2025 Predictions Revisited
Last year, while trying to kill time waiting to ring in the new year, I jotted down ten predictions that I thought might come to pass in 2025. Actually, I thought most of them wouldn't, but I thought they could. Here we are with only a few days left in the year, so let's check back in, before jotting down some new ones for 2026...
Apple buys an AI company – As we enter 2025, the consensus view on Apple's initial foray into AI seems to be indifferent at best and underwhelmed at worst. As Siri herself starts to get real upgrades in 2025, that could change a bit. But Apple is running a race quite slowly – too slowly – behind folks who are doing laps much faster. One way to leap ahead would be to buy a player. Apple famously doesn't do a lot of splashy M&A but perhaps an exception could be made for a Mistral or a Perplexity. Who knows if Apple would even be allowed to buy the former given their war with the EU. But the latter could be interesting in particular in light of the impending end of the Google default search deal/payments...
My first prediction – the one I was probably most confident in! – is the one that absolutely did not happen. To be fair, I do believe that Apple bought up some AI talent, but I was also clearly thinking this would be a larger deal. And as we got into 2025, it became clear to many people why Apple might want/need to do such a deal. And Apple apparently thought about it too – hello, Perplexity – but ultimately decided against it; something I eventually talked myself into as well – especially given the prospects of a Google partnership, which should come into play in early 2026, to bolster Siri. Finally. Anyway, smaller deals and simply thinking about doing larger ones don't give me a passing grade here. Grade: D+
Someone buys Warner Bros Discovery – David Zaslav clearly – clearly – wants to do a deal. But the music is slowing down and other players are already finding seats to the point where it's looking more like he won't be able to find a partner to buy as much as he'll be the one selling. Perhaps it's just some assets – CNN and/or HBO? – or maybe it's the whole thing. If the M&A environment truly is about to thaw, Comcast or someone else could make a move. Might Paramount aim to bulk up even more post-Skydance deal? And of course, I'm always trying to come up with ways for Apple to spend money, per above. It does make some sense...
While I guessed that it might be Skydance, fresh off of the Paramount deal – a good guess, if I do say so myself! – it was actually Netflix that bought Warner Bros. Okay, technically the deal hasn't closed. And probably won't for months – if not years. I mean, it obviously may never happen. Or Paramount Skydance may yet get that last laugh. But still, at least right now on paper, I nailed this: Grade: A
Intel gets bailed out – We'll see what ends up happening with the CHIPS Act under the new Trump administration, but regardless, I have a hard time believing they're going to just let Intel wither on the vine. Enabling/bolstering a true American player in chips is a win, no matter the political party. Could a new administration help broker a deal to get, say, Microsoft to support the rejuvenation of Intel? First things, first, Intel needs to put in place a new leadership team. I'll keep my (early) money on Lip-Bu Tan here – if he can overhaul the entire board as well...
We can argue the definition of a "bail out" I suppose. But at the highest level, clearly the government not only investing in Intel, but becoming their largest shareholder, seems like it should qualify for my notion. Further, that support clearly got others on board as well – both SoftBank and NVIDIA with money, and rumors of Apple and others looking at Intel for their manufacturing needs. And I think I get some bonus points for correctly predicting Lip-Bu Tan as the next CEO of Intel. Grade: A-
Elon Musk bails on the White House – I'm a bit torn on which side will sour on the other first here, but I'll go with Elon getting fed up with all the in-fighting – which we're seeing already, of course – and bureaucracy. He'll say the whole thing is fucked and go back to trying to bring about change in his more insurgent manner, by shitposting on Xitter. There are many reasons for him to hang on as long as he can – billions, in fact – but perhaps he'll pre-empt a move by Trump here. There can be only one main character, after all. And he only shares the spotlight for so long...
The only issue here remains the question of who dumped whom in the breakup which was utterly predictable. "Vibes" suggest it was more Trump (or his inner circle) that got sick of Elon, but there was obviously annoyances the other way as well. (And clearly, Tesla shareholders wanted him back and focused on at least one of his companies.) After Elon went nuclear a few times on Trump on Xitter, and Trump responded with reminders of who has the real control, the two seem to have reached some sort of détente. Grade: B
Amazon's Alexa overhaul proves less than "remarkable" – Not exactly shocking given all the reports and the fact that that project has clearly been delayed multiple times already. But I just have this feeling that Amazon is going to have a hell of a time trying to convince the world that a new Alexa is above and beyond the old Alexa – let alone worth paying for. Their previous success with the Echo devices is their undoing here. Might it push them ever closer to Anthropic as a result? The pieces are already in place for that relationship to be the new Microsoft/OpenAI marriage...
