F Apple, Marry Anthropic, Kill Microsoft
Predictions for Tech in 2026
Entering 2026 I believed that, by October, the AI bubble would burst and the American economy would lose its one bright spot and be exposed as a dud. One month in and I no longer think that’s true; the AI ferris wheel will go another round and much of the hype that exists today will still be here next year. But, that doesn’t mean the circus isn’t still in town. Let’s go booth by booth.
Prediction: Apple gets back on top
Apple unequivocally has been the loser of the AI Vibes Games. 2024 was their complete collapse, where they promised amazing AI that ended up bupkis. In 2025, they spent some time cleaning the house, waiting for people to stop believing in Siri, and got back to shipping banger phones. Today, the general feeling is less “Apple sucks” and more “Apple missed an opportunity,” but whether intentionally or not, the results of staying out of AI has kinda…worked?
When the story of tech — and to some degrees the entire economy — is AI, you have to respect the fact that Apple, arguably the most important tech company, has put out next to nothing AI-related and is still out there crushing it.
The bet is this: the Gemini deal (which is still only $1B) gets Siri to at least parity with the AI chat apps, and shortly afterwards becomes a magical upgrade that make iPhones once again the best devices money can buy. To reiterate, Siri x Gemini probably won’t even be better than, say, the ChatGPT app; but just being in the arena will be enough.
Prediction: Anthropic wins AI
2026 is going to be an undefeated season for Anthropic. Let’s break it down by predicting games against the other teams:
Anthropic v OpenAI
This one is a game of focus, discipline and restraint. Their operating principles are fundamentally different: OpenAI is doing a 2010s ZIRP growth at all costs strategy, while Anthropic is reining in a lot of excess to build a functional business. This goes for both the actual products they’re building and for who they are targeting as customers.
Anthropic is dominating coding for their models, and is aggressively hitting the enterprise. Consumers don’t know Claude, and that’s totally fine; basically every engineer agrees Opus 4.5 is the best coding model, and 40% of enterprise AI spend goes to Anthropic vs 27% for OpenAI. OpenAI does not seem to have a plan other than “grow more.” More on that below, but they appear primarily focused on the (free) consumer and losing out on paying customers.
The way they are distributing the product is just so different: OpenAI is giving the entire federal government ChatGPT basically for free, while Anthropic is not going to be in Siri because they wanted too much money for it. Anyone using Claude has hit a limit, whereas ChatGPT feels like you can prompt it forever. The discipline to say no to customers, and to have a clearly defined path forward, is what will push Anthropic ahead.
Anthropic v Cursor
This one is in the IDE: Claude Code against Cursor. First, Cursor essentially owes its life to Anthropic figuring out something magical for coding in Claude Sonnet 3.5. Also, it definitely got big through Twitter influencer marketing through 2024:
Cursor blew up in 2025, but now things are not looking so hot: Claude Code is the hot new thing—the new narrative is that everyone is switching to it—and the Composer model Cursor is building seems to be a bit of a nothingburger. The only “moat” was the $700M Cursor pays to Anthropic for their models, trapping Anthropic into not being able to cut them off for such a big chunk of their business, but that chunk looks smaller every day.
My prediction is that if Claude gets a Sonnet/Opus 5 that is notably better than 4.5 at coding, Anthropic won’t share it. Claude Code with it’s new models becomes SOTA for development, Cursor can’t answer, and everyone churns. Bye bye Cursor.
Anthropic v Public Markets
This last one isn’t so much against as a hurdle: Anthropic is expected to go public this year. Everyone and their grandmother will want to buy Anthropic on day 1 – obvious bet. They’re losing money, but not at a ridiculous scale like OpenAI is, and they’re going to go public at like half their valuation, a clearly better entry point. This just seems like an obvious slam dunk to me. Buy ANTH, I bet $1T market cap by this time next year.
Prediction: Nadella gets fired
Initially I was just thinking “Short MSFT” for this, but after they dropped 11% in a day (shoulda wrote this a month ago) I have to ramp up the prediction. But the sentiment remains. Microsoft did the equivalent of: draft Pat Mahomes, let him walk on his rookie contract, and then lose to his new team in the first round of the playoffs. Very few series of decisions that could’ve been a bigger biff of the opportunity here, truly.
A year ago, Satya Nadella was bragging he could pay for Stargate, Microsoft owned OpenAI’s IP, had them locked into using only Azure for compute, and basically owned the company. Today, they have a third of the company, Azure business is reliant on OpenAI but not the other way around, and the impact of AI in Microsoft products is pretty much limited to new names.
Microsoft is clearly weakest between Amazon and Google, in terms of Azure vs. AWS/GCP and Maia vs. Trainium/TPUs and partnership with Anthropic. Really unclear to me how Microsoft can come back from this without a massive shift… one example of which would be Nadella getting dumped. He has done legitimately nothing interesting to introduce AI into Microsoft, and that just feels fireable when you were the first to get there. Massive screwup. Bye bye Satya.
On OpenAI
In short, this company will keep moving forward, but the promises/reality membrane is getting closer and closer, and I really don’t think OpenAI can successfully cross it.
Start with Sam Altman. Every company he’s been at has tried to kick him out, always for reasons in the arena of being a lying grifter. Maybe he’s not like this, but based on evidence of what I’ve seen, it seems sort of true. But what also seems clearly true is that his goal still is to build AGI for everyone. While noble, his plan is pretty much straight out of a Boulder dorm room smoke sesh (you can read it still): AGI means robots do all the work, everything is cheap/free, in relative terms everyone is very rich!1 People play the greedy billionaire card on Sam, but I do think he’s motivated only by the stated mission of creating AGI for everyone, and not money. It’s actually the only way what OpenAI is doing really makes sense.
