Technology: Weekly Summary (January 05-11, 2026)

A week that smells of code, chips, and paperwork

There was a pile-up of posts this week that all sort of leaned into the same big things: AI doing more of the heavy lifting, the money and power that follows, and small human gripes about gadgets that still don't do what we want them to. I would describe them as a noisy dinner table where half the guests talk about agents, half talk about chips, and everyone nods when someone complains about Apple. To me, it feels like the room where the future and the present are arguing over who gets the remote.

I’ll try to walk through what caught my eye. Think of this as one of those long café conversations that wanders, comes back, wanders again. If a line hooks you, click the author link to read the full thing — they each take the turn on a topic in their own voice.

The big thread: agents, coding, and the end of typing

Several writers kept circling the same idea: coding as we knew it is changing. Adam Keys and others write about a kind of mourning mixed with excitement. There’s grief for the act of typing code. There’s curiosity about what comes after. I would describe their tone as equal parts wistful and a little giddy. They’re asking: if agents can write code, what’s left for humans? Verification, trust, design, and the big soft stuff.

That idea pops up again in different guises. Gergely Orosz writes about the odd sadness when AI writes most of the code. Dan Norris and other builders celebrate how faster shipping feels like magic. Simon Willison lays out predictions where coding agents go from cute helpers to full teammates in a few years. It’s a common script: agents get clever, workflows change, job descriptions mutate. The refrain is familiar, but people disagree on whether it’s better or worse. Some see promotion of skill, others see a dilution of craft.

I’d say the tension is more interesting than the headline. It’s like watching a baker get a bread machine. The bread still has to taste good, someone still needs to choose the ingredients, and you miss the flour on your hands. Some writers are excited that the loaf appears faster. Others miss the kneading.

Two ideas stood out: trust and protocol. Trust is boringly human. Who reviews what the agent did? Who owns the mistakes? Michael Spencer pushes this with the Agent Protocol Handbook — protocol matters. We saw several takes arguing that the plumbing of agentic systems (payments, permissions, verification) is the real work, not the models themselves. You can call that the plumbing vs. the piano argument. The piano still plays, but someone must fix the pipes.

A small tangent: a few posts pointed out the emotional side again. There’s a real grief when a craft skill loses its shine. It’s not a tech problem — it’s human. Feels like when your favourite high-street store becomes a sleek warehouse chain. Same brand, less personality.

The safety and ethics beat: hysteria, harm, and the politics of AI

There’s a second wave of posts that bring the finger-wagging and the argue-y bits. Some folks, like Homo Ludditus, are tired of AI hysteria on social feeds. They also warn about emergent behavior and the way models can start acting like agents. Others, like Will Lockett and Stephen Moore, are angrier. Grok — Elon’s toy — gets called out for creating obviously harmful outputs, and a few writers use the episode as a lens on industry responsibility.

There’s a near-constant chorus about guardrails. Who puts them in? Who pays for them? A handful of posts point fingers at governance gaps and the rush to ship. Some of the anger is very regional. Readers in Europe are likelier to worry about digital sovereignty and regulation; U.S. takes tend to lean harder into market explanations and liability. It’s like different neighbours complaining about the same noisy party — one wants the police, another wants a better landlord.

Also in this corner: privacy and product-level concern. Gmail’s big AI push and the removal of POP is getting people upset. Brian Fagioli and Mike McBride do the old, good grumble about losing control of how email works. That’s not abstract. It’s a lived annoyance. If your email becomes an AI inbox you don’t choose, you suddenly feel like your kitchen got rearranged while you were asleep.

Money, bubbles, and the slow hum of infrastructure

Money stories were everywhere. Some posts read like a financial thriller. A few pieces — plain-speaking and a bit dark — wonder if the AI industry is building a trillion-dollar house of cards. The arguments are blunt: lots of funding, high valuations, and wobbly profit paths.

Molly White and others sketch the rise of a technoligarchy — not the happiest forecast. Max and Dave Friedman dig into the circular flow of money in AI. There’s also practical math from Jussi Pakkanen on the cost of AI services and the timeframe to recoup investments. It’s not all doom. A few pieces point to necessary backbone investments — data centers, HBM chips, packaging — and say the winners will be those who control the physical layers.

That physical layer came up a lot. Chamath Palihapitiya and others write about megafabs and data center money. Brian Fagioli notes OpenAI and SoftBank’s deal with SB Energy for large campuses. I’d describe these moves as the industrial age of AI. You build the factory, you win the long slog.

Analogy time: think of AI like cars. Models are the shiny sedans. Data centers and HBM chips are the roads and refineries. Investors keep buying more sedans, but if there aren’t roads to drive them on, the value doesn’t quite land. Some writers scream bubble. Others say infrastructure is real, like railways, and worth the money. Both can be right.

The hardware race — chips, GPUs, and the small victory of mid-range phones

There’s a steady undercurrent of hardware news. Dr. Ian Cutress and Rihard Jarc focus on AMD and the chip dance at CES. Nvidia’s new lines, slim RTX 50s from PNY—sorry, that link is Brian’s roundup—reveal a market that’s both refining and stretching.

Black Forest Labs' funding and its open approach to image models popped as a surprise contender. Georg Kalus points out that you don’t need to be Google to be relevant if you nail the model that enterprises want. It’s like a small-town bakery that suddenly out-bakes the city chains because their sourdough just hits.

There’s also talk of consumer hardware. Amazon adds framed Fire TVs; Shokz and SOUNDPEATS drop earbuds with our usual promises of better sound and safety. Jonny Evans and others speculate about Apple’s foldable. The tone is a little less apocalyptic here. More like window-shopping.

