About SuperSeed Capital Limited

SuperSeed exists to back Europe's best B2B SaaS founders at the earliest stages and to help them build great companies. In the short term, our portfolio companies enable their customers to drive revenue growth and efficiency savings using next-generation software and AI. In the long-term, they have an opportunity to create category defining global technology companies. SuperSeed focuses on the fundamentals by helping founders build good companies with strong unit economics and sensible distribution models.

Forward-looking statements

This announcement contains statements that are or may be forward-looking statements. All statements other than statements of historical facts included in this announcement may be forward-looking statements, including statements that relate to the Company's future prospects, developments and strategies. The Company does not accept any responsibility for the accuracy or completeness of any information reported by the press or other media, nor the fairness or appropriateness of any forecasts, views or opinions express by the press or other media regarding the Group. The Company makes no representation as to the appropriateness, accuracy, completeness or reliability of any such information or publication.

Forward-looking statements are identified by their use of terms and phrases such as "believe", "targets", "expects", "aim", "anticipate", "projects", "would", "could", "envisage", "estimate", "intend", "may", "plan", "will" or the negative of those, variations or comparable expressions, including references to assumptions. The forward-looking statements in this announcement are based on current expectations and are subject to known and unknown risks and uncertainties that could cause actual results, performance and achievements to differ materially from any results, performance or achievements expressed or implied by such forward-looking statements. Factors that may cause actual results to differ materially from those expressed or implied by such forward looking statements include, but are not limited to, those described in the Risk Management Framework section of the Company's most recent Annual Report. These forward-looking statements are based on numerous assumptions regarding the present and future business strategies of the Group and the environment in which it is and will operate in the future. All subsequent oral or written forward-looking statements attributed to the Company or any persons acting on its behalf are expressly qualified in their entirety by the cautionary statement above. Each forward-looking statement speaks only as at the date of this announcement. Except as required by law, regulatory requirement, the Listing Rules and the Disclosure Guidance and Transparency Rules, neither the Company nor any other party intends to update or revise these forward-looking statements, whether as a result of new information, future events or otherwise.

Investment Manager's Review

Where Are We Right Now?

NVIDIA is down 10% over the past fortnight. Palantir has fallen 16%. The Magnificent Seven are wobbling. After a strong year, markets are taking profits and reassessing. Is this the AI bubble popping, or simply a risk-off period before year-end?

Two legitimate concerns are circulating, both worth examining carefully. The first centres on Chinese open-source models reaching parity with Western frontier labs-the top five open-source models globally are now all Chinese (DeepSeek, Kimi K2, Qwen 3, MiniMax M1, GLM-4.5), and these models are beginning to displace Western frontier labs in production workloads. The second involves GPU depreciation schedules. Michael Burry and others argue that hyperscalers are depreciating AI hardware over six years when technological obsolescence and power constraints force replacement far sooner, flattering current profitability.

Neither concern is trivial. Chinese models genuinely threaten Western frontier labs' pricing power. Anysphere (the company behind Cursor - the $29 billion AI coding platform), built its own proprietary "Composer" model (widely thought to be a fork of one Chinese open-source model Qwen3) rather than continuing to pay Anthropic's premium pricing. The displacement we predicted at year's start is materialising in production environments.

The depreciation question carries weight too. If hyperscalers must write down GPU assets faster than current schedules assume, profitability takes a hit. Markets would reprice accordingly. But even if depreciation periods shorten from six years to four, the impact shaves valuations rather than destroys them. This is accounting friction, not structural collapse.

Our assessment: risk-off sentiment, not catastrophe. Markets are pausing after strong gains, digesting genuine concerns about AI economics. The concerns are real - frontier lab pricing power is under pressure, hyperscaler profitability may prove overstated - but these are margin questions, not existential threats.

The more interesting question: where does value actually concentrate as these dynamics play out?

The Application Layer Thesis

Whilst frontier labs burn billions maintaining market position and hyperscalers commit $453 billion in 2025 capex (more than double 2024's ~$220 billion spend), a quieter shift is occurring. Applications are escaping the cost structure that made their economics impossible.

Twelve months ago, application companies like Cursor sent every revenue dollar to Anthropic for API access. Anthropic then spent 1.5x in compute costs. And the hyperscalers would invest 10-14x in capex. The unit economics were broken. Today, those same applications switch to open-source Chinese models at one-twentieth to one-fiftieth the cost. Economics invert overnight. Instead of burning money on intelligence, applications achieve positive unit economics. Instead of remaining captive to frontier lab pricing, they control their own margins.

This isn't displacement by slightly cheaper alternatives. It's fundamental margin restructuring enabled by open-source models that deliver comparable performance at radically lower cost. The irony is that Western hyperscalers' massive investment in frontier models-the hundreds of billions in infrastructure, the burn rates exceeding revenue-accelerated the Chinese response. Chinese labs learned efficiency matters more than scale.

The Implementation Bottleneck

But here's where the thesis gets interesting: whilst intelligence commoditises, implementation doesn't.

