In a filing, the agency sided with Elon Musk’s company, saying attempts to stop xAI from running the natural gas turbines “threatens American national, economic, and energy security by seeking to shut off the power supply for artificial-intelligence innovation that supports the Department of War’s military operations.”

The DOJ, along with xAI and the state of Mississippi, asked the court to dismiss the suit, filed by the NAACP in April.

Taft continues:

According to the DOJ memorandum, there are only four artificial intelligence models, including Grok, that “support mission-critical operations across Secret and Top-Secret classified networks.” A separate declaration filed by Cameron Stanley, the chief digital and artificial intelligence officer at the Department of Defense, details how the military relies on Grok’s Gov model to “support vital national security missions.” That includes using the model as part of recent strikes against Iran. Forcing xAI to stop running the gas turbines powering Colossus 2, Stanley says, “directly threatens ongoing national security interests.”

Samuel Hardiman at The Daily Memphian:

“In particular, if Colossus 2 is shut down because it cannot rely on power from the Stanton Road site, xAI would lose capacity to train and develop future improved versions of Grok,” Stanley wrote. “And if xAI is hindered from continuing to improve and upgrade Grok, including the Grok Gov Model, DoW’s ability to meet its national security mission and keep pace with adversaries will be impaired.”

Stanley said that xAI’s data centers in Memphis, including Colossus II, could be relied upon by the U.S. government in the event of another armed conflict or matter of national security.

Here is my rack as it stands today:

I am all in on UniFi gear, as you can see from all the Apple-like aluminum hardware in that photo. From top to bottom, here is how things are set up:

• Cable Modem: My Comcast connection comes straight into this, and then is passed to my network over a 2.5 GbE Ethernet connection.
• Dream Machine Special Edition: This is my gateway. In the UniFi world, this is what manages the entire network, including routing, the firewall, content filters, and more. The Dream Machine can be used to run Protect, Unifi’s home security and camera system, and even record footage with its built-in hard drive, but I have a standalone Protect box further down the stack, as when I initially built this out, I was not using a gateway with its own storage. The Dream Machine uses Comcast as its upstream Internet connection, but it can fail over to my backup ISP — a super-slow AT&T DSL connection.¹ That green fiber optic cable runs to a workshop off my garage, where I have a secondary switch running two of my cameras and one wireless access point.
• A 24-port patch panel with a hodgepodge of keystones.
• My main switch is a Pro Max 24 PoE. Yes, its name is terrible and a rip-off of Apple’s terrible naming, but it’s a heck of a switch. I use its two SFP ports to connect it to the Dream Machine and to my NAS. It has a handful of 2.5 GbE ports that I use for connections at my desk and for my access points. The colors correspond to the link speed of the connected device, which is a handy way to see if something is acting up with just a glance.
• Below the switch is my UNVR, or network video recorder. I have three cameras, and all their footage is stored locally on this device. This is overkill for just three cameras, but I already had it (and had opened it up to install quieter fans) so it is here to stay.
• The next two rack units are a bit deceiving. At the front of the rack is a mere placeholder, but at the back is Unifi’s Power Distribution Pro, which is basically a networked power strip. Combined with the UPS at the bottom of the rack, the Dream Machine can turn off network equipment individually if the power is out and the UPS is nearly depleted. I could do this with just the UPS, but it doesn’t have enough battery-backed outlets for the rest of the gear in the rack, so I run everything through the PDU Pro instead.
• That box full of hard drives is the UNAS Pro, UniFi’s older rack-mountable NAS product. It lacks some of the niceties of UniFi’s newer offerings (like the ability to use an SSD as a cache to speed up file access), but it works well enough for what I need. I should note that this is not something like a Synology that can run things like Docker containers. I just needed a bunch of storage on my network.

I haven’t mentioned it yet, but the rack itself is actually two Toolless Mini Racks stacked together. Had I known I would need so much space, I would not have taken this route, but I got a great deal on the second one, so I stayed the course.

For Wi-Fi, I am using three U6 Enterprise APs. Wi-Fi 6E is plenty fast for my needs, and my first run at Wi-Fi 7 didn’t go super well. To have coverage outside, I’ve got a set of U6 Mesh APs outside — one for the front yard and one for the back.

