We all know about "crossing the chasm" when it comes to starting with early adopters and moving on to the majority A necessary action for any startup pursuing scale But, this framework isn't helpful for industrial startups...

Acceptable TAMs for venture-backed companies over time: 2015: $1B+ 2020: $10B+ 2025: $100B+ 2026: US GDP

30 funds have raised 36% of all VC and now lead 34% of seed rounds by value. We talk about capital concentrating at the top but it's actually concentrating all the way down. PitchBook compared the Top 30 VC firms to the rest of the market at seed. (Top 30 by performance, not size, although there's heavy overlap.) These firms aren't just raising an outsized share of capital or dominating the biggest growth rounds. They're dramatically changing the dynamics at seed. The median seed round led by a Top 30 firm was historically 2x the size of overall median. This year, it's 3x. That changes pricing, runway, and how big a swing a company can take. But the bigger question is what flows from here. Is it a given that the top Series A companies will increasingly come from the top seed cohorts? Or will performance reshuffle the deck between rounds? There used to be real concern about adverse selection when a top fund invested at seed and then didn't follow on. But that was when these firms participated in ~5% of the market five years ago, not the estimated ~15% today. It will be interesting to see whether the winners at seed naturally graduate to become the winners at A and the market bifurcates further. Or if the market resets between rounds, leaving a much larger pool of companies with an even higher bar to clear given the expectations that come with brand and access to capital. Which are you seeing at A right now?

Airtable raised $1.4B. Bending Spoons just bought it for $1.28B. DroneDeploy raised $157M. Procore just acquired it for $845M. One returned less than it raised. The other, 5x. A good reminder: vertical applications are often more efficient, better at R&D, and tackle bigger markets than VCs expect. Two years ago, I compared Procore with the leading horizontal project-management platforms. Procore was the standout. The DroneDeploy and Airtable acquisitions make the case stronger. Read more in Unlocking hidden value: How niche markets are bigger opportunities than VCs think jaydimonte.substack.com/p/un…

Series A is the new B? Try again. A is the new C. The median Series A in 2026 is nearly $20M 🟰 Series C round a decade ago 🟰Series B in 2020. The median Series B is now about $40M, roughly what a late-stage growth round looked like six or seven years ago. Meanwhile... Seed rounds are up, but they haven’t kept up. 📈 The median seed round is ~$3M… up 2.3x over 10 years versus 4.3x for A and 3.6x for B. At the same time, companies are expected to accomplish far more before each subsequent round than they were a decade ago. Rational at the letter rounds, but harder to do with seed. The bar for an A has grown faster than resourcing. Perhaps one reason seed-to-A conversion keeps falling.

🦄 AI unicorns are in a league of their own. Bad news for the class of 2021. Roughly 30% of unicorns minted before 2022 still haven't raised another round. After a few years of pullback, unicorn creation is accelerating again. This year is on pace to surpass 2022. Will we see the same rate of zombiecorns out of this cohort? Maybe not. The dynamics are different. Volume is back to 2022 levels, but dollars have blown past them. The implications are significant: ➡️ Venture is becoming more concentrated. Instead of hundreds of breakout companies attracting capital from individual firms, we're increasingly seeing dozens of mega funds concentrated into dozens of companies. ➡️ Outcomes become more correlated. The industry's returns become increasingly dependent on the same small set of winners. ➡️ The incentives change. When many of the largest funds own meaningful stakes in the same companies, they also have the capital and incentive to keep supporting them.

In hardtech, engineering innovation is what gets you started but supply chain is how you win.

Angels make the world go round. 😇 Per AngelList, the median commitment size for a... <$5M fund is $10K. $5–20M fund: $50K. $20–250M fund: $100K. Fund commitments at these check sizes are overwhelmingly coming from HNW individuals… and likely a lot of overlap with folks that are (or would be) angel investors. This shows up in the data: 50% of dollars going to <$20M funds come from individuals (vs. family offices and institutions). But that participation is slowing, and that’s meaningful. I've written before about the decline of angel-led rounds in the last few years. The slowing of small fund fundraising might trace back to the same pullback. All to say, just like angels in early companies, those first LPs into emerging funds are special. At Grid, a number of those folks backed us fast and early. Forever grateful for their partnership.

