An AI Rocket (Reddit) & An AI Failure (Inflection AI): What Are The Lessons From Both These Case Studies In AI Outcomes
We had two interesting market headlines this week:
The first was a red hot tech initial public offering or IPO.
The second was a once equally hot AI startup selling itself for parts.
The IPO was Reddit's, a website that hosts millions of online forums. If you aren't very familiar you might remember that Reddit hosted the "Wall Street Bets" forum or subreddit that caused so much excitement (and pain) during the peak of the Covid-lockdown stock buying frenzy.
You can read our old coverage of that fun here.
Reddit is a cool company and has a very passionate user base. It was also famously founded just up the street from Pebble HQ in a tiny apartment by two college roommates. It is therefore a classic US startup tale made good which is both super impressive and a nice story to boot.
These days Reddit's listing on the New York Stock Exchange was highly anticipated for investors - retail and otherwise - excited to get back to the days of hotly hyped IPOs and soaring valuations.
The stock delivered. Reddit trading under the ticker RDDT soared as much as 70% above its IPO price and closed the day nearly 50% higher. An intriguing business model, a passionate retail investor fanbase and, perhaps most of all, a very constructive macroeconomic backdrop were enough to create maximum excitement around a brand new technology stock.
One of the interesting wrinkles with Reddit in particular is the idea that, surprise, surprise, they are keen to play up their artificial intelligence potential. With nearly 20 years of highly specific, user-generated content on millions of forums, the idea is that Reddit has legal ownership perfect raw material to train extremely powerful large language models. The posts are even ordered for relevance!
That is Reddit's story but it is by no means alone. Slowly but surely the IPO market is coming back from the dead. Reddit wasn't even the hottest IPO of the week. Another AI startup called Astera Labs closed the day 72% above its IPO price. You know good times are here again in the tech ecosystem and the stock market when companies you have never heard of are setting records.
As the kids say: we are so back!
The only unpleasant part of this happy tale, of course, is the fact that Reddit isn't YET making billions from AI......
That is something to remember if you are buying RDDT stock and it is also where the other headline comes in.
It got a lot less press this week but:
A very hot AI startup was quietly absorbed by Microsoft.
We say "absorbed" because we aren't entirely sure how else to put it. Technically Inflection AI wasn't acquired. Rather, Microsoft purchased the non-exclusive license to its intellectual property as well as hiring most of its key employees for a cool $650 million.
That is a huge sum but it is also quite a step down from the $4 billion dollar valuation that Inflection was raising money at just a few months ago.
$4 billion to $650 million in 6 months is something to remember before you start thinking that every hot AI venture's future is paved with gold.
There are a few interesting things to say here:
The first is that this is a novel - and cunning - attempt to get around the obvious complaints that Microsoft is using its wealth and compute power to acquire all the critical start ups.
The second and far more important one is that, for all intents and purposes, this was an acqui-hire by Microsoft. By this we mean that Microsoft is acquiring Inflection more for the parts - including its people - than for the product. Hence "acqui-hire" from acquisition-to-hire.
Third, this gets back to something important we have pointed out before which is the best strategy for the AI mania is in owning the shovel makers, not the gold miners.
The unstated and nasty takeaway from this tale is that Inflection raised a ton of money from investors and delivered very little in return. Incredibly, however, all of these investors are apparently going to be making at least some profits.
The information we have is that Investors in the early $225 million round will be getting 1.5 times their investment while those in the later $1.3 billion round will get 1.1 times their investment, according to this article by The Information.
Those are slim returns for investors who no doubt hoped to make many times their investment. They are lucky to get it, however. Another company with less connected investors might easily have been sold for a less advantageous price or simply went to 0.
Nothing new there. The rich and powerful using their wealth and power to come out ahead is a tale as old as time. What is different about the Inflection AI story is who came out most ahead. Hint: It isn't Microsoft or Inflection's employees.
No, the entity who made the most money from Inflection was Nvidia (also a shareholder apparently!). The AI startup had raised over a billion dollars of capital on a $4 billion dollar valuation and spent most of it on - you guessed it - Nvidia GPUs.
22,000 of them in fact which, no matter the class of GPU, easily cost hundreds of millions of dollars. That shocking expenditure is becoming table stakes to compete in the top echelon of AI companies which is why there are so many fears around only the deepest pockets being able to enter into this space.
That may still be true but this shows that striking AI gold isn't simply a matter of a few billion and some very talented individuals. It takes serious commitment, a fine tuned business sense and also, most likely, a healthy dose of luck to find the AI motherlode.
So, as we have argued from our very first piece on Nvidia early last year. Picking winners in a gold mining contest is very difficult. Inflection is a perfect and cautionary example here.
They had very prestigious backers (Bill Gates, Google's Eric Schmidt etc), vast amounts of capital (well over $1 billion in a 2 year span), and incredible founding team (including Deepmind's Mustafa Suleyman) and yet the inescapable lesson is they only narrowly avoided flaming out entirely.
This is why you have those big investors with powerful connections. It is hardly a surprise that Microsoft isn't opposed to making Bill Gates whole but all the good will and talent and the world haven't stopped Inflection from blowing a ton of money only to be acquired overnight to save fade.
In the end Inflection may have been a terrible investment for its investors but it was a great outcome for Nvidia's shareholders though. They are thrilled with this outcome.
That is the lesson here. Bill Gates should have just bought Nvidia.
*******
Have questions? Care to find out more? Feel free to Download our App (!!) or reach out at contact@pebble.finance or join our Slack community to meet more like-minded individuals and see what we are talking about today. All are welcome. You can also get our newsletter as an RSS feed.
