14 Comments
User's avatar
Kevin's avatar

Great article Kris! The capex spending seems obvious. They take more risk not spending than just spending. On the application winners, l do wonder if Palantir, Applovin and Shopify are not the outliers. The internet amplified the winner takes all model by introducing a network effect. The result, a power law. It seems that AI strengthens this power law. Maybe with fewer winners, but they'll win big and faster than ever.

Expand full comment
Kris's avatar

Hi Kevin

Thank you for your kind compliment. I really appreciate it.

As for your remark about the application winners, yes, for now these are definitely the outliers. And yeah, I also believe that the winners-take-most evolution will even accelerate. But, as Applovin showed, that can also be a newcomer. They will indeed win big and faster than ever.

Expand full comment
G Mamlet's avatar

Kris, I really think it would be worth pausing for a minute on Apple to hear some alternative perspectives. Apple needs to offer agentic AI to its users: AI that can perform useful tasks based on how much Apple knows about them. “Siri, buy another 100 shares of NVDA in my Roth IRA…”

The AI platforms to support this aren’t there yet but they are coming. We will see if Apple is ready by then or not.

Partnership is a fine way for them to start. Time-limited, with Apple maintaining control of the user and having full access to the data acquired. They preserve the option to replace the partner with their own technology if that becomes important. Meantime users are kept wrapped in the Apple cocoon with that nice tap into their wallet.

Perhaps you mean to suggest that if someone else succeeds in building generalized AI, then Apple will be screwed. Ok, yes, possibly.

But if you think there would be no deal feasible between that yet-to-be-named overlord of AI (as tight as the competition is right now, why do you think there would only be one player to reach that goal? Wouldn’t the next be weeks at most behind?) and Apple? I would disagree on a “buy vs build” calculus alone. Cheaper to do a deal with Apple and get all those users quickly than to try to woo them over to your ecosystem/device that’s every bit as fully built out as Apple’s.

I get your impatience with Apple. The first generation iPhones would be old enough to be in college, for chrissakes, and Apple has had nothing major to show since then. But there are multiple ways to play this out, and for Apple in particular, I think there’s a good shot at using the iPhone franchise to pull themselves into the AI era with much less sweat and more grace than the other MAG7.

Expand full comment
Kris's avatar

Thank you for your interesting comment!

I actually 100% agree with your comment about Apple, which has a fantastic platform to leverage AI. There are very few companies in the world that know its users so well. That's why I think partnerships are not the way to go. Others get the insights from Apple's data and it could be that company that screws Apple. Just like Microsoft did with IBM. To me, that's a significant danger for Apple.

Expand full comment
G Mamlet's avatar

Kris, have you ever run a business yourself? Have you ever had to negotiate and sign complex agreements with landlords, investors, partners, distributors, …?

In case you haven’t, let me tell you about the time my business partner discovered the hidden “your intellectual property will be ours” clause buried in a thick contract from a F500 company: “Notwithstanding anything else in this agreement, any intellectual property disclosed to us by you will become fully ours as well if we can understand it and describe it.”

Or the time my 2 by nothing company shared sensitive info with another F500 company under NDA, and they then told us, “we’re ripping up the NDA and will take your IP to our friends to build into their product to compete with you. You can sue us if you want to light money on fire.” That IP formed the basis for a major division of an F500 company.

The business world is full of these stories. In my experience, it’s the IBMs who screw over the weakling Microsofts, so let’s look at what happened there.

IBM couldn’t imagine a future in which the PC became a meaningful part of computer demand. They never thought to require that MSDOS only run on IBM PCs. They never thought to require Intel to only sell the 8086 to them for use in a PC. They never thought it would matter.

Microsoft had a different idea of what was possible, and they made it happen.

Coming back to the present day, we aren’t talking about a David vs Goliath situation here. We are talking about two Goliaths negotiating and contracting with each other to make a deal that is beneficial to both sides. They do it all the time without creating the kind of catastrophe you’re imagining.

If anything, Apple will have the upper hand here: there are almost certainly going to be several eager partners bringing AI technology forward for it, so it can play one off against the others.

Expand full comment
Kris's avatar

I run my own business, and my wife's a lawyer and tax specialist who was Chief Legal Officer in a multinational company. So, yeah, I'm familiar with that world.

Let me take your own words, but change them a bit:

Apple couldn’t imagine a future in which AI software became a meaningful part of demand. They never thought to build their own model. They never thought it would matter.

Of course, Apple can get any partner it wants. But that's my point: that exactly is the danger. They can be disrupted from within. That has not been the case with anything else, because no revolution has been as big as AI and LLMs since mobile. When I first tried out ChatGPT, I immediately understood (and wrote about it) that this was a huge sea change. I said it was the third big wave I lived through in my adult life, after the internet and mobile. And Apple's saying: "We'll let someone else do that for us?" That really gives me the IBM vibes of missing how important this is.

Of course, the future is inherently unknown, and eventually Apple could be a winner from AI, but I just point out that it's much riskier not to invest in AI than to invest in AI.

Expand full comment
Brian McCormick's avatar

Good article. The AI capex spend must continue.

Expand full comment
Kris's avatar

Thanks, Brian!

Expand full comment
David's avatar

I can't help but ask: what might be the reasons behind Apple's behavior? Often they don't follow the "hit first" approach, but the "hit best" one. Could it be the case in relation to AI?

Expand full comment
Kris's avatar

I think that's the case but I also think that this time, that could come to bite them. If the paradigm shifts, you can't react in the same manner than with other developments.

Expand full comment
Emerging Value's avatar

Well, AI is a commodity business. Apple is right in not going into it. They will be able to partner with a company with a model to provide experiences for apple users. Also, its just just a phone. Apple is not into complex business to business use cases.

As far as the cape trends, I think its a bubble similar to previous bubble. AI is meant to become more efficient and at some point maybe the old servers are just too outdated versus new servers. Its like buying a ton of 2008 computers. Cool in 2008 but in 2015 you are left with a large outdated asset Base.

Expand full comment
Kris's avatar

I disagree for a few reasons:

First, AI is not a commodity business if applied well. I provided examples of Palantir and AppLovin. Meta is using it to have better ads, Google as well, Amazon for better recommendations, etc. And Apple? Crickets.

Secondly, just by renting out AI capacity, the hyperscalers earn more money than they invest.

Third, if Apple has to partner, it means they lose control. It's not in their own hands. Apple's strength has always been the tight ecosystem, where it controls everything. Steve Jobs was obsessed with that thought. If they partner, they throw that Jobs paradigm under the train they should be on.

Not sure what you mean by the remark about old servers.

Expand full comment
Emerging Value's avatar

AI models are commodities. Companies that integrate their data with Ai are not AI first companies. They are data first companies integrating AI. AI is something everyone can do, students or startups know how to use AI, but they dont have the data, the customer base, or the software around it.

Apple can get many partners for AI models in a second. They can even get some open source models free. They use google as a partner for search and search is much more concentrated than AI.

And capex is profitable for now, we dont know in the future.

Expand full comment
Kris's avatar

I think you should make a distinction between AI and LLMs, which you don't seem to make. That distinction is crucial. And of course, it's about the data, but without AI, those are often useless.

I didn't say that Apple can't get partners. Of course, it can. But if they do that, they will make the same mistake IBM made with Microsoft decades ago.

Look, this is not an Apple article, so I don't feel like discussing too long about this. It's just a warning that they will have to act or they will become the Microsoft of 2000-2015, especially at this rich valuation.

Expand full comment