Top 10 UK-Based Tech Startups to Watch in 2026

The UK startup ecosystem is no longer a polite satellite of Silicon Valley. These ten companies are working on things that matter globally, from chips designed in Oxford to air defence systems built in Cambridge and orbital intelligence operated from London.

UK startups raised $23.6 billion in venture capital during 2025, up 35% from 2024, with AI companies taking 33% of that total. London ranked third globally in Startup Genome’s 2025 ecosystem report. The money is moving toward companies with defensible technical positions and genuine market proof, not toward thin applications layered on top of other companies’ models. The ten companies below reflect where that selectivity is landing.

The same warning applies here as with any startup watch list: funding tells you that credible investors made a considered bet. It does not tell you the company will succeed. Several companies below carry open critical uncertainties. This article names them rather than pretending they do not exist.

1. Ineffable Intelligence

Founded in late 2025 by David Silver, Professor of Computer Science at University College London and former head of Reinforcement Learning at Google DeepMind, Ineffable Intelligence is headquartered in London. In April 2026, the company raised $1.1 billion in what it described as the largest seed round in European history, valuing the company at $5.1 billion. The round was co-led by Sequoia Capital and Lightspeed, with participation from NVIDIA, Google, DST Global, Index Ventures, EQT, BOND Capital, the UK Sovereign AI Fund, and the British Business Bank.

The size and composition of the round reflects Silver’s specific reputation: AlphaGo, AlphaZero, AlphaProof. These are not adjacent credentials. They are the most influential results in reinforcement learning over the past decade, and they are Silver’s work.

What Ineffable is building is described as a superlearner, an AI system designed to acquire knowledge through direct experience and environmental interaction rather than through exposure to human-generated data. The distinction matters technically. Current large language models learn by imitating patterns in human text. The reinforcement learning approach, taken to its limit, would allow a system to discover knowledge that has never been written down: new mathematics, new materials, new scientific structure.

This is where honest framing must be applied most carefully. The ambition of building a superlearner is significant, and Silver is one of the very few researchers whose track record makes it credible to attempt. But the gap between the seed round and the first system that meaningfully exceeds existing capabilities is unknowable in duration and enormously difficult to bridge.

Ineffable’s founding team beyond Silver has not been publicly named, which makes an assessment of execution capacity difficult. The UK government’s backing reflects a strategic bet on keeping frontier AI research on British soil. Whether that bet pays off in science rather than just credibility will take years to determine.

Founded: 2025 | HQ: London | Founder: David Silver | Latest round: $1.1B Seed (April 2026) | Valuation: $5.1B

2. Nscale

Founded in May 2024 by Joshua Payne and Nathan Townsend, Nscale is headquartered in London and began as a spinout from Arkon Energy, an Australian Bitcoin mining operation. Payne concluded that AI compute infrastructure was a better use of the power procurement, site development, and hardware deployment skills that crypto mining had required. As of March 2026, Nscale has raised more than $3.7 billion in total funding at a $14.6 billion valuation.

The $2 billion Series C was co-led by Aker ASA and 8090 Industries, with participation from NVIDIA, Dell Technologies, Nokia, Citadel, Jane Street, Lenovo, and Point72. Sheryl Sandberg and Nick Clegg joined the board of directors in March 2026.

The commercial evidence behind the valuation is real. Nscale signed a $14.6 billion deal with Microsoft to deploy 200,000 NVIDIA GB300 accelerators across sites in Texas, the UK, and Portugal. The platform delivers vertically integrated infrastructure: GPU compute, networking, storage, orchestration software, and a managed Kubernetes service from a single provider, with all capacity marketed as carbon-neutral and located within the European Economic Area for GDPR compliance. A 60 MW site in Glomfjord, Norway, powered by hydroelectricity, began operations in early 2025. The UK site, scheduled for Q4 2026, can house up to 45,000 NVIDIA GB200 GPUs at 90 MW.

The honest tension in Nscale’s story is the relationship between capex commitment and revenue stability. Infrastructure businesses at this scale carry enormous fixed costs in land, power agreements, hardware, and financing. The IPO Payne has targeted for late 2026 is the moment when the market will price the gap between contracted revenue and the full cost of the build-out. Until then, the company’s health depends on executing the Microsoft deployment on schedule and closing additional capacity deals to fill its 1.3 GW global pipeline.

