Gene-edited banana startup Tropic raised $105M to commercialise climate-resilient fruit varieties, reflecting growing venture interest in food security technologies as supply chains face climate & geopolitical shocks (here)