Yeah, I mean, after a grandiose unveiling that reeked of over-hype, it took forever for Amazon to actually roll Alexa+ out to a meaningful swath of the user base. Once they did... the silence is sort of deafening. I have it now, it's... fine. A less capable version of ChatGPT or Gemini mixed with a more capable Alexa (though in some ways, the new functionality is predictably less capable, or at least stable). Still, it's nice to have conversational AI scattered in some rooms around my home (because it's still not on every Alexa device). But not the game-changing transformation that had Panos Panay getting emotional. As for that Amazon/Anthropic relationship, it's getting... complicated in ways that really does have echoes of Microsoft/OpenAI! Grade: B+
Microsoft and OpenAI kiss and make up, or break up – It all comes down to the negotiations to turn OpenAI into a for-profit company. Assuming this happens, and Microsoft gets a massive (actual) equity stake as a result, I think that plus the continued floundering of their own consumer AI efforts will lead to a re-kindling of the relationship. Oh yes, and the fact that OpenAI still needs another $20B+ or so to make their AGI dreams come true. SoftBank and the sovereigns can help, but they still need a true American Big Tech™ partner to compete with Google and DeepMind here. If/when the for-profit status comes into effect and the relationship is rekindled, OpenAI should shoot past a $200B valuation in new tender offers. If the rift instead deepens and Microsoft and OpenAI break up in the name of independence, look for the former to be involved in a deal to buy TikTok...
This is a tricky one to grade since they technically did kiss and make up in order to seal OpenAI's deal to transition into a PBC – thus securing Microsoft's 27% stake – but it's not like they're fully back in love. Instead, Microsoft made their moves early in the year to pretty much fully outsource any OpenAI growth commitments to SoftBank. At the same time, this frees up Microsoft to more fully and openly compete with their partner, both with their own internal push for AGI/Superintelligence but also by partnering with other OpenAI rivals like Anthropic. Still, it's not like it's a full break-up of the relationship. Again, Microsoft owns a massive 27% chunk of OpenAI. It remains... complicated. Anyway, having just unwound one insanely expensive deal, it didn't seem like Microsoft was in a hurry to jump into the TikTok deal (again) – though they seemingly did debate it! (I would also just note that it's also not fully clear who the owners of that mysterious 5% equity slug are yet...) Anyway, mainly I'm annoyed for drastically underselling OpenAI's would-be valuation. $200B? They shot past that one month into the year. Try $500B now. With talk of new money coming in at $850B! Grade: C+
NVIDIA comes back to Earth, a bit – Overall, there's almost no way that Blackwell isn't a massive hit for the company. To say there's pent-up demand undersells it, and these will certainly not be undersold! But I also wouldn't be surprised if NVIDIA's overall numbers are more muted in 2025 than people may think due to supply and complexity issues. But more so I'd chalk this prediction up to the law of large numbers simply catching up with NVIDIA. They can't keep growing as they have, so simply awesome growth will have to do instead of bonkers growth. Also, what if they're not as dominant in AI inference as they have been in pre-training? And, of course, everyone – including all of their major partners – are now sprinting full speed to try to diversify away from the reliance. Still, that's probably much further off than 2025. Again, this year already seems pretty signed and sealed...
Another tricky one to grade since it's both right and wrong. From a market cap perspective, NVIDIA has fallen a bit – from the $5T highs down to a "mere" $4.6T to end the year – but they're also well ahead of where they were a year ago (and still ahead of everyone else). The earnings remain great – they even managed to reaccelerate last quarter! – but yes, investors are getting a little more cautious on their numbers given the above (natural) issues. And mainly, there's growing concerns about the "AI Bubble" in general. The shorts are out! But the main "vibe" I think I got correct here is around Google's TPU chips as an up-and-coming challenger. They're not going to displace NVIDIA's place at the top any time soon, but they could eat into some of the business. And yes, their last-minute $20B Groq deal seems to be about both snuffing that out while also ensuring they don't lose out on the growing inference wave. Grade: C+
Threads passes Xitter in active users – This seems to be more a matter of "when" not "if" at this point. But while early models suggested it may take a few years, Meta is seemingly getting more aggressive when it comes to growth. A few more turns of their dials and I bet they pass Xitter this year. Bluesky will continue growing too, but I don't think they will be able to compete with the growth engines and talents (and cloning capabilities) that Meta possesses – even if they're making more of the right moves in the space...
