SuperReturn Berlin 2026 Debrief

Last week, the Brighton Park team attended SuperReturn Berlin for the fifth year running, but this conference felt fundamentally different. As we flew back from the conference, SpaceX executed the largest IPO in market history at a $1.8 trillion valuation, signaling that the discussions about AI and market restructuring are no longer hypothetical - they are happening in real time. The conference captured a market in transition: LPs reassessing portfolio positioning, GPs refining investment frameworks, and 10,000+ attendees working to understand a technology shift that's moving faster than any previous software cycle. Last year's cautious curiosity about AI has evolved into genuine enthusiasm about the scale of the opportunity ahead. If SuperReturn 2025 was about whether AI would prove transformative, 2026 was about who's positioned to win during the most significant value creation cycle in decades - and who's at risk. Below, we outline four key themes from our conversations with global investors around the conference:

1. AI: Understanding Where Value Accretes and Where Businesses Are Exposed

‍Investors have shifted from exploring AI's potential to dissecting the specific mechanisms of disruption vs value creation. Who are the winners and losers going to be, and how do we underwrite it?

• Portfolio defensibility: Investors are conducting comprehensive risk assessments of their existing investments: Can companies integrate AI capabilities into their core products, or will large language models make their value proposition obsolete? Are these businesses positioned at the orchestration layer - where workflows are managed and decisions are made - or are they simply storing information and vulnerable to disruption? The pattern is clear: software companies that create new categories, control critical decision points, embed themselves in workflows, and possess deep domain expertise appear defensible.
• ‍Valuation discipline: Investors have once again developed skepticism toward valuations disconnected from underlying economics. While capital has flooded toward high-profile foundation model companies and infrastructure plays, driving some valuations back up to 2021 highs, vertical AI investments have remained more grounded. This creates a bifurcated market: more horizontal, hyper-growth plays are bid up on potential and narrative, while vertical AI companies are being valued on demonstrated unit economics and customer acquisition fundamentals, making them more accessible entry points for growth equity investors.
• Recalibrating the underwrite: AI-native companies are reaching scale approximately three times faster than cloud-software businesses did, which means there's less historical data to analyze (Source). This creates real challenges: How do you assess growth durability with limited historical information? How do you model long-term profitability when costs and compute economics are rapidly changing? The frameworks developed over two decades of SaaS investing don't apply cleanly to AI infrastructure and applications. Firms are racing to build new frameworks before the knowledge gap becomes a competitive disadvantage.

The consensus: AI isn't an extinction event for software, but it is a value reallocation event. While legacy companies face real risk, AI-native and AI-enabled companies creating entirely new categories, automating / enabling workflows, and providing tangible ROI, represent exceptional investment opportunities with significant value creation opportunities.

2.“Tokenomics”: The Cost Structure Reality

‍While AI's strategic importance is no longer in question, its economics at scale remain unresolved. "Tokenomics" emerged as shorthand for a fundamental challenge: running AI inference at scale is prohibitively expensive for many applications, and current pricing models don't support sustainable economics. The evolving investment thesis:

• Applications using open-weight, domain-specific models show superior cost efficiency. Foundation models offer remarkable capabilities but carry significant compute costs. Models optimized for specific use cases can deliver comparable results while running on more economical infrastructure, fundamentally improving margins.
• ‍Owning the technology stack has become strategically important again. Companies that own their infrastructure - compute resources, storage, and model serving - can manage margin pressure as the market matures. Businesses dependent on third-party API calls face persistent margin challenges with limited ability to influence their cost structure.

Translation: Companies whose AI strategy consists primarily of routing requests via API to foundation model providers haven't articulated a defensible economic model - they've built a cost dependency without a margin advantage. The companies positioned to scale efficiently are those that either own their infrastructure stack or establish strategic partnerships with foundation model providers that enable meaningful cost optimization at scale.

‍3. The IPO Wave: Market-Redefining Public Offerings on the Horizon

The investment community is preparing for an unprecedented sequence of liquidity events, with SpaceX (now trading at a $2.1 trillion market cap), Anthropic, and OpenAI potentially going public within a 12 month window. The magnitude of capital that could return to the ecosystem in this compressed timeframe is difficult to put in historical context.

• Market absorption capacity: Can public markets efficiently absorb this volume of supply without fundamentally repricing expectations across the technology sector? The scale of these offerings - particularly given SpaceX's valuation - raises questions about timing, sequencing, and whether institutional demand can sustain multiple defining events in rapid succession without experiencing fatigue.
• ‍Liquidity decisions: Every limited partner with exposure is modeling the same questions: should distributions be held through post-IPO volatility, or should liquidity be realized immediately upon becoming available? This directly impacts distribution requirements, portfolio rebalancing timelines, and how quickly capital cycles back into private markets.

The bottom line: The market is actively trying to model what happens in the 24 months following these events. Currently, there is no consensus on how valuations, capital flows, or competitive dynamics will adjust.

4. The Liquidity Reality: Exit Paths Are Narrowing, and Strategies Are Adapting

The exit environment remains constrained, and traditional paths forward face increasing friction.

• Continuation vehicles have evolved from niche solutions to standard portfolio management tools. What was previously used as a specialized mechanism has become a normalized practice for liquidity. Limited partners expect significant expansion of the continuation vehicle market as general partners balance liquidity needs against the strategic value of maintaining ownership in high-performing assets.
• ‍Sponsor-to-sponsor transactions face new complexity. Every prospective sponsor is evaluating whether acquired software assets are built appropriately for the AI ecosystem or whether fundamental product rebuilds are required. This introduces uncertainty, drives up bid-ask spreads, and creates a bottleneck for sponsor-to-sponsor exit activity: if substantial reinvestment in product development is necessary, does the risk-adjusted return profile still support a transaction? Can the next sponsor credibly underwrite a path to returning 2-3x on invested capital given these technical requirements?
• Strategic acquisitions remain the most efficient liquidity path for differentiated assets. Corporate buyers continue to deploy capital, but their acquisition criteria have sharpened. They seek capabilities, not incremental features. Strategic acquirers will pay meaningful premiums for companies that solve mission-critical problems, possess proprietary intellectual property, or enable workflows that cannot be easily replicated.
• ‍The IPO bar continues to rise. SpaceX's $1.8 trillion valuation at IPO has effectively reset expectations for what constitutes appropriate public market scale. Public investors are looking for category-defining businesses with demonstrated paths to profitability, durable competitive advantages, and revenue growth that justifies premium valuations. Companies that would have gone public at $1-2 billion valuations in 2021 are staying private longer, supported by abundant late-stage private capital (e.g. Databricks, which recently raised a Series L round at a $134 billion valuation).
• What happens when there is no bid? Companies that cannot clear the bar for any of these exit paths face prolonged hold periods. The outcome becomes binary: discounted management buyouts that crystallize losses, or extended holds as zombie assets in portfolios for the next decade.

The takeaway: Liquidity remains available, but it is selective. Firms that can articulate clear exit strategies - and the strategic value that justifies them - operate with fundamentally different options than those dependent on improving market conditions.

The Bottom Line

The private markets landscape is being re-written in real time: AI is re-distributing where value lives within software, historic IPOs are about to reshape capital flows, and exit strategies are adapting as traditional paths narrow and new liquidity mechanisms emerge. We believe deep technology fluency, disciplined investment frameworks, and decisive action will increasingly differentiate outcomes as market structure continues to shift.