This has been the case since the summer when they were on a deal-announcing spree, but their strategy going forward is this:
Build products that use a lot of compute. Whatever it is, as long as it does more compute than before. Generate images, generate videos, generate suggestions for you while you sleep, listen to you from a device and talk to you all day—more more more compute
This “demonstrates” demand for ChatGPT and compute. Sam, whose main skill is fundraising (and he is the GOAT) goes out and does two things: convinces those who can invest to give him more money, and convinces those who can’t to build data centers.
More datacenters = cheaper compute = ability to afford making pennies on the dollar serving a billion ChatGPT users. Once inference and training is cheap, the AGI trek becomes a sprint, and the vision is fulfilled.
We have seen hints that this is their strategy (outside of Sam basically saying it multiple times): a former employee wrote a blog post about working there, with some choice quotes:
OpenAI is incredibly bottoms-up, especially in research…Rather than a grand 'master plan', progress is iterative and uncovered as new research bears fruit.
It wasn't unusual for similar teams but unrelated teams to converge on various ideas. I started out working on a parallel (but internal) effort similar to ChatGPT Connectors. There must've been ~3-4 different Codex prototypes floating around before we decided to push for a launch.
There is a strong bias to work on your own thing and see how it pans out.
OpenAI changes direction on a dime.
The company pays a lot of attention to twitter. If you tweet something related to OpenAI that goes viral, chances are good someone will read about it and consider it. A friend of mine joked, "this company runs on twitter vibes"… There's certainly still a lot of analytics around usage, user growth, and retention–but the vibes are equally as important.
I don’t think compiling this list and presenting it is taking those things out of context: seems like the author geniuely liked those qualities. As a business that loses a shitload of money while having made a trillion dollars worth of spending promises, though, feels like they might want to operate with a more clear plan?
The numbers don’t add up very well for the company, and it’s clear their only shot is to be given a couple hundred billion dollars in the next couple years to make it over the hump. Honestly—I think they’ll get it. It’s already out in the open that the biggest players will prop up the AI boom to keep it going. The issue for OpenAI is that money will just run out: they’ll get their $100B round early this year, then if they go public, we’ll see—but unless it’s a blockbuster, that will sort of be the ceiling of money they can realistically pull in, no? The drama will be when we have Anthropic and OpenAI S-1s: for me, the company with saner financials, better enterprise, clearer direction, and possibly better models is where I’d put my money.
So my prediction for OpenAI is they keep growing and keep raising money, but don’t make much progress on becoming a better business.
The rest of tech
Above were the big narrative predictions for this year. These are random thoughts and some lesser predictions.
Meta
First, it’s important to know that I fuckin hate Zuck. He’s the worst and his company is awful. He is driven by a singular goal: to have one idea on his own that works out. This will never happen, because he is a lizard who lives in a padded room. Every single thing he’s done in his life was stolen or bought from someone else, and it continues to work out because everyone is addicted to the Facebook/IG ads machines.
His “vision” for AI has no substance—“Meta's vision is to bring personal superintelligence to everyone”—and his AI team is a group that are all only there because he literally bribed them. Meta will continue making more and more money by just showing more ads, but say it’s because of AI, so he can spend more on AI.
My predictions are thus:
Alexandr Wang, the head of AI, leaves by end of year. Probably goes to some personal AI thing: the Jony/Sam device, or Friend.com, or something like Poke
IG Reels becomes the new TikTok in the US. Their algorithm shifts to the TikTok model: follower graph doesn’t matter, it’s all content-based, and as TikTok’s new MAGA ownership starts censoring in the name of free speech, Reels takes over the wheel.
Meta glasses things continue to not be bought by, talked about, or even known by anyone outside tech journalists.
Hardware devices
These really just aren’t going to kick off at all. Sam and Jony’s device thingy will be a dud. It will probably look cool and have great marketing videos, but ultimately it won’t do much, probably won’t work very well, and nobody outside of people who bought Vision Pros will buy one. Friend.com had a pretty-clearly fake guerilla ad campaign and ended with most people hating it and nobody buying it; the dude is basically just a Twitter Guy who believes good marketing will overcome bad product. I think they will close shop this year, along with Rabbit, following in the footsteps of Humane
The last bits
Buy Google, short a neocloud (Coreweave, Nebius, etc) and/or Oracle
Replit & Lovable, and vibe-coding in general, comes back down to Earth as it becomes clear they can’t penetrate the enterprise.
Someone will get caught blatantly insider trading a prediction market, and the outrage will not overcome the fact that the current government is doing everything it can to help grifters scam, and generally shitty people get money
Elon’s businesses are very unpredictable: TSLA investors do not seem to use logic, SpaceX, xAI and Tesla may all combine into one, and Grok/X might get banned in some places. Really too volatile to say what will happen there.
I don’t care about moltbot, sorry
This exercise was a way for me to put down all my thoughts about tech. I’ll come back to it in the summer and see how things are going, and then in a year and grade myself. Could be fun. Thanks for reading!
If you just read the plan, called Moore’s Law for Everything, he’s obviously a dipshit. One of the sections, titled “Capitalism for Everyone” describes how companies would be taxed by the government in shares, then equally redistribute the shares…as money…to citizens. Karl Marx liked this post.