UniFi’s management tools are pretty great. I can monitor my network, cameras, and NAS from anywhere in the world using either the web or a set of iOS applications.

A common complaint about UniFi is that its hardware often outpaces its software. For example, the ability to shut down and restart devices based on the UPS’s state was pretty broken until just recently. I also have an issue where the network dashboard retains the port assignments even after I move items to a different port.

That said, I love that I own my hardware, that my camera footage is stored locally in my house, and that accessing it doesn’t require a subscription.

Is this pile of equipment overly complicated and expensive for a home user, even if that user works from home? Sure. But we nerdy folk like that sort of thing, and I certainly enjoy the stability and flexibility this setup offers me.

(Oh, the top of the rack houses a Mac mini, a Raspberry Pi running Home Assistant, and a UniFi PoE Smart Chime that goes off when someone rings my doorbell.)

WWDC26 has come and gone, and the guys have scores to pick, topics to discuss, and Liquid Glass sliders to adjust.

I’ve been emulating the Apple Lisa recently, and I was struck by how many of its UI strings were slightly or wholly different than what we’re used to.

It makes sense. Lisa came out in 1983 as Mac’s predecessor and really the first GUI that is directly linked to what we’re using today. Even though it borrowed things from work done at Xerox, tons of conventions were not established yet.

So, I thought it would be fun to actually take a closer look.

What follows is an in-depth look at the writing across the Lisa’s operating system. It’s clear it was designed at a time when everyone was still figuring out how to communicate what would happen when a user performed certain actions.

This is my favorite example from Marcin’s wonderful post:

As Dan Moren writes, this is one of those times, with nods to both touch-enabled Macs and a foldable iPhone.

An unlikely looking flying machine stands on its tail above the watery, thicketed Florida sandscape. The time is the mid-1980s, and the Space Shuttle preparing for launch is one of a fleet of four that now plies routinely, about one round trip a week, between the United States and Earth orbit.

On average, the shuttle would fly once every three months or so, if you count the five years where it grounded for accident investigations after the Challenger and Columbia disasters in 1986 and 2003.

From that opening paragraph on, this document represents a vision that was never close to being realized. To be sure, the space shuttle’s legacy includes great achievements, including the Hubble Space Telescope and the International Space Station, but it could never live up to how it was talked about in the 1970s. It’s a heartbreaking look at a future never realized.

Memphis City Attorney Tannera Gibson said the company has recently assured the city that it would be completing the plant.

“The feelings are pretty positive and pretty strong based on recent conversations,” Gibson said.

Gibson made her comments after being questioned by members of the Memphis City Council about the paused plant.

“We’ve all gotten reassurances, but I want to hear those in public for everybody else,” Memphis City Council member Jerri Green said, referencing private conversations she’s had with SpaceX leadership. Green is a Democratic candidate for Tennessee governor.

Those conversations come after upheaval at SpaceX’s Memphis operations. Brent Mayo, the site’s former leader, is no longer with the company.

One of the first things I noticed after installing the macOS Golden Gate beta was the updated icon design. The colours are much bolder, several icons have been adjusted, and the refraction in the Liquid Glass effect has changed significantly, especially in icons like Journal.

There’s also a noticeable sharpness to the icons, along with a flattening of the Liquid Glass effect. I’m not sure yet whether this is simply an early-beta artifact or the intended final look. For example, while I really like the redesigned Finder icon, the sharp black edges around the nose currently feel a little unrefined.

Here is that change to the Finder icon:

Photos really shows the difference:

I think these look better basically across the board, and I think the change is even better on the iPhone.

I like this wallpaper; it has some Mac OS 9 vibes.

These unofficial ones from Basic Apple Guy are also great.

Apple today confirmed that watchOS 27 will not support the Apple Watch Series 9, Apple Watch Ultra (first generation), or Apple Watch SE (second generation), effectively drawing a line at devices equipped with the S9 or S10 chip.

The only Apple Watch models compatible with watchOS 27 are the Apple Watch Series 10, Series 11, Ultra 2, Ultra 3, and SE 3. Despite the fact that the Apple Watch Ultra 2 contains the S9 chip, the Series 9 which contains the same chip will no longer be supported.

