Serious About Computing? You Should Build Your Own AI.
If I had to pick an overarching theme for this month's appearance on the Big Technology Podcast,1 it would probably be "one step ahead". That is, nearly everything Alex Kantrowitz and I discussed to start the week was right back in the news in a meaningful way by the end of the week.2 Still, I think the points discussed are good ones – dare I say, prescient?
OpenAI Drama vs. Anthropic Surging
We kicked off talking about the latest drama at OpenAI – a potential rift between CEO Sam Altman and CFO Sarah Friar over the timing of their would-be IPO.3 One of my predictions heading into 2026 is that the company wouldn't go out this year – there are simply too many factors, including macro, that are unknowns. That's obviously more true today with a war happening overseas and the sudden surge in Anthropic's business.
That's something else Alex and I discussed before the numbers later in the week suggested that Anthropic may actually be ahead of OpenAI in terms of ARR now (yes, they measure them differently, but still, there's no denying the trend lines here). Is OpenAI's growth slowing while their main rival grows faster than perhaps any company in history? Regardless, OpenAI is clearly in the midst of a major shift on both the product and business fronts and so it's going to be hard, if not impossible, to model out where the business will be in a few months – which is when they'd have to be filing if they want to go out this year.
And with SpaceX having now already filed to go public, likely in June, that will steal some of the pent-up demand for an AI play in the public markets. (Though, to be clear, xAI is clearly not OpenAI or Anthropic in terms of both usage and business.)
Look, OpenAI just raised $122B – not only the largest private round of financing ever, but even larger than the amount SpaceX is said to be targeting to raise in that IPO. That plus the executive turnover/leaves of absence would undoubtedly make anyone put a pause on going public. Then again, OpenAI undoubtedly has to at some point. Because even that $122B isn't going to be enough to get them to profitability! And while they can undoubtedly find more money privately, we're also clearly near the upper bounds. And this may be a unique moment in time to go public – certainly if the macro picture starts to turn. And yes, they'd obviously want to beat Anthropic out to market given those comps.
I'm talking both sides here. I just wouldn't be shocked if they didn't list in 2026, and I'm not sure they should, but they might feel like they have to.
Lastly, we discuss if the pivot to the "Super App" strategy is the right one. I tend to think it's sound to combat Anthropic – leverage your strength, which is ChatGPT's massive user base – but there's real risk in terms of complexity. And while Claude is surging, we talked about how their lack of infrastructure spend relative to OpenAI – which is obviously what makes Anthropic's bottom line numbers look better – may be about to bite them in the ass, causing them to scramble for new data center deals. Which again, by week's end sure seems like it was happening!
Siri To Suck Less?
From here we discussed if Apple truly is about to fix Siri. The move to create a stand-alone app certainly seems like the right one now given the table stakes laid out there by ChatGPT, Claude, and the like. And it's seemingly a good sign that Apple is recognizing this now and not a year from now after another failed AI strategy.
I'm feeling more optimistic that Apple is going to get Siri right this time, but I've been getting fooled on that front for 15 years running, so... Alex brings up a good point that it's interesting how Apple is finally acknowledging the AI chatbot reality just as OpenAI, Anthropic, and others are moving on to these more all-encompassing AI suites of services. Is Apple going to be behind yet again?
Maybe. It's still obviously very early on the "agentic" front. But still, there are also advantages to being a first-mover. Just as there has been with AI overall. And Apple, famously, doesn't typically strive to be the first-mover (though ironically they were first with Siri!), which could continue to hurt their chances in AI.
While they may look smart at some point for not pouring hundreds of billions into CapEx spend, that could come back to bite them in ways that are more tangential. Including, culturally, if the DNA of the company is never rewired to operate in the Age of AI. "People who are really serious about software should make their own hardware," Alan Kay famously declared in 1982. What if the modern day version is something like: "People who are really serious about computing should make their own AI"?
That leads directly into a discussion prompted by my purposefully provocative piece wondering if Apple should buy Anthropic. Sure, it's not going to happen, but that doesn't mean it shouldn't – perhaps for both sides. Yes, even at or near the trillion-dollar price tags. It would immediately reboot Apple to be ready for the Age of AI, and it would give Anthropic all the resources they need – and protection from the government seeking to destroy them. Again, it's not going to happen. But it's not completely crazy!
Google has DeepMind. Microsoft had OpenAI, and now is building their own in-house frontier models. Meta bought Scale to do the same. Amazon has a huge stake in Anthropic, but is also now building in-house. Etc. Apple is a glaring outlier here, which is either going to be insanely prudent or completely catastrophic.
Meta's Misses
Speaking of Meta, we talk through the notion that they keep missing on their big picture initiatives. The metaverse, crypto, encryption, etc. The first swing at AI was clearly a miss despite Mark Zuckerberg having some level of vision for where computing was heading as he tried to buy DeepMind before Google did, and as a result, brought Yann LeCun in-house at Meta well over a decade ago. Again, it hasn't worked out, but what about this new AI reboot?
Well, a couple days after we recorded guess what? Here's 'Muse Spark'. Silly name aside, it seems like the model is solid, though not quite on par with the true frontier players. As we discussed, Meta was clearly laying the narrative down and downplaying this initial release ahead of pushing it out the door.
Still, it's an impressive turnaround time to build a model from scratch. Then again, xAI did the same and that hasn't really mattered. So will it matter for Meta?
Or do they risk becoming the next Yahoo, as Alex frames it? It's an interesting comparison, though I do think Zuckerberg being in place is the thing that probably saves them. While Jerry Yang tried to come back and save Yahoo back in the day – and did from a sale to Microsoft! – Zuck has complete founder control of his company. He just needs the company to execute better on these big initiatives. The entire business is still advertising-based, which sure, is a nice problem to have, but it's also a single choke point risk.
Can the Meta Ray-Bans help with that? Perhaps, but they're still reliant on the iPhone and Apple is coming...
1 Also, please enjoy my significantly upgraded camera in the new Studio Display. ↩
2 That included kicking things off talking about the Michigan Wolverines in the NCAA title game and sure enough... Hail to the Victors. ↩
3 Which obviously pales in comparison to the serious situation that unfolded later in the week when someone threw a Molotov cocktail at Altman's home in San Francisco. ↩