It's always crazy in the post-whatever-we're-in era. Let's go over some thoughts and topics.
I might make a few soft predictions, just so we can benchmark at the end of 2026...but please, don't time any investments or big moves to what I say. Let's allow ourselves to have a bit of harmless fun speculating in a blog post. 😅
AI
It goes without saying, but the largest trend to note is that of AI and its undeniable gravity. I think it would be naive to omit or ignore, so let's give it its due.
Hyep!1!!
"Markets can remain irrational longer than you can remain solvent."
Whether you agree with the concept or not, we're seeing something akin to the hype cycle for AI, much like we have seen in the past with other technologies. The one thing that makes AI a little different is the broader climate of wealth consolidation in the global economy. Where the industry was able to place many smaller bets at once, that beam has tightened into a singular and dogmatic focus. For better and mostly for worse.
My sense of it is that we're somewhere on either side of the trough of disillusionment as of the time of this blog post. Though we spent most of the year perched atop the peak of inflated expectations, which seems to have caused a money glitch between the hardware and software vendors, prompting fears of a bubble-style collapse.
Predictably, people who get their news on a bit of a lag may still feel like we're still in that early stage of the hype cycle.
You'll observe this in the form of your less tech savvy family members and acquaintances asking you for "your opinion on AI" during coffee talk.
Perhaps that tradesperson sparks up the discussion based on something they saw on television, or your hair stylist has it on their roster of go-to conversation starters.
Generally, I've found that once the discourse reaches the less information-literate or hard-to-reach mainstream - where knowledge travels at the speed of in-flight-magazines - things are actually much further along than they are aware.
AI Credit Deflation
This started happening in October and November, but your "existing members will get some loss-leader AI credits to play with" have started to depreciate.
Now it's the end of December and what seemed like a curious value-add to something you were already paying for is starting to feel like a hard upsell.
This thrashing is an attempt by companies to recoup the inordinate costs of operating competitive AI functionality. It might also be C-suite, VPs and product managers desperately trying to shore up the bad bet they made chasing a LinkedIn popularity contest. It belies a stillborn business model where users are happy to use AI functionality, but refuse to be metered on its usage.
Which makes sense, really.
This nickle and diming mimics the reluctance for people to adopt cloud-hosted, instanced productivity tools.
If every action I take in a tool causes a little credit in the corner of my window to tick, that's a hard pass.
Any product that tries to put a tightly metered tax on peoples' core work loop is ultimately doomed to fail.
Democratization of AI
Local AI is going to keep on quietly getting better, but this will be stunted by the AI blowout that's starting right now as hardware - RAM & SSD particularly - prices go through their own little artificial shortage.
The integration patterns, unfortunately, are a little bit of an unknown, and I'm not familiar enough with the conversations going on to know if we'll make it much further past MCP for 2026. As someone who has been waiting on the sidelines for his moment to jump in, I still have much to learn.
Still, pain can be felt by entities trying to implement voice-to-voice and realtime conversation functionality.
That's a hyper proprietary and computationally intensive space right now, and the gatekeepers are doing their best to keep it that way for as long as possible.
Which means it's also ripe for disruption.
A practical AI
On a more positive note, the word circling in business circles is around "how can we make AI useful". So less unserious people are finally taking a look, and topics like RAG and MCP will probably do very well. It makes sense, because those are the only two things right now helping AI stay coherent in everyday use.
This plays nicely into the next section...
The impact of AI on jobs
Now that we know hyperactive light-switch-flipping robots can't always be trusted to tell the truth or avoid being sociopaths, we're curbing our appetites.
I can't guarantee this or predict exactly when, but at some point either in 2026 or 2027 (more on the obnoxious prescience of six-seven below), tech professionals might feel a bit less of a pinch.
The one thing that might threaten this is are broader global factors... 😐
Bagagwa help us all if there's war, but at least it'll be a #weird-war, waged on behalf of brain worms.
AI has prematurely reaped jobs out of the industry, and because large corporations live in that mysticized information vacuum I described earlier, it's going to take some time for the pendulum to swing back.
But I think it will, and people will be expected to be better at building systems and doing analysis. Wishful thinking on my part? Perhaps! Hooray! 🎊
Closing thoughts on AI
It's very-good glorified autocomplete. It is not the sentient dystopian singularity some want it to be, and we need to not let our imaginations get carried away.
People are talking about the bubble popping soon - let's hope they're right so that money can finally be used for things like food, shelter and taxes again.
