Blain’s Morning Porridge 31st July 2024 – Pop goes the AI Bubble.

 “Half a pound of tuppenny rice, half a pound of treacle, that’s the way the money goes, Pop goes the Weasel.”

There was a Judder Moment in AI last night. Microsoft’s numbers confirm the hype is played out – the reality is the AI everything rally is bleeding revenues and momentum. AI could see a cascading failure event bringing down startups, and triggering a reassessment of AI linked firms. Tesla – as the most hollow set of undelivered promises – is set to feel most pain.

But First: Bill’s Pub to Club Swim: Let me remind readers my Bill Blain’s Pub to Club Swim Charity event is open in support of Wessex Heartbeat, the cardiac care charity.. You don’t have to anything except click the link – here – and make a donation. Without Heartbeat, well, you would not be reading the Morning Porridge this morning.

AI Bubble? 

As the stock market wakes up to a thumping great Microsoft and AMD AI hangover this morning, a whole series of threads are coming together in the AI Happy Space suggesting a Judder Moment has occurred – when the tectonic plates supporting a widely accepted thesis suddenly shift and wobble, exposing the underlying weakness and limits of the idea. Generative AI has been the biggest new, new thang in years, and has triggered an incredible wave of investment into new AI infrastructure: chips, servers, datacentres and energy security, all of which have multiple issues. That sound that woke you up in the middle of the night… was the pop of a bursting BUBBLE.

Everyone is acutely aware the whole mood of the market has been set by Nvidia – the dominant cog in the AI Chip supply chain – and, basically, the entire stock market these last 12 months. The simplistic read was: without Nvidia chips the AI revolution could not happen. It rallied 800% in the 18 months from Dec 2022 to its peak a month ago. Rival chip maker Intel’s stock rose last night – because its slashing costs and jobs. Samsung is up because is now selling AI chips. Nvidia is down 20% since June. BUBBLE.

Back in early June I pieced together some date from Pitchbook on Unicorns – Of Unicorns and Real Asset Values. (Unicorns are start-ups worth more than $1 bln.) Since 2016, nearly 1500 new firms survive as Unicorns today. The pace of Unicorn Formation has accelerated since the Generative AI boom began in 2023 – according to Edge Delta there were already 214 AI Unicorns in March this year. That number is even higher today as the way to get rich is to call everything and anything AI based. (Would you like to buy some of my new AI generated Gin?)  BUBBLE.

As we’re all aware, at the core of the AI Everything Boom is OpenAI, which not only now intends to compete with Google in search engines – but loses $5 bln per annum while receiving revenues of less than $3 bln. From my perspective as an old bond dog, that doesn’t sound a particularly sustainable proposition – but apparently equity investors think that’s exceedingly normal. ChatGPT, and the multiple iterations and copies have repeatedly proved they aren’t clever, smart or particularly intelligent. It hasn’t been the productivity game changer we were told to expect. The Generative AI models crunch numbers (at enormous energy cost), to produce is the most plausible string of words as dumb answers to lazy questions. BUBBLE.

Last night it was Microsoft in the headlines. The stock crashed after the earnings report – its Azure cloud unit saw revenues miss (29% up against 30% expectation), panicking investors the momentum of the AI everything rally might be slipping. The real issue is how much Microsoft is spending on its AI infrastructure – chips, servers, datacentres and issues like energy security. (Wonder why Bill Gates is spending money on innovating small modular nuclear power modules?) Its bought nearly 2 million chips and, while the numbers are contradictory, but it’s spending somewhere between $10-15 bln building or leasing data-centres to service its AI ambitions. BUBBLE.

It’s not just Microsoft spending big. Alphabet is also in the spotlight for its AI spend. The rest of this week will focus on just how much the rest of the Mega 7 have blown on AI. The answer is they are all desperately seeking to associate and align their products, businesses and image with the new, new thang – AI. Since I barely use much of the functionality on my iPhone.. do I really care if its AI enabled…? BUBBLE.

