Humanoid Robots Meet the Performance Review
Congratulations on the backflip. Now finish your shift.
Humanoid robots face a broader factory and warehouse test in 2027, with logistics giant GXO planning to deploy Agility Robotics’ Digit at two more US facilities. GXO says Digit has already moved more than 100,000 totes at its Georgia warehouse, proving it can do more than just star in viral tech demos. The new assignments are expected to start in early 2027.
BMW offers a different benchmark. The automaker says Figure AI’s Figure 02 robot handled over 90,000 components during a ten-month production pilot. Figure 03 has since demonstrated more complicated parts sorting there, but comparable sustained results haven’t been disclosed. The successor has a résumé to live up to before it earns a permanent badge.
Elsewhere, onboarding is practically an industry. Elon Musk says Tesla needs 10,000–20,000 robots practicing tasks in an Optimus Academy to generate training data. He targets outside business deliveries around late 2027, with hands, durability, and AI still needing work—minor details, really.
GM’s chief product officer, Sterling Anderson, challenges the premise itself. He told The Verge humanoid advocates put “form over function,” favoring specialized machines tailored to factory tasks. That complicates the industry’s sales pitch. Even a capable robot needs a reason to look like us when a cheaper, ceiling-mounted octopus arm might work just fine.
Buyers need useful output at an acceptable cost once maintenance, charging, integration, and human intervention are included. The real 2027 contest is whether humanoids can justify their complexity against existing automation. A robot’s resemblance to the employee badge photo won’t settle that.
XREAL AURA Puts Your Office on Your Face
Your office has found another way to follow you home.
Google and XREAL have launched AURA, wired Android XR glasses that slap floating workspaces and Gemini AI right onto your eyeballs, starting at $1,099 through December. Shipments are expected to begin in mid-November, assuming you want to dodge the regular $1,279 price tag.
The frames weigh under 95 grams, mostly because the battery and heavy compute components have been banished to a separate, cabled puck. So while your face enjoys the weight reduction, your pockets inherit a tethered brick. On the bright side, the transparent lenses let you stare directly at your surroundings without a camera feed.
The real parlor trick is hoarding your work everywhere you go. You can plug in a compatible laptop via USB-C to cast your screen without extra software, arrange Android apps around it like digital wallpaper, and connect wireless peripherals. Google claims Gemini can interpret what you’re working on—even inside that casted desktop window—to visually highlight details. Congratulations, your spreadsheet now has an AI backseat driver.
For the beleaguered IT crowd, there’s an actual enterprise angle: Google says Android Enterprise brings zero-touch enrollment, managed application distribution, and remote device policies to AURA. That could make mass deployments easier to administer, though the actual comfort and workflow benefits remain to be seen. Because heaven forbid you leave the desk without taking the corporate rulebook with you.
When you finally log off, more than 150 XR experiences join your standard Google Play apps. And just in case you thought you could fully escape reality, Google programmed the dimmed lenses to automatically turn transparent the second someone walks into your view.
Even your private virtual theater can’t protect you from “got a minute?”
DoorDash Orders Instacart for $17 Billion
DoorDash has agreed to buy Instacart for a cool $17 billion, mashing up its restaurant-delivery dominance with Instacart’s grocery-aisle prowess. Assuming regulators don’t toss the deal out of the cart, it’s slated to close sometime between late 2027 and early 2028.
The corporate pitch is all about survival of the fittest against retail apex predators like Amazon and Walmart. By combining logistics, ad networks, and AI tools, DoorDash claims it can arm local grocers for battle without forcing them into exclusive delivery marriages. So, Main Street gets a shiny new toolbox—courtesy of a massive, consolidated gatekeeper.
On the consumer side, the Instacart app survives the merger, while DashPass and Instacart+ will merge into a single subscription. DoorDash estimates $100 million in annual membership savings collectively, unlocking 50,000 extra stores for Instacart regulars. It’s a definite win if you were paying for both, though don’t hold your breath for cheaper avocados.
Drivers are being wooed with promises of endless order options, and Instacart loyalists can choose to stick solely to grocery runs. Naturally, guaranteed higher wages aren’t part of the announcement. Antitrust lawyer Ethan Litwin points out this team-up could trap gig workers with fewer alternatives, proving that more deliveries don’t automatically equal more bargaining power.
