Apple Weighs the Voting Machine

In the short run, the market is a voting machine but in the long run, it is a weighing machine.

If you've worked in finance or any tangential industry, you know this quote from Benjamin Graham. Warren Buffett cites it often. Is it mis-quoted? It doesn't matter because it's true. But it's a great quote because while it's easy to understand, it's incredibly hard to adhere to. Both if you're an investor and if you're a company.

My mind went to this quote when reading the recent report from Aaron Tilley, Wayne Ma, and Valida Pau for The Information about how "Apple’s Aversion to Big Deals Could Thwart Its AI Push". The key high level points aren't necessarily new – there's a ongoing debate inside of Apple if they should do a big AI deal to try to help them "catch up" in the race – but some of the details are. And it paints a pretty clear picture of why this is such a debate.

Everyone knows about Apple's trials and tribulations when it comes to AI. But it's also not entirely clear that they would be able to fix such issues simply by making an acquisition. Further, making a big one – the report cites interest in both Mistral and Perplexity; again both known, but now with new details – could actually backfire in a major way. Why? Because Apple's famous internal culture is quite different from that of other cultures – and certainly of most startups. If most of those have gone to the Zuckerbergian school of "move fast and break things", Apple is basically the opposite.

Call it: move methodically and make sure everything is perfect.

This has come back to bite Apple a few times in the past, but in the current Age of AI, we're seeing some real outward signs of tensions and challenges because the entire industry is now moving at light speed. I'd argue – and have, many times – that at the highest level, that's why Apple has fumbled the bag so badly with AI. Right now at least, it's an industry that requires speed and experimentation.

Oh yes, and money. A shit ton of money. Something else which Apple has historically struggled with – not making it, the opposite: spending it.

Case in point: as their Big Tech peers rocket past $100B a year in CapEx spend, Apple holds slow and steady between one-fifth and one-tenth of that. That's obviously mostly because Apple isn't building and buying data centers to train large language models, and in fact, even most of their cloud infrastructure is outsourced. But there have certainly been factions inside the company over the past couple of years that have seemingly been begging them to buy up some NVIDIA GPUs and get going. The CFO said "nope" and Tim Cook agreed.

Maybe that was a mistake – certainly given Apple's current predicament in AI, it seems like it may have been. But that's also the short-term vantage point. There is a world, perhaps a few years from now, where that lack of massive CapEx spend seems smart – and is rewarded by Wall Street.

There's also the opposite case to be made, of course. If AI truly does transform everything and Apple is left building on others' technology – something they historically hate. But I'm just saying there's a case that they end up on the right side of history, at least with regard to these early foundation model build outs.

And tangential to that, there's a case to be made that they actually shouldn't do a big acquisition in AI. I've long been laying out the plan to make such a deal, listing the obvious targets – including, of course, the two listed above – but I do think it's worth thinking about if Apple's own internal hesitation here is warranted.

Apple's largest deal to date remains Beats, done over a decade ago for $3B. That is less than many of these AI companies now pull in within single rounds – not the valuation, the amount raised. OpenAI is in the midst of raising an order of magnitude more than that. At more than two orders of magnitude the valuation. So yeah, that seems like a mental hurdle the company would have to get over to do a big deal in AI. Cook is saying they're not averse to it, but they clearly are, per the report.

And again, perhaps for good reason. While from the outside we may see the Beats deal as a success – the brand is still around, giving Apple an outlet to experiment in some hardware (and accessories) outside of their own pristine branding. But that was only a piece of why Apple bought Beats – another deal that was being pushed by Eddy Cue, just as these current crop of AI deals apparently are:

In another instance, though, Cue successfully pushed Apple to do a deal that was outside its comfort zone. In 2014, at a moment when Apple was falling behind Spotify in the music business, Cue got Cook to agree to buy Beats, despite skepticism from Apple’s old guard about the $3 billion deal. Cue’s plan with Beats was partly to bring about a cultural shift in Apple’s music business, people familiar with his thinking said.