AI has been great for some stocks. We know that road transport is incredibly complex – if we can’t even make simple point-to-point railways work, how will AI self-driving on roads work? The only factor holding up Tesla’s battered and questionable stock valuation has been Musk promoting the lie AI can match or even better human intelligence. While he has been telling us for years Full Self Driving will be “jam tomorrow” – he knows his camera tech is pants, he knows it will never really work safely, and is very aware Generative AI is just processing power with no innate real intelligence. But who is the guy telling us AI General Intelligence is just around the corner? Musk. Why? Because he knows the safest way to safely drive a car is a human at the wheel. Commercially, he can’t admit or say that. Musk is telling us the porky that AI (which doesn’t work) is safer and better. He says it again and again till poor fools like Cathie Wood actually believe him and parrot him. Tesla’s claims to enormous value from FSD and RoboTaxis is bunkum and a complete BUBBLE.

Could we see a contagion effect spread through the AI driven markets? Many AI startups have financed their purchases of Nvidia chips and specialist servers from the alternative private credit markets. These funders have security on the chips, and in many cases will have negotiated equity kickers and profit shares on the revenues they generate, but these chips a) become less scarce when AI start-ups start failing, so prices will drop, and b) the chip makers juice their revenues by ensuring this year’s chip is obsolete the next. These startups are also the ones that have been driving the cloud revenues at Microsoft and Google – when they start dropping then cloud revenues could tumble – becoming a cascading landslide of AI crises. BUBBLE.

Just over a week ago I was wondering about how the consequences of the CrowdStrike crash 11 days ago could further change the Tech and AI landscape. It may force CTOs to reconsider stability and security around their global tech infrastructure. If an AI bubble pops, how do they ensure their IR remains resilient if the tech ecosystem is in crisis? How vulnerable are their own firms if IT suppliers are in trouble? The mega techs may find they have lot of surplus data-centre capacity if the customer firms start re-onboarding servers and data.

Watch the IT/AI space… it could get messy quite quickly…

Bill Blain

CEO and Founder,

Wind Shift Capital

www.windshift.capital

 

Blain’s Pub to Club Swim.

One Comment

  1. Bill Blain July 31, 2024 at 5:20 pm

    From a posting on Linked in by Pete Jarvis:

    You know I always appreciate Bill Blain’s thoughts; his recent newsletter reminded me of a wonderful New Yorker cartoon, “What Lemming’s Believe.” it is a wonderful piece of art that fits well with Bill’s recent newsletter. I do not always agree with his thinking, but I always appreciate his writing and thinking – a recommended subscription.

    My personal sense is that Meta and Google are terrified of what OpenAI and Microsoft are doing due to the potential impact on additional revenues (though I feel Meta and Google’s reactions will eventually accelerate revenue decline as everyone will ultimately be shouting, which will mean no one will listen) – their response has been a massive investment in LLM related technology, as well as a push to give the models away (in my view) to take the oxygen out of the room for any other aspiring AI startups.

    Consider the parallels with Zoom. Zoom charged for video, Google’s response was to offer free video for 30 minutes, Microsoft made similiar moves with teams which they bundled in office. Meta is playing the same game with its LLM models. New entrants cannot compete with free; Meta, Google, and others will simply wait and mop up everything after the money is spent; hence, spending is a tad high right now.

    My personal sense is that Microsoft is very well-positioned, and I’ve never sold my stock. Why? It is a well-run and ecologically diverse company. Google and Meta are not ecologically diverse from a revenue perspective. This means that they have to fight and kill to maintain their ecological/ market niche. Any change in the environment will cause their costs to consume the host. A.K.A BlackBerry versus the iPhone. The difference per unit was a $149 dollar BES license versus no BES license for the iPhone. For a bank, that’s a $500,000 difference in license renewal cost. BlackBerry’s costs killed them, not the iPhone – specifically their Management costs.

    This reminds me of R. H. Donnelley (RHD), Yellow Pages; more than half a million advertisers paid $3,500 a year for space in the Yellow Pages as recently as 2006, and total circulation reached 80 million copies. The RHD CEO’s presentation to investors in March 2006 was titled “Positioned for Continued Success.” Just three years later, in May 2009, R. H. Donnelley went bankrupt.

    That, to me, is why Meta and Google are spending money from a gushing fire hydrant – they have one revenue source – advertising. They must kill to maintain that revenue stream; they have none to replace it. To be clear, this neither good nor bad, it simply is.

    What Lemmings believe by Robert Mankoff.
    https://lnkd.in/g-XhWFZg

Comments are closed.