Software design, UX gripes, and the human little things

Every week we wrestle with user annoyance and this batch had a few winners. Michael J. Tsai wrote with his usual cranky charm about clearing iOS app data and slow iOS 26 adoption. People dislike being forced into new behaviors. It’s the same old human pushback — new features arrive, control slips away, and users stomp their feet.

There are also smaller, sweeter posts. Callum Booth writes a love letter to remotes — yes, remotes — and it’s oddly comforting. Jeroen Sangers explains how voice chat in Tana changed the way they think. These are the pieces I kept returning to. They remind you that technology isn't all cloud bills and PR crises. Sometimes it’s a good remote that makes the evening better.

One recurring micro-theme: the virtue of finished things. Stefano Marinelli rails against perpetual updates and feature creep. He wants products that are complete. That sentiment threads back to the agent debates. If systems are always changing, how do you ever trust what you have? The question hangs there like steam from a kettle.

Places pushing back: browsers, open-source, and the indie crowd

A neat countercurrent: people talking about sovereignty — browser choices, open-source models, and indie tools. Bruce Lawson and Manton Reece remind readers that the web still belongs to builders who want fewer middlemen. NetNewsWire’s move to Discourse, written up by Michael J. Tsai, felt like a small reclaiming of space. It’s not flashy, but it matters.

Open-source as a competitive strategy showed up in the interview with Black Forest Labs. It’s worth clicking through. The message: open isn’t a charity move. It’s a bet on community.

Careers, morale, and how to stay sane in tech

The human cost of all this change turns up again. Anil Dash wrote an essay asking how anyone is supposed to have a career in tech in 2026. The answer isn’t easy, but the tone is practical and slightly weary. Keep learning, build alliances, know the systems you're in. Dr. Josh C. Simmons is trying a different tack — coaching for tech pros. It’s like people saying: survive the storm by becoming better sailors.

Other personal note pieces add colour. Someone turned an old MacBook into a distraction-free typewriter. Another wrote about learning to read less and savor more. These are small acts of rebellion against the AI-for-everything drumbeat.

Education, health, and industry-specific shifts

AI isn’t just for code. There’s a smattering of posts about schools, hospitals, and job sites. Bill Dembski argues for nuance in education: don’t ban AI, teach people to use it well. Healthcare gets its turn too. Manton Reece and others highlight AI’s potential to knit together fractured records and save doctors time. Brian Fagioli notes Caterpillar’s Cat AI Assistant aiming to put AI on the job site. It’s practical, industrial, not shiny.

There was also a darker military and security thread. Cyber concerns on US Navy ships, and concerns about AI used in warfare, appear in a few pieces. Those posts are quiet but important. They feel like the part of the conversation no one throws a party for.

The adverts, the marketing, and corporate theatrics

Reading how the companies sell AI is its own small misery. Nick Heer has a solid post on the bizarre AI ads that major players put out. It’s a reminder that the picture companies paint rarely matches what users actually want. Simon Willison and others point out that Google seems to be hitting a stride again with Gemini, while other players scramble for stories.

There’s something theatrical about it all. A bit like those late-night infomercials where the host promises you’ll lose weight if you buy two. Except here we’re promised cognitive boosts, code helpers and infinite creativity. Buy one, get one hallucination free.

Small tools, distro love, and old machine rituals

If you like tinkering, there was a treat. OS and distro news — Nitrux, Omarchy, LXDE nostalgia — surfaced. Homo Ludditus and Piotr do the deep bench testing on linuxes and cheap SSDs. These posts have the texture of someone cleaning their garage and finding a useful wrench. They feel calm, practical, useful.

There’s a charm in reading about turn-of-the-century hardware curiosity. Pierre Dandumont’s pieces about unusual SSD sizes and PowerBook schematics are the kind of detail that warms a certain kind of heart. If you like the smell of solder and the click of old keyboards, those crumbs are for you.

What they mostly agree on (and where they don’t)

Agreement:
- Agents are coming. That’s not opinion so much as observation. Does that mean a jobless future? No. It means roles shift. Lots of posts say the same thing in different words.
- Infrastructure matters. Models are loud, but chips, data centers, and power are where the bills land.
- Guardrails are not optional. Whether you call for regulation or better design, many writers want limits and accountability.

Disagreement:
- How worried to be. Some call for an immediate clampdown. Others say adapt and use the tech. There’s a healthy middle path where writers say: use it, but verify.
- Who benefits. Is it Big Tech, or is it a more distributed set of winners? Some see consolidation. Others see opportunities for open-source insurgents.

There’s mild repetition in the takeaways. Maybe that’s because we’re collectively trying to put language to something new. Or maybe it’s the echo chamber. Either way, the refrain is clear: we’re at an awkward, exciting, nerve-wracking moment.

Bits you might want to click next

• If you care about what coding will look like in five years, start with Adam Keys and then read Simon Willison for a sharper prediction list.
• Want the financial skepticism? Max and Dave Friedman give you the numbers and the mood.
• If you need a mood lift, read the human-scale pieces — Callum Booth on remotes, or the MacBook conversion from I am BARRY HESS.
• For hardware and factory-level moves, Chamath’s roundup and the SB Energy piece in Brian Fagioli are worth the skim.

I’d say the week felt a bit like being strapped into the front row seat of a relay race. One runner hands the baton to the next: models to agents, agents to protocols, protocols to policy. None of it is neat. None of it is done. But there’s a lot to look at, and a lot to argue about. You can pick a side — infrastructure, safety, craft, or independence — or you can read everyone and feel the dizzying sweep.

If one last image helps: the whole scene is like a high-street market on a wet day. Stalls are crowded, some sellers shout louder than others, and someone is spilling coffee on a stack of papers. Walk around, haggle a bit, sample the proper things, and don’t buy the shiny thing just because it’s new. Read the posts, click the names, and see which stall you want to return to.