The evidence is stark. MIT's 2025 report on generative AI in business found that 95% of enterprise AI pilots fail to reach production. A separate study by S&P Global reveals that 42% of companies now abandon the majority of AI initiatives before production-a dramatic surge from just 17% the previous year. Overall, 88% of AI pilots never make it to production, and 80% of AI projects fail outright-twice the failure rate of non-AI technology projects.

The challenge isn't models. The top obstacles are data quality and readiness (43% of respondents), lack of technical maturity (43%), and skills shortage (35%). In other words: enterprises have access to powerful foundation models, but they lack the absorptive capacity-the infrastructure, frameworks, domain expertise, and organisational readiness-to deploy them successfully. Technology alone doesn't create value. Implementation does.

Physical AI: Implementation as Moat

Physical AI exemplifies this pattern. Teaching autonomous systems to function in unpredictable environments-warehouses, construction sites, industrial facilities-requires more than models. It requires domain expertise in how these environments actually work, regulatory validation that can't be shortcut, iterative learning in physical settings where you can't A/B test at scale, and customer-specific deployment because every facility differs.

The foundation model becomes commodity infrastructure-necessary but insufficient. Defensible value lives in the application layer where software meets physics. This is why European robotics company Hive Autonomy secured a strategic partnership with Toyota Material Handling in Q3, beating competitors from the US. They're not competing on model parameters. They're solving implementation: making autonomous systems work where traditional automation fails.

When Toyota Material Handling-the world's number one forklift supplier for over two decades-chooses a European startup over in-house development or Silicon Valley alternatives, they're choosing implementation expertise over pure-play AI model capabilities. The model is commodity; the application is the moat.

Fund Portfolio Update

The portfolio revenue has grown 45% year-over year. Performance tracks top quartile benchmarks globally. More significantly, six portfolio companies enter Series A fundraising over the next two quarters, which will drive substantial TVPI expansion in the quarters to come as valuations reflect progress since Seed rounds.

Key Q3 Developments

Growth leaders this quarter include Popp (213% annualised growth) and FreightSuite (152% annualised growth). Both demonstrate the pattern we're backing: vertical AI applications solving specific problems in established industries where domain expertise creates moats.

The Series A pipeline accelerates. Popp, ThingTrax, OctaiPipe, Finteum, FreightSuite, and Hive progress toward raises over the next two quarters. These valuations will reflect the substantial progress companies have made since our Seed investments.

Portfolio Spotlight: Hive Autonomy

The Company

Founded in 2020, Hive provides autonomous and remote-controlled load handling for forklifts, wheel loaders, and construction equipment. Machine-agnostic retrofit solution enabling single operators to control multiple machines simultaneously from centralised stations-toggling seamlessly between autonomous operation, remote assistance, and full remote takeover when required.

This isn't simply remote control. It's the critical bridge between one-operator-per-machine and full autonomy-precisely what's required in environments where autonomous operation can't be deployed overnight.

Q3 Breakthrough

Hive secured strategic partnerships with Toyota Material Handling (plug-and-play automation for Toyota forklifts), with more vehicle manufacturers on the way. Unlike AGVs requiring controlled environments and fixed paths, Hive's AI operates in unpredictable real-world conditions: active warehouses, construction sites, industrial facilities.

Toyota chose Hive because the platform enables forklifts to operate autonomously where traditional automation fails. Hive solves implementation, not just technology.

Why It Matters

Hive exemplifies where defensible value lives. They solve the genuinely hard problem: making autonomous systems function in environments too variable for scripted automation.

Every warehouse differs. Every construction site presents unique challenges. Every industrial facility has distinct constraints. Hive's expertise isn't in the model-it's making autonomy work where traditional automation breaks down.

When global industrial leaders choose a European startup over in-house development or Silicon Valley alternatives, they're choosing implementation expertise. The model is commodity; the application is the moat.

Having just closed their seed round, Series A is expected in the first half of 2026, with strong strategic interest recognising Physical AI applications as where lasting value concentrates.

Forward Outlook

Investment Period

SuperSeed II's investment period concludes Q1 2026. We anticipate two final investments in Q4 2025, both in Physical AI applications where domain expertise creates defensible moats.

Market Positioning

While the current market wobbles are resting on legitimate concerns (Chinese open source models compressing frontier lab pricing power, and potential GPU depreciation impacts on hyperscaler profitability), neither fundamentally threatens the application layer thesis.

Validating The SuperSeed Thesis

In fact, both dynamics validate our positioning. As intelligence commoditises through open source models, applications benefit from radically lower costs whilst maintaining defensible positions through implementation expertise. As hyperscalers face margin pressure, they increasingly recognise value in application-layer optimisation.

Physical AI-where software meets atoms-doesn't follow foundation model economics. Domain expertise in manufacturing, logistics, construction, and industrial automation creates moats that Chinese open source can't replicate. Europe's manufacturing base transforms from perceived liability to genuine advantage. Companies bridging bits and atoms demonstrate where lasting value concentrates.

As the portfolio matures and valuations reset at Series A rounds, metrics will reflect the progress companies have made-not just in revenue growth, but in building defensible positions within industries undergoing fundamental transformation.

The AI revolution continues. We're positioned where value concentrates rather than where capital burns.