This sort of happened. On mobile, by some metrics, Threads is now bigger than Xitter at least in terms of DAUs. By other counts, it's very close. But on the web, it almost certainly is not. Overall? It's hard to say for sure, but most estimates still have Xitter ahead by around 100M (MAUs) or so. Call it 500M to 400M (maybe as much as 600M to 350M or maybe as little as 450M to 400M). Close, but no cigar. Also, it seems like Xitter may be growing again, so it's a matter of if Threads can keep up the faster growth rate in 2026 to fully overtake them. Will the merger with xAI help or be a distraction? Is this whole thing just gonna be a part of Tesla and/or SpaceX one day anyway? As for Bluesky, that prediction seems accurate: still growing, but not nearly at the rate of Threads. Grade: C
Google starts to feel real pressure on search – This is a tricky prediction to make because everyone makes it every year – and has for years. And yet it's never true. But I do think that the various AI players may finally start to make some inroads here in 2025. And that could also be related to the end of the default search contracts – that will undoubtedly be tied up in courts for years, but just the threat of it may open some players – like Apple – to start to look at other options here. Perhaps this is just anecdotal, but I'm using ChatGPT for a lot more queries these days... Of course, Google could combat this with a bold placement for Gemini...
Even though Google continues to maintain that Search is healthy and there's "nothing to see here", simply watching their actions – or listening to Apple exec Eddy Cue, under oath, no less – would seem to suggest otherwise. 2025 was the year that Google pivoted hard to shove AI into basically every surface of Search. And it seems to be working – as OpenAI's "Code Red" at the end of the year is testament to. While the judge in the antitrust case altered the details of the default search deals a bit, he ultimately allowed such deals to stay in place – a huge win for both Google and Apple (and Mozilla). Why? Largely thanks to the looming threat of AI disrupting everything, of course. So yeah, this one feels mostly correct. Grade: A-
Mark Zuckerberg unchained – While it all felt a bit too calculated for my taste, the "Zuckaissance" of 2024 was nevertheless impressive. But come on, this is PR 101. The media builds someone up to then knock them down. Zuck has lived this cycle a few times now and he's due to be knocked down again after a year of the press eating up the gold chains and wakeboarding. Not even sucking up to the incoming president can stop this inevitability. It's just a question of what causes the backlash. Something with Llama? Instagram? WhatsApp? Good old Facebook? Something political? Maybe flooding all our feeds with AI bot nonsense? Anyway, this knock down needs to happen in 2025 to get ahead of the next Zuck build up as the "Orion" AR glasses get closer to launch...
It took a whole week for the 10th and final prediction for 2025 to start coming true. After a 2024 in which Mark Zuckerberg seemingly got endless love in the press and was riding high, 2025 was indeed the knock-down, wake-up call. Beyond the blowback he got (from both sides) for his political wishy-washy-ness, he also got served a huge piece of humble pie in the form of his AI ambitions for Meta. Once Llama 4 failed to compete with (let alone overtake) their main competitors in AI, Zuck had to spend billions to blow up their efforts and start over. Maybe it works, but I'm skeptical. Regardless, it's a bad look when you spent the previous year talking up the importance of "open source" AI, then you basically throw that out the window to bring on a team aspiring to build behind closed doors – which is also seemingly causing a lot of internal strife, as you might imagine. But hey, could have been worse: they could have lost their own antitrust case and been forced to sell/spin-off Instagram and/or WhatsApp. But the FTC's case was so weak that the judge sort of laughed it off. As a result, Meta's stock, which ended up slightly for the year, could have fared far worse – especially with Wall Street growing skeptical of their CapEx spend in particular. That is, of course, also why we're seeing Meta start to pivot away from the metaverse. Awk. Grade: A
One more thing: I also had a bonus bet...
One more thing: My bet would be that after some strong initial buzz amongst the Apple diehards – including myself – Apple's new 'FacePod' home hub thing fails to catch on in a major way as well. Beyond people who buy everything Apple – again, that's me – it's not clear what it will be used for. Though I'm not sure how much it will matter as it's just an initial attempt to enter the home in a major way (well, beyond HomePod, I guess), but still, if it's not a hit out of the gate, there will be pressure on Apple following the fizzle of the first Vision Pro as well...
It's true that Apple's 'FacePod'/'HomePad' failed to catch on – because they never shipped it. It was seemingly delayed as a result of the AI fiasco and should instead come early next year. I'm still skeptical of the device, so let's call this a "push" for now.
Overall, ChatGPT tells me my GPA would be 2.96. We'll round up slightly for a solid B on the year. I also went ahead and published a page with a running tally of predictions throughout the year. Well, the correct ones, at least 😝. On to 2026 🥂.
M.G. Siegler
M.G. Siegler