Over in New York, Mayor Zohran Mamdani is already ringing the alarm bells, blasting the buyout as a recipe for “higher prices, harsher exploitation, and greater economic inequality.” He happily pointed out the glaring irony of DoorDash coughing up a recent $131.5 million worker-pay settlement with the city just before finding $17 billion in the couch cushions for a mega-merger. Clearly, the corporate shopping budget remains untouched.
Please review the competition before proceeding to checkout.
The MacBook Pro Finally Gets Touchy
Apple has finally approved fingers for professional use.
Apple unveiled its first touchscreen MacBook Pro yesterday, pairing a PROLED XDR display with M5 Pro or M5 Max chips. The 14-inch starts at a hefty $2,899; the 16-inch at $3,399. Maxing out the specs pushes those numbers to an absurd $14,249 and $14,549, respectively—just in case you were deciding between a new Mac and a decent used car. Preorders are open, with availability beginning Nov. 6.
After years of insisting that poking a laptop screen is ergonomically terrible, Apple has suddenly seen the light. Its new philosophy? Keep the keyboard central, fatten the menus for clumsy digits, and stiffen the hinge to stop screen wobble. The long-standing philosophical objection now comes with a shipping date.
Early testers are drooling over the screen while mostly shrugging at the touch input. Ars Technica’s Samuel Axon praised the absolute blacks but noted that pawing the display mostly duplicates trackpad gestures. WIRED’s Steven Levy admired the visuals and noticed UI elements politely enlarging for fingers.
The caveats? It isn’t a convertible, and the Apple Pencil is strictly forbidden. While macOS generously interprets taps as basic mouse clicks, true touch-first optimizations are still rolling out. Apple’s Creator Studio patch lands in November, with third-party developer updates dragging in by year’s end.
Apple is blurring the sacred boundary between Mac and iPad while preserving desktop workflows. For creative pros, the gorgeous OLED panel might justify the massive upgrade tax before the touch features ever do; test your apps before expensing Apple’s sudden change of heart.
The Midterms Have an AI Verification Problem
Even autocomplete needs a campaign manager.
Artificial intelligence is officially on the ballot for the Nov. 3 midterms, juggling jobs as a synthetic ad generator, a hallucinating voter guide, and an occasional security headache. But let’s burst one panic bubble right away: Georgia officials confirmed Anthropic’s Claude retrieved free public statistics without actually touching voter rolls or jumping a paywall.
Beyond those accidental data scares, AI has become a deliberate target for political spin.
Campaigns are aggressively trying to influence chatbot recommendations so your digital assistant votes for their guy. It’s a bold strategy, especially since the States United Democracy Center found that ChatGPT omitted candidates in 88.9% of gubernatorial queries during early-year tests. A confidently written answer is still a terrible voter guide if it casually forgets who is running.
In a frantic bid to keep things sane, OpenAI is launching an election war room and forcing ChatGPT to render election-interference prompts as goofy cartoons. CNET found the protection held; Gizmodo easily found a bypass by simply asking for photorealism. Clearly, Big Tech’s safeguards are more like speed bumps than brick walls.
It’s not all doom and gloom. An Election Assistance Commission guide documents AI-assisted notices, training, and poll-worker scheduling to help swamped local offices. The catch? “The chatbot ate my homework” doesn’t hold up in election law, so humans still have to check the outputs.
Separately, an attempted hack triggered the FEC’s electronic-filing shutdown, abruptly blinding the public to late-stage campaign finance disclosures. While no AI link is established—and actual ballot counting is entirely unaffected—it proves transparency is fragile right when we need it most.
Bottom line for voters? Double-check your deadlines and polling places with actual human election officials. For everyone deploying AI, preparation means constantly checking outputs and controlling actions, not just applauding corporate safety announcements.
Your chatbot doesn’t get a vote. Don’t give it yours.
25,000 Job Cuts, AI Details to Follow
The Netherlands is trimming the bureaucratic fat, planning to cut roughly 25,000 full-time government jobs over the next decade in an AI-led efficiency crusade. The goal? Pocketing €900 million (about $1 billion) in 2027 and €3.5 billion (roughly $3.9 billion) annually by 2035. Police and defense are being spared the algorithmic ax.
Exactly how this digital utopia will function remains a mystery, with official AI blueprints not arriving until early 2027. Which systems will handle the grunt work—and whether this means brutal layoffs or just natural attrition—is anyone’s guess. For now, the budget spreadsheet is leading the charge.
This overhaul goes beyond slapping a chatbot on a public website. Officials want to slash 500 regulations annually, standardize IT, and force agencies to finally share data. Even physical office furniture isn’t safe—the government is mandating a game of musical chairs with a target of one workstation per two employees.
