The Excuses Change, the Hackquisitions Remain the Same
Ivan Mehta
One's a dot. Two is a line. Three is a trend:
In a big move, Character.AI co-founder and CEO Noam Shazeer is returning to Google after leaving the company in October 2021 to found the a16z-backed startup. In his previous stint, Shazeer spearheaded the team of researchers that built LaMDA (Language Model for Dialogue Applications), a language model that was used for conversational AI tools.
Character.AI co-founder Daniel De Freitas is also joining Google with some other employees from the startup. Dominic Perella, Character.AI’s general counsel, is becoming an interim CEO at the startup. The company noted that most of the staff is staying at Character.AI.
Google is also signing a non-exclusive agreement with Character.AI to use its tech.
If this all sounds familiar, it's because it is familiar. Microsoft did not acquire Inflection, they hired some of the team – notably the co-founders – while paying the company for a non-exclusive licensing arrangement. Amazon did not acquire Adept, they hired some of the team – notably the co-founders – while paying the company for a non-exclusive licensing arrangement. Considering that there's not even an attempt to hide the similarities here, presumably Google (and Microsoft and Amazon) think they have solid legal footing to do such deals in this way. The various regulators will (and already are with Microsoft and Amazon) take a look, but clearly these three companies, three of the largest companies in the world, with all the legal resources in the world, think this deal framework can pass muster.
But make no mistake, this is a hackquisition.
In a "normal" regulatory environment, Microsoft would have bought Inflection, Amazon would have bought Adept, and Google now would have bought Character. And actually, Microsoft probably would have bought OpenAI at some point already rather than just buying up the rights to the very precise on purpose 49% of the eventual profits. Which is effectively the same thing as buying 49% of the company. Which is effectively the same thing as buying 1% less of the company than would have made such a deal an obvious actual acquisition.
And actually, you may recall that Microsoft was really the one who pioneered this hackquisition model when OpenAI nearly imploded after its board ousted Sam Altman. Microsoft was set to bring over Altman and his co-founders alongside the rest of the team – again, by hiring them. Satya Nadella was able to put the pieces back together and keep OpenAI as a merely 49% owned entity, which was obviously preferable from a legal standpoint.1 But they decided to test their new hackquisition model just a few weeks later with Inflection. And we were off...
I do appreciate that each of these deals has a slightly different flavor of rationale from the startups' perspective. Inflection decided it was better to go down the enterprise path, not the consumer path. Adept decided to go down the "agentic" path rather than the AGI path. And now Character has decided they don't need their own models for their tech and will instead use third-party models. Each of these excuses are important because they provide cover for why the founders of these companies in particular are leaving to go to the totally not acquiring entity. They just want to continue the world they set out to do, you see. And the companies they've founded realized that was not the best path forward.
Well that and boatloads of cash helped. Per Kalley Huang, Natasha Mascarenhas, and Stephanie Palazzolo at The Information:
Character’s leaders told staff on Friday that investors would be bought out at a valuation of about $88 per share. That’s about 2.5 times the value of shares in Character’s 2023 Series A, which valued the company at $1 billion, they said.
And it's even more nuanced than that:
Unvested employee options will continue to vest at $88 per share through the end of July 2026; they will be paid via a fund with money from the licensing agreement, the leaders said. Options will continue to vest after two years, but there will no longer be a guaranteed payout from the fund.
One thing that I suspect will be looked at in each of these cases is if any of the tech is actually used via these licensing arrangements – i.e. what is actually being licensed here for all this money, and why? But I also suspect that Microsoft, Amazon, and Google will figure out a way to use some of the tech they're buying the rights to in some absolutely bare-minimum way. But is there some way to gauge an actual fair market value for the cost of these licenses? Or is this really expensive coffee? With this new deal, it seems like really, really expensive coffee.
As I ended my post a month ago:
I suspect this will not be the last "hackquisition" we see...
It took a whole month. But now who will be next? NVIDIA? They tried to make an actual acquisition recently and it was quickly jumped all over. Apple? Usually their deals are too small to garner regulatory interest, but this is a brave new world.2 My money is on Meta.
And if you put your ear to the ground and just listen for the whispers of which massively buzzed about and funded AI startup is actually burning too much money and has no real path to profits any time soon while the funding music starts to slow... you too can figure out who will be the next to be hackquired.
M.G. Siegler
M.G. Siegler
M.G. Siegler
1 This is why I was surprised when they took any sort of board role at OpenAI. But they clearly felt like they had to after such a debacle. But it's also why I was not at all surprised when they were out of the board room again after only a few months.
2 Also, if they took me up on the idea to buy HBO and/or all of Warner Bros Discovery, that would obviously be looked at. Hard. And I don't think the hackquisition framework would work for such a complex deal, sadly.