Startup brief
UK Sovereign AI and Grid Software Surge as Fintech Venture Capital Plunges
State-backed equity investments and energy infrastructure startups drive deal activity while traditional late-stage funding drops to a decade low.
Startup
The UK Sovereign AI Fund makes its first equity bet — backing London deeptech to lower enterprise compute costs across domestic hardware clusters.
BackgroundFounded by Cambridge neuroscientists, Callosum develops software that routes AI workloads across mixed computing chips to maximize performance. The UK Sovereign AI Fund was created in early 2026 to back domestic deeptech infrastructure directly.
- Atomico led the $100M seed round alongside Plural, DCVC, and state backing, funding expansion across European enterprise networks.
- Callosum's orchestration software optimizes chip workloads across mixed GPU and NPU hardware, reducing server compute overhead by up to 70%.
- South Korean chipmaker Rebellions agreed to supply 100 NPU racks running Callosum's platform, expanding UK government-backed AI cluster capacity.
Startup
Late-stage UK fintech capital dries up — forcing firms into consolidations while AI-native financial software absorbs available funding.
BackgroundBritain remains Europe's primary financial technology hub, but high interest rates have compressed valuations for late-stage private companies. Investors are requiring clear paths to net profitability rather than prioritizing rapid user acquisition.
- Late-stage growth funding dropped 45% to $830 million, pushing mature scale-ups toward M&A exits rather than public listing attempts.
- AI-driven financial software captured 25% of all UK fintech capital, rising from 16% in H1 2025 as traditional models struggled.
- Britain retained its position as Europe's largest fintech destination despite overall capital contractions, outstripping continental rivals in early deal counts.
Startup
Emerald AI proves data centers can dynamically shed power — giving UK grid operators a crucial tool to prevent localized blackout risks.
BackgroundData centers draw massive electrical loads that threaten power grid stability during peak consumption hours. Dynamic software throttling allows operators to drop power consumption briefly without disrupting active high-density computing tasks.
- Energize Capital and DCVC co-led the $150 million Series A alongside corporate investors including Nvidia, Siemens, and Salesforce Ventures.
- The London trial showed Emerald's software can cut data center GPU power draw by 25% to 40% during local transmission spikes.
- Dynamic throttling prevents localized blackout risks while preserving active high-density model execution, unlocking new regional data center capacity.
Startup
Systemiq backs eComID to solve retail's costliest headache — deploying AI recommendations to cut expensive e-commerce return rates.
BackgroundOnline apparel retailers lose billions each year due to high customer return rates and poor fit conversion. Machine learning models analyze cross-merchant sizing and return histories to deliver accurate purchase recommendations.
- London-based Systemiq Capital led the seed investment alongside H&M Group Ventures and former Depop chief executive Maria Raga.
- eComID's Vera AI agent delivers personalized fit recommendations across 20 million active shoppers on partner fashion sites.
- Cross-merchant sizing data reduced clothing returns by 30% while boosting overall checkout conversion rates by 10% during retail trials.
Unlock the full brief
Sign in to read every signal, takeaway, and source. Free account — Apple, Google, or email.