But integrating Beats with Apple proved difficult. Apple staffers viewed the Beats engineering team responsible for its streaming music service as below Apple’s standards, according to former employees who worked on Apple Music. Apple ended up building its own streaming music service, Apple Music, from scratch rather than relying on Beats’ software. Beats’ headphones business has been more successful for Apple.

Buying Beats to "bring about a cultural shift" within Apple is similar to an argument I've been making as to why buying a big AI company might make sense. And while bringing in people like Jimmy Iovine and Dr. Dre clearly did have an impact internally, it's not entirely clear that it was a good one. And the fact that they couldn't convert the Beats software into the cloud-based iTunes also doesn't speak well to any AI product Apple would buy here. Again, it's just a culture that is very insular and likes to do things their way, from the ground up and top down.

The most successful bit of M&A Apple has done (beyond buying NeXT, of course) is likely a smaller, more low-key deal: PA Semi. That deal, done 17 years ago, laid the foundation for Apple's move to build their own chips, with many of the employees brought over laying the groundwork for Apple Silicon. So clearly, that would seem to be the type of deal that Apple would want to do in AI. But it's not clear that Apple could do such a deal in this space because it's so red-hot with investors that any company of that pedigree will have already been bid up well into the billions.

In hindsight, Apple probably would have loved to have done the DeepMind deal. That deal was done right before the Beats deal at nearly 1/10th the price. Good luck finding that deal in 2025. DeepMind today would undoubtedly be worth hundreds of billions of dollars as a stand-alone company.¹

Buying Perplexity, which just raised at an $18B valuation, would probably cost at least double that, in order to convince them to sell. Mistral is valued slightly lower, apparently now raising at $10B. Still, assuming this round gets done, the price tag would be at least $20B, and probably higher given the money will have just been raised, and presuming there would be other bidders around the table. So we're likely looking at a big AI deal that would be an order of magnitude larger than any deal Apple has ever done. Certainly Apple has the money to do it, but these aren't just turnkey-and-get-caught-up-in-AI deals.

Perplexity, potential cultural issues aside, would undoubtedly require a lot of work from a technical perspective to become an Apple product. And that's if none of their current partners balked at working with an Apple-owned company.

Mistral, which unlike Perplexity, is actually building their own frontier models, would undoubtedly require billions more in ongoing spend from Apple to keep doing what they're doing. You could say goodbye to that small CapEx number.

With all that in mind, there would certainly seem to be a case to be made that Apple would be better off continuing down their current path – not exactly going it alone, but continuing to partner with various AI players to get their products up to speed while they work on their own technology behind-the-scenes. Again, the hesitation to that strategy is first and foremost that it hasn't worked yet. And related, that Apple may not have the right mentality, culturally, for it to ever work in the AI space. At the same time, there's no guarantee that a big acquisition would change any of that either – and it may make things worse.

Beyond Beats, the Information piece goes into some other troubled past acquisitions such as LuxVue and even Siri itself. And none of this even speaks to the potential regulatory challenges Apple may face in trying to do a deal. Would Apple dare do a "hackquisition" to get around such concerns?

But there's another layer to all of this that Cook and Apple must be weighing: the optics. Wall Street clearly wants to see Apple get back into the AI race and you could imagine that if a deal for Perplexity or Mistral or the like was announced tomorrow, Apple's market cap would immediately shoot past whatever the cost of the deal was. That's not a reason to do the deal, but it has to be a part of an equation. Apple needs to be seen as taking this challenge seriously – all the more so if the Google Search deal goes away – and such a deal would clearly send a big signal.

It would be Apple focusing on the voting machine and not the weighing machine. Which sounds decidedly short-term, but there are long-term benefits to that mentality too. All of this Apple must weigh. And then vote.

¹ That alone probably makes Google look awfully undervalued by the market.