Founded: 2024 | HQ: London | Founders: Joshua Payne, Nathan Townsend | Latest round: $2B Series C (March 2026) | Valuation: $14.6B

3. CuspAI

Founded in 2024 by Dr. Chad Edwards (CEO) and Professor Max Welling (CTO), CuspAI is headquartered in Cambridge with offices in London, Singapore, Amsterdam, Berlin, and Tokyo. In July 2026, the company raised $450 million in a Series B led by Kleiner Perkins and NEA, with participation from Bezos Expeditions, AMD Ventures, the UK Sovereign AI Venture Fund, Temasek, Prosus, Glade Brook Capital Partners, and Lux Capital.

The round valued CuspAI at $2.6 billion, up from $520 million at the Series A nine months earlier. Total funding now exceeds $650 million. The advisory board includes Nobel laureate Geoffrey Hinton, former ASML President and CTO Martin van den Brink, and Abhi Talwalkar, AMD Board Member and Chairman of Lam Research.

CuspAI describes itself as a search engine for the material world. The underlying approach applies generative AI to materials science: modelling how proposed new materials would perform chemically and physically before any physical synthesis takes place, filtering a search space of trillions of possible molecular and structural combinations down to the candidates most likely to behave as intended.

The platform is synthesis-aware, prioritising candidates that can actually be manufactured under real conditions rather than generating theoretically interesting but physically unrealisable structures.

Alongside the Series B, CuspAI launched the AI Materials Foundry, a coalition of more than 48 organisations including NVIDIA, Meta, Samsung, Hyundai Motor Group, and Lam Research. Semiconductors will absorb 80% of the company’s research bandwidth in 2026, specifically work aimed at finding materials that could reduce reliance on supply-constrained rare metals such as ruthenium and iridium. Sifted ranked CuspAI first on its AI 100 list of rising European AI companies.

The question a materials discovery company must eventually answer is whether AI-identified candidates can survive physical synthesis, safety testing, and manufacturing qualification, a process that takes years even when the initial discovery is sound. CuspAI’s closed-loop validation with physical laboratory partners is the design decision that separates credible materials AI from theoretical performance claims.

Founded: 2024 | HQ: Cambridge, UK | Founders: Dr. Chad Edwards, Professor Max Welling | Latest round: $450M Series B (July 2026) | Valuation: $2.6B

4. Olix

Founded in 2024 in London by James Dacombe, Olix is a chip company building inference hardware that departs from the dominant HBM-based architecture. In August 2026, the company raised a $312 million Series B led by Fundomo, with participation from Arm, Hudson River Trading, Reed Hastings (Netflix co-founder), and the UK Sovereign AI Venture Fund. The round valued Olix at $3.3 billion, more than tripling its $1 billion valuation from a $220 million raise in February 2026. Total funding now exceeds $530 million. Professor Nick McKeown, Professor Emeritus of Computer Science and Electrical Engineering at Stanford and winner of the 2025 Marconi Prize, joined the board alongside the round.

The technical argument Olix makes is architectural. Standard AI inference hardware keeps model weights in high-bandwidth memory, creating a bandwidth bottleneck as model sizes grow. Olix integrates SRAM-architecture with photonics, placing more compute directly alongside memory and using light rather than electrical signals to move data between components. The company argues this surpasses HBM-based architectures on throughput per megawatt and total cost of ownership, while also being better insulated from supply chain constraints that currently affect HBM, dominated by Samsung and SK Hynix.

First customer delivery is targeted for H2 2027. That timeline is the most important single datapoint for anyone evaluating Olix: the chips do not yet exist in customer hands. Semiconductor startups routinely face delays between investor milestones and silicon milestones. Dacombe is 25 years old and has no prior semiconductor company founding experience, which makes the quality of his engineering team the primary indicator of execution credibility. Arm’s participation in the round is a meaningful technical signal: Arm invested in the architecture, not just the narrative.

Founded: 2024 | HQ: London | Founder: James Dacombe | Latest round: $312M Series B (August 2026) | Valuation: $3.3B

5. Cambridge Aerospace

Founded in September 2024 by Steven Barrett (CEO), Christopher Sylvan (CCO), Junaid Hussain, and Tess Van Stekelenburg, Cambridge Aerospace is a Cambridge-based defence technology company building affordable interceptor systems for drone and missile threats. In August 2026, the company raised $300 million in a Series C led by DFJ Growth, valuing it at $3.4 billion. Prior investors include Lux Capital, Accel, Lakestar, Elad Gil, and Spark Capital. Grant Shapps, the former UK Defence Secretary, served as chair until May 2026.

