Artificial Intelligence: Are we bubbly yet?

Photo by Derek Story on Unsplash

When we first started playing around with Chat GPT late last year, we had an inkling that this was one of those applications that would capture the imagination. Sure, the base code had been around for a while, and a small set of users also had access to the rest of the OpenAI suite, including image generation tools like Dall-E.

Fast forward a couple of months and the timeline is a familiar one: first, the niche; then it gets on some segments of social media; then it goes public – the kids start talking about it, then some parents, then ALL parents, then it hits the financial media, then the mass media and then popularity grows exponentially to the point where servers need to be rate limited to prevent the entire application from grinding to a halt.

The cynic in our heads would immediately recognise that as the same trajectory the Metaverse narrative took. And in many ways, ChatGPT is to AI what Axie Infinity was to the Metaverse concept. Both weren’t the first, and both came at a time when their respectively underlying concepts weren’t exactly new – the metaverse wasn’t a brand new concept, and likewise Artificial Intelligence isn’t something we’ve just thought up. But where their forebears failed to catch the narrative wave, they succeeded and became the poster children for the mania that ensued.

Of course, we all know how the Metaverse bubble ended – we even wrote explicitly about it being a bubble back in Nov 2021. Yet, that didn’t stop us from making some good money from that trade – the market and reality don’t always need to match up, and they rarely do.

Ultimately, the prospect of staring into another bubble – one that potentially could eclipse the prior ones we’ve seen – is enticing. The question that needs to be answered is in fact not “is this a bubble”, rather it’s “can this bubble endure and for how long?”

We’d surmise that the answer is found in first revisiting the ingredients for a bubble.

Cooking up a storm

At its very base, getting a bubble going requires our favourite ingredient: narratives. A compelling story – not just about doing things better, faster, cheaper. It needs much more – a blue sky storyline that promises fundamental change to the way the world works, often latching onto an aspect of the zeitgeist that is poignantly emotional.

Think back to the days of Axie Infinity and “Play to earn”, and the narrative that accompanied it: Filipinos who were unable to travel and work due to the pandemic were able to make a living playing a game. Playing and earning, the complete opposite of what our parents used to tell us, that playing games was a waste of time and money, and that homework should be finished first. The resonance of that narrative: villain (work) and hero (games), and the prospect of a fairy tale ending (everyone happily playing games and making money from home) echoed through the market.

By any measure, the Metaverse story spread relatively far and wide: from the Philippines, it went all the way to Facebook changing its name and ticker to Meta.

So one can easily imagine how Artificial Intelligence, as a technology that has arguably even greater (theoretical) impact could surpass the imagination impact of the Metaverse.

The story probably isn’t yet fully formed, but the characters are in place. ChatGPT, presented in its chatbot form, made AI a personable counterpart that the masses could relate to. Children are using it to do homework, teachers are finding ways to counter ChatGPT produced homework; devs are using it to write and troubleshoot code segments, while others are using role play on ChatGPT to turn it into its evil twin, DAN.

This is a plot summary that belongs to some science fiction novel from the 80s or 90s, and just as everyone had given up on the promises of AI made in those early years, they’re starting to come true.

Sure, there are also accompanying issues of copyright infringement claims, royalties and attribution matters and regulation, but even these “challenges” are being seen as the “establishment” fighting back: the luddites who are pushing against new technology to protect their soon-obsolete livelihoods.

It’s all fitting into a nice storyline that, as it happens, we are already VERY familiar with.

AI will take over all the boring work that needs to be done, leaving humanity emancipated and free to pursue leisure without limits.

And yes, AI can even tell you what to cook given the ingredients in your fridge.

Talk about cooking up a storm.

A splash of nitroglycerine

This recipe card, very much like the metaverse one, features an extra ingredient: VC nitroglycerine.

It’s easy to write VCs off especially after the terrible year they’ve had, from metaverse startups to crypto projects that have cratered to even the not-so-small names that have dominated our headlines.

But they’re still around, and according to Pitchbook, there’s still about $585bn of dry powder sitting around, and that dry powder’s staying dry – looking for deals on increasingly investor-friendly terms. All said and done, unallocated dry powder is anathema for any fund manager: capital needs to be put to work, but where?

The combination of capital seeking an investor-friendly home (i.e. target companies that have relatively weaker bargaining power in a negotiation of terms) and a need to deploy A LOT of money before it gets redeemed makes for nitroglycerine when it comes to fuelling bubbles.

Once more, a quick glance at the Metaverse plot line provides hints as to where this capital might go: the big projects like Meta’s Horizon Worlds and Oculus typically don’t need VC money, and even if they did, they are leaders in the space and can dictate terms. In this case, the analogue would be OpenAI: the company behind ChatGPT, partnered up with Microsoft, but in pole position in this race.

This isn’t VC territory.

In order to get the friendly terms they need, VCs need to go for the challengers: the next OpenAI, the OpenAI of __________, the _________ of AI, OpenAI but different because ___________. Insert anything into those blanks and start looking around, and chances are there will be some startup that fits the description.

Even better if it’s in crypto: add “decentralised” to any of those descriptions, toss in a handful of token economic plans and spin up a discord group with 20k members, including a sub-channel called “moonbois” and suddenly there’s a community that is captive and can be marketed to.

Piece it all together, spin up a “roadmap”, click and drag on excel and on the other side pops out an investment case for a potential 20 bagger.

There couldn’t be a better recipe for a bubble. And just like with the Metaverse – if there’s a bubble in stocks, there’s a mega-bubble in crypto.

The Soros Line

Cynical as we may sound, we’re actually not – long-time readers would recognise how despite all our cynicism around crypto (and our long-standing view that 99% of tokens that currently exist are worth zero, even if they may not trade at zero), we are constantly on the lookout for new developments in the space, eagerly awaiting the day it finds proper product-market fit.

The same can be said of our views of the Metaverse and of Artificial Intelligence. They may not work at the first attempt, or even the second, or third, but sometime in the future they will.

But as we’ve said above, and will continue to insist, the fundamental reality doesn’t need to line up with the market.

Things are never truly black and white, and they come in a spectrum: on the AI front, there are businesses with applications that have been literally battle tested, albeit lacking in the sexy personal touch – some are listed, like Palantir and C3.ai, even if trying to analyse them has historically proven to be pretty much a black box, coming with the territory of US government contracts.

These aren’t limited to US names – Chinese companies like iFlytek, Sensetime and even Baidu have been building on AI for years now. Their narratives have been around for years, but no one really bothered until the zeitgeist caught up. Now could be their time to shine – certainly at the minimum for their stock prices.

And then there are the heroes that capture the public imagination, like OpenAI, providing the skeleton around which the entire story can develop. Private or public alike, they are the protagonists of our story.

And then there are the challengers: some may have valid technologies that simply need more work, some are trading up just because somewhere in their name, URL or website text it says “AI”. Chances are they’re very early stage – maybe they come good in the end, maybe they’re just vapourware, who knows.

But it is in this last group that the bubbling is going to be most pronounced.

This much is clear, though: the number of people calling this a bubble and deciding to stay out suggests that this bubble could well have some way to go.

Where one decides to apply the principle of that famous Soros quote regarding bubbles is then simply a function of risk appetite.