This Week in Tech 129

What we explore this week:

Dario Amodei just published a major essay arguing that AI is advancing so fast it has fundamentally outpaced the government's ability to regulate or respond to it. Read the full essay here. This is the CEO of one of the most powerful AI labs in the world saying the gap between technological capability and policy readiness is a genuine crisis, not an abstract concern. When the person building the technology is sounding the alarm in public, in writing, it deserves more than a scroll-past. The uncomfortable implication is that the window for getting governance right may already be closing.

Cursor just announced Origin, a git hosting and code collaboration platform launching this fall. They're no longer just an AI code editor; they're gunning directly for GitHub's territory. The logic is clean: if AI agents are writing and reviewing code, it makes sense to also control where that code lives.

Fresh off a $900M raise at a $9B valuation and recent SpaceX acquisition news, Cursor has the runway to fight this war on two fronts. The company that owns both the editor and the repository has leverage over the entire software development workflow, and Microsoft should be watching this closely.

NVIDIA unveiled MotionBricks at SIGGRAPH 2026, a physics-based motion AI system that gives game characters and robots access to over 350,000 motion skills — running, stumbling, adapting, transitioning — all in real time at 15,000 FPS with 2ms latency. This goes well beyond smoother animations. It represents a fundamental shift in how digital and physical agents understand movement as a composable skill set, more like muscle memory than scripted behavior.

For game developers and roboticists alike, this collapses the gap between canned behavior trees and genuinely lifelike motion, and the robotics applications may matter more than the gaming ones.

2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.

Download the Toolkit

Snap made its boldest hardware move yet, officially opening preorders for Specs: fully standalone true AR glasses shipping this fall in the US, UK, and France. UploadVR broke the preorder news, while @NathieVR shared a first look at the device targeting productivity, streaming, gaming, and everyday use. The technical specs, detailed by @SadlyItsBradley, are serious: LCoS waveguides, a 51-degree field of view, dual-chip architecture, electro-chromatic dimming, and 132 grams total weight across 47mm and 52mm sizes.

The official Snap AR developer page confirms the $2,195 starting price and Fall 2026 ship window. At this price and with these specs, Snap isn't chasing the mass market yet. They're planting a flag that AR is no longer a prototype promise, and putting themselves directly in the ring with Meta and Apple for the future of how we see the world.

Qualcomm announced the Snapdragon Reality Elite, a new flagship XR chipset delivering 60% better GPU and 30% better CPU performance over its predecessor, built from the ground up for both VR and passthrough AR displays. As reported by UploadVR and detailed by @vrgamesshowcase, the chip debuts in the XREAL Aura, an Android XR glasses-plus-puck device opening reservations now. XREAL's own announcement pitches the Aura as a pair of lightweight smart glasses capable of turning any room into a multi-monitor spatial workspace.

The tight integration with Google's AndroidXR platform is the strategic tell: the Qualcomm-Google-XREAL alliance is quietly becoming the Android of the spatial computing world, and if the Aura delivers, it could be the first Android XR device that actually matters.

While everyone debated whether AR glasses were ready, Meta Ray-Bans tripled their daily active users year-over-year, according to Meta's Q1 2025 earnings. That's not a fluke; that's a trend. The secret weapon isn't a display (there isn't one): it's the AI assistant baked in, making these the first wearable where the software actually justifies the hardware.

As WIRED detailed, Meta cracked the problem previous smart glasses couldn't: making the thing on your face actually useful without looking absurd. If this growth rate holds, we're watching the early innings of the post-smartphone era take shape on people's faces.

VITURE just unveiled Helix, AI safety glasses built on NVIDIA's XR AI platform that see and hear everything you do, positioning wearable AI as a serious workplace tool rather than a gadget. The safety glasses form factor is deliberate: it targets enterprise and industrial environments where previous smart glasses flopped trying to be consumer products. NVIDIA's hardware muscle as the platform provider gives Helix a credibility advantage that Google Glass never had.

The race to put an AI co-pilot on your face just got a major industrial backer, and the enterprise adoption path here is more plausible than anything we've seen from smart eyewear before.

UploadVR dug into the full visionOS 2 feature list and found a stack of meaningful additions Apple quietly buried beneath the keynote highlights. The full rundown is worth reading if you own a Vision Pro or are tracking the platform closely. Apple tends to under-announce and over-deliver on software, and this update appears to be the one that starts making the hardware feel complete rather than impressive-but-limited. The deeper the software gets, the harder it becomes for competitors to close the gap on the experience side even if they close it on specs.

If a robot can build a KALLAX shelf from scratch, it can probably do a lot of other things we currently pay humans to do. Cointelegraph reports that IROS 2026 is throwing down the gauntlet with a humanoid robotics challenge centered on autonomous IKEA furniture assembly, a task that demands vision, dexterity, spatial reasoning, and a tolerance for ambiguity that robots have historically failed at spectacularly.

