The Value of Distribution is Increasing in the AI Boom
As an extreme simplification, there are two things that a company must get right when going-to-market with a new offering: product and distribution.
Different companies are stronger or weaker on these two dimensions. Incumbents in a market generally have strong distribution. Meanwhile, in most industries, the nimbleness and talent profile of startups enables them to build strong products more rapidly to adapt to technology changes. But a better product alone doesn’t win, as many wide-eyed entrepreneurs have learned.
During past technological paradigm shifts, speed of building a strong product was more important in determining the winners — and startups benefited. During the dot-com boom and the mobile wave, the biggest returns went to the most successful startups rather than major incumbents, and the VC investment class had the greatest success.
This is because there were natural network effects built into most of the products that were created in those waves. During mobile, the biggest winners were companies like Uber and Instagram — which moved quickly and nimbly to create great products, well before any incumbents could enter the market. By the time the incumbents could react, the market was already locked up because of the extreme network effects.
AI businesses don’t have the same inherent network effects. Unlike in past technological paradigm shifts, the companies spawned by this shift don’t have obvious network effects to propel them if they are the first-movers. No vertical AI CEOs have a good answer at this point for why their business is defensible; it’s usually a version of “it’s a hard problem” rather than the obvious network effects of an Instagram or Uber.
So that means the best products will be commoditized. The lack of network effects means that, instead of there being one runaway category-winner, it’s likely that every category has dozens of companies with comparable products. Uber and Instagram were winner-take-all, there’s no reason that “vertical AI to automate X workflow” won’t have 20 equivalent companies fighting over the same share.
And it means that incumbents have an advantage. The lack of network effects also means that, even if a nimble startup builds an incredible product two years ahead of the incumbent, the incumbent will still be able to use its distribution advantage to win the market after it eventually catches up. Two years was an eternity because of Instagram’s network effects (there was no way Facebook could have built a competitor with that deficit, or Google Photos could have prevailed). Without network effects, a two-year headstart simply doesn’t matter that much. If a healthtech company builds “AI for XYZ,” and Epic builds it two years later at 80% the quality, I would bet on Epic winning.
The economics of AI favor incumbents: if real innovation is being produced, but the product itself is commoditized — the value of distribution will increase. AI will create massive winners: real innovation is going to come to market, and real value will be created — so buyers are eager to step forward for a variety of new capabilities. But the beneficiaries will be the incumbent with strong distribution; they can either fast-follow on product development, or (more often) acquire from amongst the 20 slight variations of startups in a given category, making the acquisition relatively inexpensive.
***
As a practical example of all this, the medical coding space for providers in healthcare (i.e. converting medical records into billing codes) is an example of a sector where i) AI is going to be transformational on the industry, ii) there are a flurry of innovative companies going after the problem, and iii) I would bet on an incumbent with good distribution ultimately winning (either by building its own product or doing a tuck-in acquisition).
I believe the AI boom will create at least as much value as the dot-com boom and mobile wave, but I think it is less clear that the traditional VC asset class will do well or that as many incredible startup successes will be minted in vertical AI plays [though I am still extremely bullish on startups that are infrastructure for AI, or that are complimentary to AI]. Instead, for vertical success, I would bet on incumbent companies that figure out how to hire well and acquire well, PE strategies of combining AI companies with incumbents, and tech-enabled rollup plays.
