Who wins generative AI markets?
There are two types of generative AI market: new markets being made from totally new products (like rideshare was in 2010, or Midjourney today), and markets of existing products that incorporate generative AI into their products (like any conversational AI company rolling out LLMs into their chatbots)
Totally new markets
The virtue of winning a new, growing, and valuable market is well understood. Startups can be advantaged in new markets because they are more nimble, can be more appropriately capitalized, can have better focus without competing organizational priorities, etc. All of the typical virtues that enable startups to win. The only exception is if the startup needs abundant capital to train a foundation model, in which case incumbents have lower cost of and better access to capital.
The macroeconomic scale is also tipped in favor of startups. Larger companies are often encouraged these days to cut costs in order to enhance profitability. Those that trim R&D spend too much are probably most at risk of unsuccessfully entering totally new markets. I expect that incumbents that shell out to hire great AI talent and fund relevant R&D will successfully enter new markets at a greater rate than companies that trimmed down relevant spend.
It’s not the end of the world though if an incumbent does not successfully enter new markets so long as they can continue to win and defend against gen-AI native startups in their existing market.
Existing markets
The virtue of being a re-envisioning of current markets is poorly understood. I think the best analogy is what happened with cloud migration over the past ~18 years.
For any given existing product, the key questions are, to what degree do products in this market need gen AI, and to what degree do they need to be rebuilt ground up to incorporate gen AI? Not at all? A lot? Somewhat?
If it’s marginally more valuable to use the version of the product that incorporates gen AI than it is to use the version of the product that doesn’t incorporate gen AI, the market will probably adopt the version of the product that incorporates gen AI.
The question is, for products that are better when gen AI is incorporated, do startups win or do incumbents win? I think this is where cloud migration becomes a good analogue. Sometimes products that were rebuilt cloud native from the ground up won, and sometimes incumbents merely needed to rearchitect to transition to cloud to retain dominance.
I think this paradigm works for gen AI. Sometimes the incumbent will incorporate gen AI and then beat challenger startups, because the product didn’t need to be rebuilt as gen AI native, and because the incumbent had distribution that crushed the startup. And sometimes the startup will reimagine the “v2” of the product that is gen AI native to which the market responds more favorably.
There are a lot of permutations of how markets could play out so I wouldn’t anchor on the extremes here. You could have a market where the incumbent doesn’t go after the new “v2” of the product, because that “v2” of the product is only demanded by 5% of their market, so it might not be worth it for them to retain that customer. One of a million possible hypotheticals.
Winning startups
These are both interesting market dynamics - they’re just a different game. I think that if you find market pull, it’s probably easier to win in the totally new market because the advantage of the incumbent is much higher in markets where they are already dominant than it is in totally new markets.