The commercial traction is where Cambridge Aerospace’s story is most concrete. In April 2026, the UK Ministry of Defence contracted its Skyhammer system, a low-cost modular interceptor designed to neutralise Shahed-type attack drones. Skyhammer is already in the market. A rocket-powered interceptor called Starhammer, designed to handle higher-value missile threats, is in development targeting 2027 release. Two further products have been announced without confirmed delivery dates: Looking Glass, a multistatic radar system, and Nightstar, sovereign-built solid rocket motors.

Barrett was an aeronautics professor at MIT before joining the University of Cambridge in April 2024, and his prior published research on electrically propelled aircraft established his technical standing before the company launched. Sylvan previously led UK business development at Anduril, giving the commercial side defence industry experience from inside an operation that had already navigated the same market.

The demand driver, the proliferation of low-cost attack drones in modern warfare, is not speculative. The question is whether a two-year-old company can scale manufacturing and extend its product range at the pace a $3.4 billion valuation implies.

Founded: 2024 | HQ: Cambridge, UK | Founders: Steven Barrett, Christopher Sylvan, Junaid Hussain, Tess Van Stekelenburg | Latest round: $300M Series C (August 2026) | Valuation: $3.4B

6. Fractile

Founded in 2022 by Dr. Walter Goodwin, then a PhD student at the University of Oxford’s Robotics Institute, Fractile is headquartered in London with engineering presence in Bristol. In May 2026, the company raised $220 million in a Series B led by Accel, Factorial Funds, and Peter Thiel’s Founders Fund, with participation from Conviction, Gigascale, Felicis, Buckley Ventures, and 8VC. The round valued Fractile at approximately $1 billion. Pat Gelsinger, former Intel CEO, and Stan Boland, former Arm and Acorn Computers executive, are both early investors. In February 2026, Fractile announced plans to invest £100 million in its UK operations over three years.

Fractile builds inference chips using in-memory compute: an architecture that places processing directly within memory rather than between separate compute and memory units. The memory bandwidth bottleneck has become the primary constraint on LLM inference hardware speed as models generate longer outputs at scale. Fractile’s approach eliminates that data transfer at the architectural level. The company claims its chips can support inference for models with up to 16 trillion parameters.

Chips do not exist in customer hands yet. No UK or European customer pilots have been publicly announced, which is the datapoint most worth watching through the rest of 2026 and into 2027. Goodwin’s founding at 22 as an Oxford PhD student is an unusual starting position for a semiconductor company. It is also the starting position of the founders of Arm, which is not a trivial point. The investor list, particularly Founders Fund and Accel, reflects conviction that the architecture is credible.

Founded: 2022 | HQ: London and Bristol | Founder: Dr. Walter Goodwin | Latest round: $220M Series B (May 2026) | Valuation: approximately $1B

7. Granola

Founded in 2023 by Christopher Pedregal (CEO) and Sam Stephenson, Granola is headquartered in London. In March 2026, the company raised $125 million in a Series C led by Danny Rimer at Index Ventures, with participation from Mamoon Hamid at Kleiner Perkins, Lightspeed, Spark Capital, and NFDG. The round valued Granola at $1.5 billion, up from $250 million as recently as May 2025. Total funding stands at $192 million. Revenue grew 250% in the quarter preceding the Series C.

Granola’s product is an AI meeting notepad that records audio locally on the user’s device rather than joining a call as a visible bot. The distinction from competitors is behavioural rather than purely technical. Professionals in legal, executive, sales, and similar functions find meeting bots intrusive in ways that affect the authenticity of conversations.

Granola operates invisibly, transcribes the session, generates structured notes, and makes those notes searchable across an organisation. In February 2026, Granola launched a Model Context Protocol server, enabling AI tools to connect directly to Granola’s data layer. Enterprise customers include Vanta, Gusto, Asana, and Mistral AI.

