🌱 5-Bit Fridays: The gaming gambit, the AI price-fixing dys...

WHY INVESTORS DON'T FUND AI STARTUPS you might know A16Z is running a seed-stage startup accelerator called SPEEDRUN where we're funding 40+ companies in the next 2 months with $750k per company. (apply here: https://t.co/lExlSgiixi). As part of running this accelerator, we get thousands of new startup ideas sent to us. These days, the majority of them, like >60%, have a major component of AI. Yet when I talk to other investors — angels, seed funds, other multi-stage VCs — it can be a challenge to sort through all of these early startups and find the ones that stand out. Even though AI is the focus in the entire tech industry right now, and of course plenty of companies are being funded, the majority of startups won’t find investors. So why do AI startups end up getting turned down, even when the teams and ideas are good? And what differentiates the ones that succeed? Some common reasons: - Everyone says they’re an AI startup - Often it’s not clear which will win- the weak form and strong form - Everyone can integrate into the same APIs, what’s the defensibility over time? - There’s a real difference between AI apps and foundational models - Lots of growth driven by novelty but what about retention? - Hard to pick winners when it’s early. Mobile wave with Flipboard and Foursquare. Better to wait - Lack of proven business models - Concerns about hype and overvaluation Every startup is now an AI startup Yes, we are at the phase in the market where every startup has now taken the hot buzzword and incorporated all through their slides. This is the case even if the product doesn’t use an LLMs or recent breakthroughs in generative AI. Or it might just be a little side feature that was built quickly in the past 3 months — yet the entire startup and product will be positioned as “we are an AI-first XYZ that does ABC.” This creates a ton of noise, and makes it much harder to tell what’s actually going to work. Yes, your photo editing app might include generative fill as one of the features, but if 99.9% of the product is the same as it was 12 months ago, that’s hardly AI-first. A year ago, an AI startup would clearly be one doing research and building one of the obviously very valuable foundational models. But these days, we are starting to run out. As text, photos, video, 3D, music, etc all start to mature, it becomes a little harder to know what types of models will make as big of an impact. As a result, we are moving further up the apps layer in the next 18 months, which is great for all of us consumers! However, the dynamics of AI apps and foundational models are very different, even if both sets of startups call themselves AI startups. For apps, we often talk about weak form and strong form. This is especially true when it’s sometimes unclear which class of products will win: The AI-native version of the incumbent who adds AI features. In the industry where I spend most of my time, and where we’re running the SPEEDRUN accelerator — the intersection of Tech x Games — there’s an open question of whether AI will be infused into products like Roblox, and if that’s the winning combination. Or will it be smaller startups that invent new genres that incorporate AI in novel ways. Or both? It’s hard to tell. Defensibility is also a major issue in the debate. If many of the AI apps in particular are just wrappers of ChatGPT, and every startup has access to the same APIs just with different UX — if AI is a primary interface, will everything become commoditized? We just ran through a decade plus in the mobile industry where defensibility mostly came from factors like network effects, and so maybe there’s a way to combine the two. But in a world where many of the AI features come from single player tools and less from social/collaborative apps, there’s a question of what will work. Lots of growth but what about retention? At the moment when you look at a lot of data rooms for AI products, you’ll see a TON of growth — amazing hockey sticks going 0 to $1M and beyond — but also very high churn rates. This comes the sheer amount of excitement in the market, where people are trying a ton of different products. It’s not a bad thing. But also it causes a bit of cognitive dissonance because the top line number often implies a level of product/market fit that isn’t there — if you were to compare to traditional non-AI products. This is why a lot of folks, rightly so, have moved towards AI prosumer products where you can at least monetize users right away. And that monetization might be much stickier than an AI app that’s just focused on consumer social interactions, or frankly, even gaming experiences. I think the next phase of growth will be around “Applied AI” categories where you really do see strong product/market fit, and a lot of fast-followers into those spaces. And there will be plenty of categories where the current AI SOTA will not be quite good enough to disrupt. Some industries will take time. Nevertheless I fully expect that over the next 2-3 years, some of the novelty around AI is bound to wear off as new products arrive in the market. Given that, we may see acquisition tail off, and the market will become dominated by the products that are high retention, not just lots of new users. And yet, there’s no better time to invest Because the entire market is so buzzed about all the innovation happening in AI, investors are also talking about lot about timing. That is, if you got into the mobile wave and did all the early investments in consumer mobile apps like Flipboard or Foursquare, but then sat out the later waves for Uber or Instagram, that wouldn’t be good! One view is that instead of trying to time the market, you just need to invest all the way through. I think this is right. But as valuations are often sky high, smaller funds and angels may not be able to pull that off. So instead, folks might invest a little too early, which might be the same as just being wrong. Nevertheless, investors are obviously very excited about the entire AI category — it’s where founders are spending their time, and it’s really the main hot spot. So while most AI startups are not getting funded, most of the funding is going towards AI startups :) For the accelerator, SPEEDRUN, that we’ve been running for the past year, we’re excited to partner up with founders working at the very very earliest stages of their AI startups. We’re focused on the intersection of TECH x GAMES, which includes AI/infra, 3D, VR/AR, game studios, gamified consumer apps, and much more. We plan to invest $30M over the next 40ish days — $750k a company — and for those of you who are super early and thinking about starting a new company, please join us! Apply here: https://t.co/lExlSgiixi