The ever-shrinking investable box

The largest technology shift since the Internet hit the scene in the late 90’s is upon us. AI is proving to be similar to the Internet, a utility-like intelligence layer with an adoption rate that is generating eye-popping growth for the frontier labs leading the charge. Anthropic’s $50B revenue run-rate (up 5x YTD) is evidence that LLMs are delivering new value. The majority of this revenue run rate is generated from automated software development, a consumption of labor value versus displacement of legacy software. With just the software development use case, it’s very clear that AI will expand the software market size as it enables new degrees of productivity.

Yet, despite the clarity that AI will drive market size expansion, the disruption of the opportunity is causing meaningful confusion and anxiety for investors in existing software companies. The dual conversation of AI code automation combined with Anthropic’s landgrab into specific verticals and business functions has put most classic SaaS investors on ice. The big PE and growth equity push into software over the last 15 years is now largely on pause, as the market tries to make sense with the future terminal value of their long-software portfolios. 

While LLMs were widely commercialized starting in 2022, the power of AI became obviously clear to everyone during 2025. Companies that understood the opportunity in the ‘22-’25 period got a jump in the large market opportunities such as software coding and legal-tech, and now there is a race to bring an “AI-native” product to market across all verticals. We’re calling all these new opportunities AI-Forward companies.

We recently met with a software investor that joked about how not long ago investors looked at “Chat-GPT Wrapper” as negative, and now this whole architecture approach has been rebranded as an “LLM Harness” - and it’s a great thing. It only took a year for everyone to realize that LLMs and the resulting agentic approach to products is the definition of AI-Native, and the teams operating traditional deterministic software products are working hard to be relevant in the eyes of investors looking for AI-Forward opportunities.

Given all of this movement in the market and the caution for anything but true AI-Forward companies, the deal specification “box” has become incredibly small. During the last 15 years, any reasonable cloud software / SaaS business could get funded. The box is so small today, that only the highest quality early-stage businesses are investable.

Said another way, in today’s market, investor expectations are much higher than in prior periods. For example, we’ve adjusted our criteria for greater levels of growth - whereas we previously would consider a business that is doubling year-over-year, at our stage we’re now expecting companies to be growing at 3-10x annually. The expectation is that AI applications will be able to either take share from incumbents or create new market expansion via selling labor displacement value, and this advantage should produce more momentum. We’re seeing this is true in our portfolio - Madison AI is growing at a 10x Y-o-Y clip in the local government market, for example.

Gross retention is the other performance measure where expectations are now higher. Software investors have always preferred 90%+ GRR businesses, but in today’s market this 90% GRR threshold is a hard line for many investors. GRR is the best indicator of a business’ market-market quality, solution value proposition, and product competitiveness all wrapped into an easily measured and tracked metric. SaaS investors have always known that it’s much easier to grow when there is no leaky bucket, but now this metric is more in focus than ever. We would summarize the current market’s view on GRR as follows: 1) >90% GRR (and high growth) = 15x ARR early stage valuations; 2) 80-90% GRR (and high growth) = 8x ARR; 3) <80% GRR (and high growth) is no bid. So for a company with $5M ARR and >100% growth, there is a ~50% discount applied to a business that churns 10% more of revenue than the acceptable >90% GRR threshold.

CreativeCo is typically the first institutional capital into a seed-stage business. Given that our best performing companies need incremental capital beyond our check to grow, how does this shrinking investable box impact our investing point of view?

First, the strength of a startups team, and in particular their product team, has never been more important. For years, a domain expert could team up with a reasonably talented technical person and build a SaaS solution simply by following well-established architecture and UI design patterns. SaaS became so mature, there was a clear template to essentially copy for a market’s requirements. That is not the case today. Agentic solutions based on new LLM architectures is an entirely new paradigm. Teams need to be creators / inventors / fast-learners to operate in today’s shifting market.

The second point is related to the first. There is an observation that engineers inside the frontier AI labs are 3-6 months ahead of Bay Area engineers, who are in turn 3 months ahead of New York engineers, who are 3-6 months ahead of the rest of the geos. If you believe that learning curve, that leaves engineers in places like Atlanta, Dallas, or Tampa a good year+ behind in adoption and skill, which compounds as time advances. We believe the idea that we can uncover amazing AI-Forward companies in secondary markets is somewhat a false assumption, and accordingly we are increasingly researching major markets from a sourcing perspective.

The third observation from our POV is that west-coast VC is now targeting niche market / small TAMs. We are seeing very early stage companies raise meaningful west coast money for AI-Forward businesses into markets that used to be off limits to VC due to assumed market size. Bay-area startups are coming out of y-Combinator etc into landscaping, home services, self storage, construction niches, and various other smaller vertical TAMs that used to be areas of low product competition. If you take this trend of VC dollars funding Bay Area teams into smaller TAMs, you can see why we’re leaning into the importance of better technical teams coming out of bigger markets for AI-Forward businesses.

We see these trends clearly in play. So what’s our opportunity? We believe the answer is greater upside potential for companies that are excellent. Valuations for premium quality companies will continue to extend. These premium companies will be mostly coming out of major tech geographies, will be top technical teams, will have AI-Forward products, and will have the growth and GRR metrics to command 15x or better valuations. Then we’ll have the binary opposite of this high valuation profile - the pre AI product, lower GRR, lesser team, etc. These businesses will find it difficult to raise new capital and will be forced into a sideways pattern with no bid until they re-invent their story. It’s going to be a tough market for lower quality deals. But the premium deals will get bid up more than ever. We really need to stay on the premium side of this equation.