Revenue, user count, and retention figures are not published. The AI meeting assistant market is projected to grow from roughly $3.5 billion in 2025 to more than $34 billion by 2035, a category large enough to support multiple large companies. What Granola has that most entrants do not is a specific user population (professionals who reject visible bots) that has proved large enough to drive 250% quarterly revenue growth. Granola’s durability rests on whether its local-recording, privacy-forward architecture creates a switching cost that enterprise customers would rather pay to preserve than abandon.

Founded: 2023 | HQ: London | Founders: Christopher Pedregal, Sam Stephenson | Latest round: $125M Series C (March 2026) | Valuation: $1.5B

8. Conduct

Headquartered in London, Conduct describes itself as an AI operating system for enterprise software, starting with SAP. The company has raised $60 million in a Series A. The founders have not been widely profiled in UK tech press at the time of writing.

The problem Conduct addresses is concrete, expensive, and structurally underserved. SAP systems run the operational core of approximately 77% of the world’s transaction revenue. They are also among the most expensive and difficult enterprise systems to modify, migrate, or fully understand. Large SAP landscapes accumulate decades of custom code, documented in ways that reflect the knowledge of the engineers who wrote it rather than the needs of the organisations that now depend on it.

When a company needs to change how its SAP system works, whether to respond to regulatory requirements, add new products, or migrate to SAP’s S/4HANA cloud architecture, the first obstacle is often that no one can fully map what the existing system does and why. Conduct reads the underlying code, reveals dependencies and downstream effects, generates documentation automatically, and accelerates both day-to-day operations and large-scale transformation projects.

The company claims 83% faster S/4HANA migration planning, 80% less manual system analysis, and 50% faster delivery of new features across its documented case studies. These figures would need independent verification to treat as benchmarks, but the directionality is consistent with what enterprise technology consultants describe as the primary friction in SAP modernisation projects. The addressable market is enormous: tens of thousands of organisations need to move from legacy SAP to S/4HANA before the mainstream maintenance deadline, and the bottleneck is not willingness but the complexity of knowing what they currently have.

HQ: London | Latest round: $60M Series A

9. Dwelly

Founded in 2024 by founders with backgrounds at Uber and Gett, Dwelly is headquartered in London. In July 2026, the company raised $170 million in a Series B led by EQT Growth, with participation from General Catalyst, s16vc, Begin Capital, DVC, and a cohort of founders including the CEOs of Legora, Synthesia, and ElevenLabs, alongside KKR partner Philipp Freise. The round included $95 million in equity and a $75 million debt facility led by Trinity Capital. A February 2026 Series A from General Catalyst at $69 million preceded the Series B. Total funding including debt now approaches $240 million.

Dwelly is a rollup of UK letting agencies, but the rollup is a means rather than the strategy. The company acquires independent rental agencies, integrates them onto its AI operating system, and uses automation to reduce the manual overhead that makes small lettings agencies economically fragile at scale. So far Dwelly has acquired eight agencies and manages over £200 million in gross merchandise volume.

The AI systems automate the routine work of letting: maintenance coordination, rent collection, communications between landlords, tenants, and contractors, compliance documentation, and inspection scheduling. Human agents remain responsible for property inspections, landlord relationships, and judgment calls that require local knowledge.

The UK private rental sector covers approximately 5.5 million properties managed by roughly 20,000 agencies, the majority of which are small independent businesses without the capital to invest in technology or the scale to benefit from it. General Catalyst’s managing director Marc Bhargava described Dwelly as having reached top-15 scale in under two years while improving outcomes across the board. The risk in any rollup model is integration speed outpacing operational quality.

The debt component of the Series B, used to fund acquisitions rather than diluting equity, accelerates the rollup pace. Whether the acquisition pipeline can maintain quality as it scales is the question that no funding round can answer in advance.

Founded: 2024 | HQ: London | Latest round: $170M Series B (July 2026, equity and debt)

10. Spaceflux

Founded in 2022 by Dr. Marco Rocchetto (CEO), Dr. Ingo Waldmann, and Marcel Debczynski, Spaceflux is headquartered in London. As of May 2026, the company has raised approximately £9 million in total seed funding, the most recent being a £3.5 million extension led by Blackfinch Ventures with participation from Foresight Group, the UK Innovation and Science Seed Fund, Future Planet Capital, and SPARX Asset Management.