Popped or not though, I will definitely be dedicating some cycles to learning about practical ways of using AI in 2026.
Alright, now that the ritual keyword offerings have been made and the algorithm is placated, let's move on to some other thoughts.
Post-enshittificationalism
Anything and everything Cory Doctorow has said is now accepted as true.
"The only way you can make it a mystery is if you have this specific kind of brain injury that you can only get by doing an economics degree."
- Cory Doctorow on Consolidation
The question now becomes "So what are we going to do about it?"
🤷♂️ [Insert the future here.]
If you want to at least prepare a little for whatever that future is, go read his stuff and watch him on YouTube.
Decentralization: Tech Ethics and Data Sovreignty
Admittedly, this builds off of the last section. But this is a topic near and dear to me, and we're at least finally starting to talk about it!
We're becoming more accepting of discussions about how we can use technology in ways that don't create as much risk, suffering and displacement.
We need to think about tech supply chains, people need to take an active stake in their social graphs, companies need to see the value in retaining some technology muscle.
Okay, that last one is maybe a bit more aspirational, but you get my tone.
I don't think we're going to end 2026 with it all sorted out, but as I said: We're at least finally starting to talk about it. Who should own the last mile to your house? What algorithms are you feeding? Who develops and hosts the hardware or software your government runs? Is that data centre in your backyard a good thing or a bad thing, and was it actually good for your local economy or did it just spike your power bills?
These are the tech-flavoured questions that come as we start to shake ourselves free from the growth scam. A core tenet economists have been ramming down our throats over the last 50 years.
I've been wishing for this dialogue since at least 2004 when it was taboo and imprudent to blend tech and ethics.
Remember those days?
Early adopters self hosting
That is to say, where technologists have been doing it for a while, early adopters might be dipping more than just their toes in. This continues the theme of data sovereignty, but only if the software gets better.
I'm curious about where technologies like TrueNAS, NextCloud and other similar self hosting tools are going to be by the end of 2026. They might be realizing that they're part of a moment, but the extent to which they have the resources to respond could be a limiting factor.
Will they seize the opportunity? Are the technologies built well enough to integrate in the ways people will want, or will it be a patchwork nightmare?
Bingo Card Ideas
Enjoy some or none of these, I might add some more after publishing.
- We'll call it "2026/7" -- Even though the schoolyard is moving on from the whole six-seven thing over the Winter holidays, I've already seen it being used as an anti-nostalgia. Two years in one! Or more hopefully, one year in two...
- Anti-nostalgia! -- We spent 2025 farming lots of dread and memes, time to harvest that meta gold from them thar crop-hills.
- From brainrot to brainshit. -- Make of that what you will, but the big shift in my view is intent. Where brainrot seemed spontaneous and self-unaware, brainshit is a more obvious and contrived attempt to emulate brainrot for gain. I do not think it will be effective and we will have an easier time ignoring it compared to prior forms of manipulative content.
- "Autogenic" -- Now there's some arable linguistic land that hasn't been claimed yet. For what? I don't know yet, but you heard it here, an operator is standing by!
- Linux. -- Not like "year of the Linux desktop". That's already happened, you just don't know it yet and I'm not talking about Android. But 2025 saw many popular personalities give Linux airtime that would have otherwise been considered socially unacceptable. Linux desktop share climbed, and whether you consider this due to the Steam Deck or not, it will count either way, especially with the upcoming Steam Box. Harass me for a blog post on this.
- "Napster good!" -- Yeah, piracy has already been trending, but nobody feels the obligation to brag about it except in the Reddit comments about a streaming service increasing their prices.
- "I cancelled Spotify" -- This will be a good move, but people will force it to be maladaptive, probably by going over to Google or Apple. See prior notes on digital sovereignty and piracy.
- Homestar Runner -- Except we only want Strongbad and he'll be selling The Cheat plushies. RIP Senor Cardgage and Hey Steve.
- There aren't enough people on Mastodon -- Which means Reddit is safe for now, so long as they don't engage in more censorship. I'd love to be wrong on this, but the Fediverse is still baking. This might be a more interesting topic in 2027.
- RISC-V will continue screwing around -- Kind of similar to Mastodon, it will defer its own destiny. Single board computers can't iterate cheap enough or fast enough in a world with a dysfunctional supply chain. The planet is overdue to replatform on RISC-V, but we just can't seem to cut the nonsense.
- A browser engine war -- But I don't think there will be a winner or a loser.