Is AI delivering real value? The answer may surprise you. Almost 20% of the 81K AI users Anthropic surveyed say AI has not delivered value to them. That's more than people who say AI is a great thinking partner (17%), helps them learn new things (10%), or synthesizes research (7%). The one exception is productivity, with 30% of users seeing gains. For me, the biggest takeaway is that we're still experimenting with what AI is, how to use it, and where to extract value. And as models get better and more capable, they restart the experimentation process entirely. What features were not helpful a month ago are dramatically more useful today. Will be interesting to see the next time they publish data. Will "no value" still be among the top rankings?

SPVs are booming. But where the money is going tells a different story than where the deals are. Great data from Sydecar on SPV activity last year: 📈 SPV deal count 2x+ last year and capital invested 2.5x+ But look at the split: 🔵 Primary SPVs account for 75% of deals but 54% of value 🔴 Secondary SPVs made up 25% of deals but 46% of value In other words, secondary SPVs are fewer deals with much bigger checks: 🔵 Primary SPVs tend to be angel groups and smaller syndicates pooling capital into early rounds 🔴 Secondary SPVs are for accessing proven (hot?) opportunities As the exit slump continues, my guess is secondary SPVs are becoming one of the primary tools for getting capital in (and out) of late-stage positions. Expect 2026 to be more of the same.

We’re in the early innings of AI, and no one really knows what comes next. Ramp published data on the percentage of their customers purchasing AI tools, and it's wild. Some notes: Almost 50% of businesses are paying for AI, up from 5% roughly three years ago. Increased adoption of 🟢 OpenAI and 🔵 Claude is fueling most of the growth. The former was quick out of the gate, the latter is accelerating fast. 🟡 Google, despite a built-in distribution advantage, hasn't seen meaningful incremental spend. What's different this last year is that businesses are spending on 🟢 🔵 🟡 MULTIPLE platforms. Early last year was the first time we saw meaningful overlap. Now, nearly 20% of volume comes from companies paying for more than one provider. We've framed this market as a battle of the models, perhaps so much so that there will be one "winner." In reality, there'svery little loyalty and data moats aren't creating lock-in. But unlike prior startup battles (i.e. Uber vs. Lyft), competition is happening through product advancement, not price. LLMs are not commodities (yet?). With a prize so big, platforms are sprinting at it. New capabilities on what feels like a daily basis. For now, the consumer is the winner. Excited to see how the next 3 years play out.

📊 45%+ of workers say they “use AI.” The real number is so much lower than that. Most people aren’t using AI the way people in tech mean it 𝘵𝘰𝘥𝘢𝘺. They’re using LLMs to edit emails, summarize documents, or replace search. Meanwhile, the pace of change is so fast that “using AI” isn’t a one-time adoption event. It requires constant learning, unlearning, and tinkering. Every new wave of AI capability resets the adoption curve. So, where we think we’ve hit the majority, it’s still the early adopters leading the charge. I’m seeing two opposing forces at work: 📉 Some people fall off the learning curve completely, adopting one use case and getting stuck there. 📈 Each advancement makes the most powerful features easier to access, pulling new users in. A lot of the anxiety about the future comes from confusing those two curves. We may think we’re "late" but I think we’re just getting started. Think of... what's the best thing you can do with AI now that you couldn't three months ago?

AI slop is coming for the enterprise. Today, the 𝘱𝘦𝘰𝘱𝘭𝘦 𝘣𝘶𝘪𝘭𝘥𝘪𝘯𝘨 𝘵𝘩𝘦 𝘮𝘰𝘴𝘵 𝘸𝘪𝘵𝘩 𝘈𝘐 are early adopters. The 𝘤𝘰𝘮𝘱𝘢𝘯𝘪𝘦𝘴 𝘸𝘪𝘵𝘩 𝘵𝘩𝘦 𝘮𝘰𝘴𝘵 𝘱𝘦𝘰𝘱𝘭𝘦 building with AI are early-adopter companies. That won't last. As these tools improve, more people will use them across more companies. It won't just be the tech-natives. It'll be everyone. 🙅‍♀️ It'll be Pam from procurement who builds her own agent… that promptly rejects a PO from a new vendor for critical parts. 🙋‍♂️ It'll be Max from marketing who spins up a web app… that burns through $100K of compute when his branded meme generator goes viral overnight. Now multiply that across every employee, every department. Think about how many mistakes we've already seen very capable people make with the release of clawdbot. Then scale that to the rest of the enterprise. We're not yet in the age of "everyone builds." But when we get there, the real question won't be how. It'll be for how long. Then it's back to permissioning, security, and… SaaS again. It's one of the reasons behind why Grid is focused on net new operating systems and outsourced services [🔗substack.com/@jaydimonte/p-1…]. Everything in between feels somewhat cyclical at the enterprise.