CB Insights

Anthropic. Databricks. Glean. Fabrica Ventures bought into all three in the secondary market — without ever seeing a P&L. Here's how predictive intelligence turns external signals into multimillion-dollar conviction: Fabrica doesn't analyze thousands of companies. They set tight criteria upfront — Series C and beyond, backed by strong funds, showing recurring revenues. That gets them to about 150 companies. Then they apply their quality floor: a Mosaic Score of 800 or above (the top percentile of private companies). What's left are the high-conviction targets worth building a thesis around. Every investment decision starts as a 70-page document built entirely from external signals. No data room. No financial statements. Just predictive intelligence — revenue signals, customer sentiment, leadership moves, market positioning, investor history, and commercial maturity scores. It's how they construct conviction from the outside in. As Co-founder Alvaro Filpo puts it: "We're a small fund. We don't have the luxury of analyzing thousands of companies — we have to get it right." This is what "gray mass density" looks like — capital clustering where top founders, products, and investors converge. When you can predict which private companies will become generational winners, you don't need access to the data room. You just need better signals than everyone else. Read the full case study: https://lnkd.in/ghWe5s3e

After analyzing 150,000 companies over two years, we found that AI + data are 4.7x more predictive of startup success than being funded by top-decile VCs. CB Insights AI + data, that is :) For years, investors and biz dev teams used “smart money” funding as a proxy for potential. But that’s a backward-looking signal -- one that misses great companies and hides red flags. Our Mosaic Score flips the script. It applies data and AI to track every funded startup daily across the growth, financial, management, and industry signals most predictive of success. Our customers use it to augment their own judgment and data, spotting opportunities and risks earlier -- including the ones they might have otherwise missed. But the shift underway is bigger than any single model. Private market sourcing is moving from reactive to predictive. From who you know to what the data shows.

AI companies represent 10% of all startups… and 36% of top startups. The higher the success probability, the more AI dominates. AI companies are over-indexed at every tier: ↳ Top quartile: 16.3% AI (+26% since 2023) ↳ Top decile: 26.1% AI (+31% since 2023) ↳ Top percentile: 36.2% AI (+49% since 2023) AI's dominance amongst top startups is growing... fast. What's driving this? This isn't just about funding hype (although AI is dominating there too). Mosaic measures the fundamentals that actually predict a company's success – financial strength, team quality, market health, and momentum. The best AI companies are attracting elite capital, recruiting world-class technical teams, and most importantly, forming the business relationships that prove market traction. The companies winning now are coupling AI capabilities with enterprise distribution and real GTM execution. Track the AI companies dominating startup success for *free* with CB Insights' AI signal tracker: https://lnkd.in/d5HbT7GX

The future belongs to those who act on signals, not press releases. At our Money20/20 panel "Private is the New Public," CB Insights CEO Manlio Carrelli was joined by Nasdaq (Garrick S.), Forge (Vidya E.), Itaú Unibanco (Mauricio Correa), and Tiger Global (Sara Eadie) to discuss how predictive intelligence is turning private-market noise into clarity. For years, private markets were the blind spot of finance. Now, AI is decoding the patterns others miss, revealing tomorrow’s trillion-dollar companies before the headlines ever hit. See what others can’t. Start your free CB Insights trial and explore which private companies are showing breakout signals → https://lnkd.in/ezFRNcnW

Every month, we launch new predictive intelligence capabilities that help our customers make their next move first — identifying the right companies, markets, and opportunities before competitors do. Here are the latest upgrades: 🗺️ 𝐕𝐢𝐬𝐮𝐚𝐥𝐢𝐳𝐞 𝐭𝐡𝐞 𝐬𝐭𝐫𝐚𝐭𝐞𝐠𝐲 𝐛𝐞𝐡𝐢𝐧𝐝 𝐞𝐯𝐞𝐫𝐲 𝐜𝐨𝐦𝐩𝐞𝐭𝐢𝐭𝐨𝐫 𝐦𝐨𝐯𝐞 Introducing Strategy Maps. They connect acquisitions, partnerships, and investments into one visual story — showing how your competitors are positioning before it’s public. So you can respond instantly, or move first. 🔍 𝐔𝐩𝐠𝐫𝐚𝐝𝐞𝐝 𝐀𝐝𝐯𝐚𝐧𝐜𝐞𝐝 𝐒𝐞𝐚𝐫𝐜𝐡 — 𝐭𝐡𝐞 𝐦𝐨𝐬𝐭 𝐩𝐨𝐰𝐞𝐫𝐟𝐮𝐥 𝐜𝐨𝐦𝐩𝐚𝐧𝐲 𝐬𝐨𝐮𝐫𝐜𝐢𝐧𝐠 𝐞𝐯𝐞𝐫 We’re taking your feedback and upgrading Advanced Search so that it’s more powerful, intuitive, and precise than ever. Zero in on the right companies by combining filters and tags you won’t find anywhere else — everything from success probability to projected revenue. Save your column set up. Use Boolean logic. Advanced Search gives you company sourcing superpowers. 🧠 𝐂𝐡𝐚𝐭𝐆𝐏𝐓 𝐩𝐨𝐰𝐞𝐫𝐞𝐝 𝐛𝐲 𝐂𝐁 𝐈𝐧𝐬𝐢𝐠𝐡𝐭𝐬 𝐝𝐚𝐭𝐚 Ask questions in ChatGPT and get answers powered by CB Insights’ predictive intelligence data. The new CB Insights for ChatGPT API is a turnkey way to plug CB Insights directly into your company’s existing AI tools. Not a ChatGPT shop? We support other AI models, too. 🔮 𝐏𝐫𝐞𝐝𝐢𝐜𝐭𝐢𝐯𝐞 𝐒𝐢𝐠𝐧𝐚𝐥 𝐟𝐞𝐞𝐝𝐬 𝐬𝐡𝐨𝐰 𝐲𝐨𝐮 𝐭𝐨𝐦𝐨𝐫𝐫𝐨𝐰’𝐬 𝐨𝐩𝐩𝐨𝐫𝐭𝐮𝐧𝐢𝐭𝐢𝐞𝐬, 𝐭𝐨𝐝𝐚𝐲 Get daily alerts on potential partners, fundraising targets, and acquisition prospects — months before they’re on anyone else’s radar. Each signal comes with clear analysis on why they’re flagged, why they matter, and next steps for you to take. Learn more: https://cbi.team/3J6smZd

The CB Insights team has landed in Vegas for Money20/20 and we’re ready to talk predictive intelligence. If you’re at Money20/20, Stop by Booth #18051 to connect with our strategists and learn how to get the most out of your 30-day trial (see below). We’ll walk you through how to: 🔎 Scout and evaluate private companies in seconds 📊 Size markets and spot breakout trends before they’re obvious 💥 Use AI to separate real signals from noise in private markets Meet the team: John P. Kelly, Aisha Chandraker, Jason Saltzman, Collin Levandowski, Nathan Lowery, Patrick Besse, Yael Sassoon → Activate your 30-day trial: https://lnkd.in/gBpTeFXk

2PM today on the Off the Record Stage at Money20/20 - I'll be moderating "Inside the Jury Room - the Money Awards" with three INCREDIBLE panelists: Diamond Jury President Michael Harris, Vice Chair & Global Head, Capital Markets, NYSE, Payments Jury President Howard R. Fields, Chief Ethics & Compliance Officer, Mastercard Banking Jury President Joanne Hannaford, Chief Information Officer & Chief Product Officer, Corporate Bank, Deutsche Bank Join us to get firsthand insights into why this year's Jury decided on the work that will set the benchmark for the year ahead!

Come see CB Insights in the Money Hall today at Money20/20!! We’ll talk you through our fintech team’s recent research and show you our data platform.

The limit of AI is compute. Especially when it comes to inference... Reasoning AI, demanding 100-1000x more compute than simple chatbots, faces a bottleneck: GPUs optimized for parallel training struggle with sequential inference workloads. NVIDIA projects a $3-4T AI infrastructure market by 2030, where inference capabilities and cost economics will be the determinants of AI application viability at scale. Investors have already poured over $9.5B into AI inference and training processor startups in 2024. Many of these leading players are rapidly scaling to $1B+ revenue (<2 years), and are delivering 10× performance gains and 3× efficiency improvements versus traditional GPU-based systems. Key innovations that will drive the next wave of inference: ↳ In-memory compute eliminates data movement between memory and compute cores ↳ Wafer-scale integration delivers 3,715× better chip-to-chip bandwidth than GPUs ↳ Deterministic architectures optimize for sequential processing vs. parallel workloads ↳ HBM + chiplet designs achieve performance-per-watt gains that make inference economically viable ↳ 2nm foundry capacity creates supply chain alternatives beyond the TSMC/Samsung duopoly The leading startups solving for inference infrastructure at scale: → Groq: 2M+ developers on GroqCloud delivering 10× faster token generation through deterministic LPU architecture, strategic investments from Saudi Arabia and telecoms for data center expansion → Rebellions: Production LLM services running at SK Telecom and Aramco with 3× performance-per-watt via HBM + chiplet architecture, Samsung, SK hynix, and Arm investments validate both technology and go-to-market → Rapidus: $12B government-backed 2nm foundry with Tenstorrent as first customer, Toyota, Sony, and NTT consortium backing positions Japan as TSMC/Samsung alternative → d-Matrix: In-memory compute achieving 150 TB/s bandwidth and 20× lower latency, secured Supermicro and GigaIO partnerships for data center integration → Cerebras: Scaling from $350M to projected $1B revenue in 2025 with deployments at AWS Marketplace, Meta AI, and Sandia National Labs powering frontier AI research We're in the middle of an architectural rethinking of the compute-memory hierarchy for inference workloads. As tech companies build their custom silicon and traditional chip makers fight to stay relevant against NVIDIA, companies with differentiated inference IP and proven production deployments become acquisition targets. P.S. Comment "inference" for *free* access to CB Insights' full market intelligence on the companies building the future of AI inference and processing.

The next trillion-dollar companies are being built in the dark. Join Manlio Carrelli and executives from Nasdaq, Forge, Banco Itaú, and Tiger Global for a discussion on how AI and predictive intelligence are dragging private markets out of the dark and into focus. Private markets have been the blind spot of finance… until now. → Details here: https://lnkd.in/gmrEjX7a