While dire for civil servants, Channel Insider notes this could be a goldmine for technology partners. But the invisible costs of babysitting those algorithms—licensing, endless integration, and human oversight—will undoubtedly chew into those projected savings. Slashing headcount doesn’t magically guarantee cheaper government operations.
Then there’s the inevitable workload boomerang. A recent Bloomberg analysis highlights “agentic flooding,” where citizens use their own AI to mass-generate complaints. Automating the clerk behind the counter is essentially pointless if the public automates the mountain of paperwork thrown at them. Success ultimately hinges on whether citizens can get human help when the bots inevitably break.
Apparently, bureaucracy is perfectly capable of scaling on both sides of the desk.
Hackers Just Stole $6.3 Billion in Crypto
Billions left Binance’s wallets, and the way in was one engineer’s laptop.
- Hackers tied to North Korea are blamed for stealing at least $6.3 billion, the largest crypto theft on record. About half of it was frozen as it moved across blockchains.
- Binance says attackers used a compromised laptop to reach a service container and submit unauthorized withdrawals. It reports no sign that private keys leaked.
- Binance has suspended deposits and withdrawals and says user funds are safe, though it hasn’t shown proof yet. An updated proof of reserves is promised after cleanup.
TRM Labs clocks the whole theft at about 20 minutes.
Creators and builders: this week, check which laptops can reach your production keys or payout systems, and cut that access down.
Anthropic Is Now SpaceX’s Biggest Customer
Elon Musk trashed Anthropic for months, then became its landlord.
- Musk publicly said he was clearly wrong about Anthropic, called it the current AI leader, and praised its Mythos and Fable models.
- Anthropic pays SpaceX $1.25 billion a month for 300 megawatts at the Colossus 1 data center, and either side can end the deal with 90 days’ notice.
- Anthropic has also said it’s interested in working with SpaceX on space-based orbital data centers.
How friendly can a rival be when it hosts your servers?
Creators and builders: this week, check which compute your main AI provider actually runs on and keep a backup model wired into your workflow.
Google Just Put Its First Data Center in Space
Google just put a TPU in orbit and asked it to write a haiku about being there.
- The Suncatcher prototype launched earlier this month on a Falcon 9 and spent about three weeks in commissioning. Its first prompt this week went to Gemma 4, which answered with a haiku about starlight feeding its core.
- The satellite makes up to 1 kW, so the chip can only run about 20 minutes at a time. The test campaign now checks whether radiation flips bits and corrupts data without tripping any monitoring, a risk ground tests suggested hits training harder than inference.
- Two more satellites launch in 2027 with higher power and optical links between them. Scaling beyond that needs Starship fully reusable, which Suncatcher researchers say could take about 1,800 flights.
A 2030s bet, with a haiku as the proof of life.
Creators and builders: this week, run Gemma 4 on your smallest workload and see how much of it actually needs a frontier model.
Reddit Says Google’s AI Is Killing Its Traffic
Reddit CEO Steve Huffman says Google’s AI search is sending far fewer people to Reddit, and the company will start making visitors sign in to read most content.
- Google search referrals fell by more than 1 million visits a week on average in October.
- Huffman said Google’s AI Mode “does not drive traffic, not for us and not to anybody,” even though Google is crawling Reddit as hard as ever.
- Reddit will test a logged-in-only experience in some markets outside the US this quarter, ending about 20 years of open browsing.
- US daily users fell 2% to 50.7 million, Reddit’s first drop since going public. Shares fell 18% after hours.
- Revenue still grew 53% to $895 million, helped by data deals with OpenAI and Google.
When Google answers the question itself, people stop clicking through to the sites that wrote the answer.
If even Reddit is locking its doors, expect more of the open web to go behind logins, making it harder for AI tools and casual readers alike.
Google’s AI Spending Heads Past $300B
Alphabet reported $127.6 billion in quarterly revenue, up 25%, as its AI business grows fast but costs even more to build.
- Google Cloud revenue more than doubled to $30.7 billion, driven by AI demand.
- The Gemini app passed 1 billion users.
- Google’s models now process nearly 32 billion tokens per minute, 4x more than a year ago.
- Alphabet spent $51.4 billion on infrastructure this quarter and raised its 2026 forecast to $195 to $205 billion.
- It signaled spending will top $300 billion in 2027.
Google is betting the company on AI, and users are showing up.
But the bill is growing faster than the profits, and that’s the question investors will keep asking: when does all this spending pay off?