A 2018 milestone study on robotic furniture assembly showed how hard this problem really is, which makes it the right kind of benchmark: too hard to fake, grounded enough to matter. The field needs challenges like this to stop overpromising and start delivering proof.

The humanoid robot race has been obsessed with making machines look like us, but Genesis is betting that's the wrong instinct entirely. Their new general-purpose robot Eno ditches the uncanny valley aesthetic for something more honest: a capable, non-anthropomorphic body built around practical intelligence.

IEEE Spectrum has documented the growing case that human form factors often work against robots in real-world tasks rather than for them. If Genesis is right, the future of robotics looks less like a scene from Westworld and more like a useful tool that doesn't require you to suspend disbelief to use it.

AMD CEO Lisa Su held a mini PC on stage and made the case that your AI stack doesn't need a data center or a $440/month subscription. As @starmexxx detailed, AMD's Ryzen AI Max+ 395 is the first x86 chip to run a 200 billion parameter model on a single piece of silicon, with CPU and GPU sharing 128GB of unified memory.

A palm-sized box running a 235B parameter model entirely on-device means local AI stops being a hobbyist experiment and starts threatening the SaaS model built on usage fees. For anyone paying per-token for enterprise AI tools, this chip is a negotiating lever that didn't exist six months ago.

Midjourney announced it will reveal its first hardware project at a live event in San Francisco. The company has been purely software since day one, making images on demand with no physical presence whatsoever. We don't know what the hardware is yet, but an AI-native creative company going physical in 2025 is a signal worth tracking. CEO David Holz has hinted at broader ambitions beyond image generation, and owning more of the creative experience end-to-end would be a logical direction. Whatever it is, Midjourney has earned enough trust with its user base to get a serious look.

@the2ndfloorguy used Fable to reverse-engineer Whoop's API, pull per-minute heart rate data, and cross-reference stress spikes with calendar events and attendees, building a personal leaderboard of who's wrecking their nervous system at work. This is the logical endpoint of wearable data meeting workplace self-surveillance. When biometric data gets this granular and connectable, every meeting becomes a health event on the record. The more interesting question is what happens when HR departments start running the same analysis.

A Social Network sequel starring Jeremy Strong as Zuckerberg arrives as Meta faces scrutiny over AI, political alignment, and platform accountability. The film follows an engineer who becomes a whistleblower on Facebook's most guarded secrets, a premise that echoes Frances Haugen's 2021 testimony without being bound by it. The original film defined how a generation thinks about Silicon Valley hubris. A sequel framed around surveillance capitalism and institutional secrecy, with Strong's track record of total character immersion, could do the same for this era.

Brian Armstrong shared a major update from NewLimit, the longevity biotech he co-founded, focused on extending human healthspan by reprogramming the epigenome. This is molecular-level science targeting why cells lose function as we age, not wellness-app longevity theater. Nature Reviews has covered epigenetic reprogramming as one of the most scientifically credible paths to meaningful longevity extension. Armstrong seeded the company with $100M, and the progress they're describing suggests this is no longer speculative. If the science holds, the implications dwarf anything happening in consumer tech right now.

BMW is officially killing the combustion engine in its most iconic performance car, with the new M3 previewed by the M Concept Neue Klasse. Four motors, likely north of 800hp, a ducktail spoiler, wide arches, and an aggressive front end signal that BMW isn't interested in making a polite EV. The Neue Klasse platform is purpose-built for electric performance rather than adapted from an existing chassis. Legacy performance brands have been on trial with enthusiasts since electrification began; this is BMW's closing argument.

The big theme of this week is distribution. Not distribution of content, but distribution of capability: AI running locally on a chip you own, spatial computing moving from expensive headsets to glasses you'd actually wear, and the control layer of software development shifting toward whoever controls the full stack. AMD's mini PC and Cursor's Origin announcement are both versions of the same story. The rent-seeking layer of cloud-dependent AI is under pressure from below, and the developers who own their own infrastructure will have a structural advantage over those paying by the token forever.

Spatial computing had an unusually loud week, and the noise is meaningful. Snap shipping real AR glasses with real specs and a real price is different from a developer preview. Qualcomm building a purpose-made XR chipset is different from recycling phone silicon. XREAL pitching a glasses-first workspace is different from a headset you strap on for demos. All of these are companies making commitments with capital on the line, not roadmap slides. The combined effect is a market that's starting to look less like a science experiment and more like a platform war with clear sides forming.

The Fable story deserves more credit than it got as a punchline. A developer used a browser automation AI agent to reverse-engineer a fitness API, join it to calendar data, and generate personal biometric intelligence from the result, all without a data science team or a single line of server code. That's the actual story. Fable is what happens when AI agents gain the ability to reach across tools, APIs, and data sources that were never designed to talk to each other. The use case here is mildly funny. The next hundred use cases will get more interesting.

Attio is the AI CRM that keeps you ten steps ahead.

Ask Attio anything. Where should I focus? What deals are at risk? Search, update, and create across your customer data.

Ask more from CRM. Ask Attio.

Try Attio for free

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide