Introduction

AI is no longer theory - it’s shaping business, life, and entire industries. The rules are still unclear, but the race is on.

What’s the problem it solves?

Startups and investors struggle to separate hype from real, lasting value in AI. This report maps out where AI is stable, where it’s chaotic, and what founders should do to build companies that last.

Quick Summary

The report compares AI today to a new universe forming. After the “Big Bang” of ChatGPT in 2023, we’re now in the “First Light” stage where clusters of stable companies and clear benchmarks are starting to emerge. But it’s still messy, with low margins, intense competition, and “dark matter” - unknowns like memory, context, and governance.

Two kinds of startups stand out:

• Supernovas: Explosive growth to $100M ARR in 1-2 years, but fragile and risky.
• Shooting Stars: Grow like SaaS but faster, with solid retention, healthier margins, and a better long-term path.

Beyond benchmarks, AI is reshaping software itself. Old “systems of record” like Salesforce and Oracle are being replaced by “systems of action” that don’t just store data but act on it. Entire verticals once seen as “too old-school” (healthcare, legal, education, real estate) are rapidly adopting AI-native tools that deliver instant ROI.

Consumer AI is also shifting. General assistants like ChatGPT and Gemini dominate, but specialized tools for therapy, wellness, and creative work are gaining traction. The next major social platform could even come from AI-native interaction.

Looking forward, the authors predict five big shifts: the rise of AI browsers, generative video’s breakout in 2026, evaluation and data lineage as critical infrastructure, the possible birth of an AI-native social media giant, and a wave of M&A as incumbents buy their way back into the game.

Key Takeaways

• Superfast growth is possible, but lasting success looks more like Shooting Stars than fragile Supernovas.
• Memory and context will be the real competitive advantage in AI products.
• AI-native apps are moving from record-keeping to taking real action.
• Narrow, high-value wedges beat broad “AI for everything” launches.
• The browser may become the main stage for AI agents.
• Private, continuous evaluation beats public benchmarks.
• Vertical AI is becoming the new SaaS.
• Big incumbents will fight back through acquisitions.
• Human judgment and taste remain key to building winning AI companies.

What to do

• Don’t chase vanity growth-aim for durable “Shooting Star” success.
• Build products with memory and personalization baked in.
• Replace workflows, don’t just add AI features to old systems.
• Start with a clear wedge problem and deliver 10x value fast.
• Design for the browser as a core interface.
• Bake evals and data lineage into your product from day one.
• Look for opportunities in verticals that were slow to digitize.
• Be M&A-ready, but operate like you’ll lead the category.
• Use speed, taste, and judgment as your edge-technology alone isn’t enough.

‍