The funding total is modest relative to every other company on this list. Spaceflux earns its place here on a different kind of evidence. In November 2025, Spaceflux was awarded three major UK government contracts to deliver sovereign space surveillance and tracking services to the National Space Operations Centre and the UK Ministry of Defence. It won them on competitive tender.

The company operates a network of optical telescopes across five continents, combined with an AI analytics platform called Cortex that processes telescope data into actionable orbital intelligence. The sensors combine visible and short-wave infrared imaging, enabling tracking during daylight hours, which extends observation coverage beyond the night-time window that limits most optical surveillance systems. By 2026, the network was expanding toward 25 global sites.

The UK Space Agency estimates 140 million pieces of debris in Earth’s orbit. A one-week disruption to satellite navigation would cost the UK economy an estimated £7.6 billion. Collision prediction and avoidance require continuous tracking data that neither military radar alone nor existing commercial satellite-tracking services provide at the resolution the growing orbital population demands.

Spaceflux’s government contracts represent operational deployment, not a research collaboration. Three production contracts won on competitive tender at seed stage is a more meaningful commercial signal than a large private round at an uncertain valuation.

Founded: 2022 | HQ: London | Founders: Dr. Marco Rocchetto, Dr. Ingo Waldmann, Marcel Debczynski | Latest round: £9M total seed (May 2026 extension)

Why These UK Startups Could Matter Beyond 2026?

Reading across this list, a structural pattern emerges that is specific to the UK and that differs from the patterns visible in the comparable US list. Several of the most interesting companies here are not competing in spaces where US giants are already dominant.

They are working on problems where the UK has genuine and specific institutional advantages: the proximity of Cambridge and Oxford to commercial spinout ecosystems, the DeepMind alumni network seeding multiple companies on this list, the UK Ministry of Defence and UK Space Agency as early-stage customers willing to grant government contracts to companies that are not yet at the scale US defence contractors require, and the University College London reinforcement learning research lineage that extends directly to Ineffable Intelligence.

CuspAI sits at Cambridge. Fractile grew from Oxford. Ineffable Intelligence is built on UCL. Cambridge Aerospace is explicitly named after its founder’s institutional affiliation. The UK’s research universities are functioning as genuine commercial founders in ways that go beyond licensing and spinout structures. The researchers are building the companies themselves.

Nscale, Olix, and Fractile together represent the UK’s determination to build its own AI infrastructure rather than exclusively consuming infrastructure built elsewhere. All three raised capital from the UK Sovereign AI Venture Fund alongside private investors, reflecting a government strategy that is more focused than previous industrial policy efforts: back the infrastructure layer with public capital, and make the sovereignty argument to European enterprise customers increasingly concerned about data residency and supply chain concentration.

If any of those three companies ships working hardware on announced timelines, the UK will have a credible domestic AI compute position for the first time.

Dwelly and Conduct are both tackling entrenched, highly fragmented markets that technology companies have repeatedly tried and failed to transform. Lettings agency rollups and SAP modernisation are not glamorous problems. They are enormous, sticky, economically critical, and resistant to change precisely because incumbents have low fixed costs and high customer inertia.

Companies that establish AI operating systems for these markets before the large platforms reach adequate depth in the same use cases carry durable commercial positions. Whether Dwelly and Conduct are those companies depends on execution quality over the next 24 months.

Spaceflux is the company on this list with the smallest funding and the most certain commercial footing. Government contracts require security clearances, technical audits, and competitive tender processes that private market customers do not impose. Winning three at seed stage means something specific about the quality of what Spaceflux has built. In an orbital environment becoming more contested every year, that foundation could become significantly larger than the funding history currently suggests.

The honest final observation is that the UK startup market in 2026 is doing something observers predicted would not happen without fundamental reform of exit pathways and public market liquidity. Companies like Nscale are contemplating IPOs. Granola went from seed to unicorn in three years. Cambridge Aerospace grew to a $3.4 billion valuation in 22 months.

The institutional infrastructure for building large companies in the UK has improved, the talent from DeepMind, Wayve, Revolut, and the broader London ecosystem has begun recycling into new companies, and the Sovereign AI Fund is providing a government backstop that makes the UK competitive for specific categories of deep technical founding teams. None of that guarantees any individual company on this list will survive the decade. It does suggest the UK ecosystem is producing companies worth taking seriously beyond its borders in a way that was not reliably true five years ago.