Introducing Zero Index

Zero Index is a first check, high conviction venture fund focused on emerging companies in this massive AI wave. We write the earliest money into a company, at day zero, before the lead investor, before the traction chart, sometimes before the product.

My first neural net

I took my first AI class at the University of Michigan in 2001 and it captivated me completely. I told my academic counselor this was what I wanted to spend my career on, neural nets and how far we could push machines toward the way a brain operates. After I graduated, I spent a year building predictive systems for financial markets, hand-coding my own neural nets in Microsoft Visual Basic (don't ask). There were no frameworks to import and no datasets to download. If I wanted sentiment analysis on a Yahoo Finance article, I wrote it myself. The system ended up working reasonably well. I didn't have the capital to scale it, and I didn't yet know how to raise any.

In 2004 I started a company with Markus Nordvik and Zaw Thet. 4INFO began as a location-based services play. The FCC's E911 rules meant carriers would soon be able to pinpoint any phone, and our brains lit up with everything you could build on top of that. The market corrected us quickly. The phones of 2004 could not do location, and their screens could barely do anything. The one thing every phone could do was text. So we built one of the first search engines that ran over text messaging. You texted us a question and got back directory listings, sports scores, stock quotes, flight status, weather, etc. To make all that work, we had to figure out what a user was asking for. I built the language processing and the neural nets that did it, and we filed multiple patents for that work.

Twenty-five years later, that promise is coming true, and it is more incredible than anything we imagined at the time. The neural nets I hand-rolled in Visual Basic have become models that reason, write, and see. Nobody hand-rolls a neural net anymore. Honestly, nobody hand-rolls much of anything anymore. I watch Claude Code write software while I think about what to build next. What captivated me in that Michigan classroom can finally do the work. And the market that unlocks is the biggest opportunity I have seen in my lifetime.

AI is bigger than software

In 2011, Marc Andreessen wrote that software was eating the world. It looked aggressive at the time. The entire global SaaS industry did about $12 billion in revenue that year, and skeptics sized the opportunity off the visible market for tools. They were wrong. Software spend now runs $1.23 trillion a year, SaaS grew roughly 25x, and the companies that proved him right, Stripe in payments, Uber in taxis, Shopify in retail, became the defining growth stories of the era.

But notice what all of that software had in common. It sold tools. The customer still did the work. AI breaks that boundary. As Sequoia's Julien Bek puts it, a copilot sells the tool, an autopilot sells the work. And the market for work dwarfs the market for tools. For every dollar spent on software, six are spent on services. Even inside IT budgets alone, services outspend software.

The capture is already underway, and it is fast. Stripe's cohort data shows top AI companies hitting revenue milestones about four months faster than the best SaaS companies ever did. Harvey passed a reported $300 million in annual revenue doing legal work. Anthropic crossed a $47 billion revenue run rate this May. Read that number against the baseline. One AI company is running at roughly four times the revenue of the entire SaaS industry at the moment the software-eats-the-world thesis was written.

Yes, prices will deflate as AI does the work more cheaply than the people it replaces. But the pool it is draining is six times the size of the one software drank from. Bek's conclusion is the one we would underline. The next trillion-dollar company will be a software company masquerading as a services firm.

The returns went private

As a kid growing up in Redmond, Washington in the 1990's, I watched a disproportionately large number of my friends' parents go to work for a company called Microsoft, so I bought some shares with my savings. They ended up paying for my first car.

That trade was available to a teenager because Microsoft went public early. It raised $61 million in 1986 at a valuation around $520 million, and every multiple it earned came after that day, in public, where anyone with a brokerage account could ride it. SpaceX went public this summer at a valuation near $1.8 trillion, twenty-four years after its founding, instantly one of the ten most valuable companies in the world. Its multiple was already made. A buyer at the IPO gets whatever is left.

The median company now goes public at 14 years old, up from 8 in the 1980s. And the math at today's entry points is unforgiving. For a buyer at SpaceX's opening valuation to make 100x, the company would need to reach a market value around 1.5 times world GDP.

Where is this era's value sitting instead? Crunchbase counts private unicorns collectively valued at more than $8 trillion. OpenAI was last valued at $852 billion and Anthropic at $965 billion, and both have now filed to go public, which means their first thousand-x will have happened entirely on the private side of the wall. And it is a real wall. By the SEC's own count, only 18.5 percent of American households qualify as accredited investors. For everyone else, the companies defining this era are legally out of reach until the compounding is over. The kid I was in Redmond bought the future with a savings account. Today, that kid can only watch.

Seed outperforms every other stage

Venture, taken as a whole, is a mediocre asset class. Cambridge Associates' index of more than 2,600 venture funds returned 8.0 percent annually over the past 25 years, net of fees, against 8.9 percent for the S&P 500. If someone offers you the average of venture, keep your index funds.

But that index is weighted by dollars, and most of the dollars sit in large late-stage funds. Break venture apart by stage and the picture inverts. PitchBook's stage benchmarks put seed around 36 percent annualized, Series A at 25, Series B at 18, declining from there, and PitchBook's own summary is that seed has historically offered the highest annualized returns of any stage. AngelList's analysis of thousands of early-stage deals found the same thing from the bottom up. Holding the companies constant and varying only the entry round, seed entries draw from a far heavier-tailed distribution than later entries, so much so that the authors conclude early and late stage venture should be treated as distinct asset classes.

The mechanism is entry price. The median seed round today is priced around $24 million. The same company at Series A is near $79 million, and by Series D past $460 million. A $10 billion outcome pays roughly 417x from seed and roughly 22x from Series D. One seed check can return a small fund several times over. No late-stage check can do that for a growth fund. Fund size data says the same thing, small funds beat large ones at the 90th percentile in every recent vintage Carta tracks.

Seed also loses more often than any other stage. Roughly 65 percent of early-stage checks lose money. Both facts are true, and they point at the same strategy. You win at seed on the tail, not on per-check odds, which is why we hold a broad portfolio of them.

How we work

We work the way we wished investors worked when we were founders. One conversation. An answer within 24 hours. Your standard SAFE, no board seat. After the wire we help you fill out the round, find customers, and tee up the downstream investors actually hunting in your space, and we show up for the unglamorous parts too, even if it is just emotional support on a rough day. Like the companies we back, we run lean and AI-native, so our time goes where the models can't, to founders.

We also built Capvia, a platform that gives founders honest automated feedback on their deck, lets them rehearse their pitch against AI investors, and helps them find and manage the right investors for their round. It is there whether or not you ever pitch us.

Day zero

We have never been more excited about anything in our professional lives. The promise we saw back in 2001 is finally here, and the best seats are at the beginning.

If you are building at day zero, we would like to meet you. Write us at founders@zeroindex.vc.