Of note, a bunch of iPads also got the axe, while iOS 27 will support the same devices iOS 26 did, reaching back further than ever before, to the iPhone 11.

Update: It looks like the Series 9 is supported.

David Smith:

Based on my analytics it looks like watchOS 27 will be available for download on around 40% of my Pedometer++ user’s watches. Which feels pretty small. That will certainly be boosted by whatever new devices get announced this fall, but it feels like I’ll be supporting watchOS 26 for a very long time as a huge number of devices will be staying there.

Today’s keynote didn’t include a formal handoff between Tim Cook and John Ternus (who wasn’t even in the video), but Cook did offer some closing remarks. As is his style, it was more about Apple than himself:

On a personal note, some of the greatest highlights of my time as CEO have been events like this. Sharing powerful new tools with all of you and then seeing what you create with them has been a constant reminder has no limits. Over the years you have helped people connect, create, learn, and experience the world in extraordinary new ways, and with the incredible capabilities we introduced today, and so many more still to come, I truly believe the best is still ahead.

At Apple, creating the best products in the world to deliver experiences that enrich people’s lives has always been our North Star.

It’s been the honor of a lifetime to help advance that mission with teams whose creativity, care, and conviction continue to make a lasting difference in people’s lives.

That had broken down over time as more and more features were cross-platform. For example, Apple would announce something in its iPadOS section just for it to popup in macOS. Something in Reminders on the iPhone would also be present on the iPad.

I have been hoping for a change for years, and this year, I got it. After the customary naming of macOS, Apple broke its keynote down into just three major topics, including:

• Platform improvements
• Trust and Safety
• Apple Intelligence and Siri

Platform improvements included much-needed refinements to Liquid Glass (especially on the Mac), better performance across OSes, and some quality-of-life items like updates to cycle tracking in Apple Health.

Trust and Safety was all about better controls and tools for parents to manage their kids’ devices. As someone with two teenagers and a middle-schooler, I am truly looking forward to how these new features will work.

Lastly, the longest section covers Apple Intelligence and the new Siri AI. There is a lot to unpack here, but I think Apple built a case that it can do what it has promised. Google Gemini was mentioned by name, and all of the demos were shot with the presenter using the device in their hands, avoiding some of the issues present two years ago.

It will take some time to sort out what other features are coming in the OS Class of 27, but I like this new direction for WWDC, and I hope it continues in years that it makes sense.

macOS Tahoe will be the last release for Intel-based Mac computers. Those systems will continue to receive security updates for 3 years.

Rosetta was designed to make the transition to Apple silicon easier, and we plan to make it available for the next two major macOS releases – through macOS 27 – as a general-purpose tool for Intel apps to help developers complete the migration of their apps. Beyond this timeframe, we will keep a subset of Rosetta functionality aimed at supporting older unmaintained gaming titles, that rely on Intel-based frameworks.

As such, the compatibility list for macOS 27 Golden Gate is refreshingly simple:

• MacBook Neo (2026)
• MacBook Air with Apple silicon (2020 and later)
• MacBook Pro with Apple silicon (2020 and later)
• iMac with Apple silicon (2021 and later)
• Mac mini with Apple silicon (2020 and later)
• Mac Studio (2022 and later)
• Mac Pro with Apple silicon (2023)

The last couple of machines to ship with an x86 processor were the high-end Intel Mac mini that hung out until January 2023, and Mac Pro (2019) was replaced in June 2023 with the M2 Ultra model. Three years is also how much time elapsed between the final PowerPC Mac being sold and when Mac OS X went Intel-only with Snow Leopard.

The technical trends discussed in this piece suggest that AI systems are going to become much more capable in coming years. These trends have huge implications. AI that can build itself would be a major development in the history of technology—one that could bring enormous good for the world in science, healthcare, and beyond. But full recursive self-improvement also might increase the risks of humans losing control over AI systems. If systems are capable of fully building their own successors, the ways we secure them, monitor them, and shape their behavior all grow much more important.

Then, scrolling waaaaaaay down, they lay out three possible futures that could play out as AI systems become better and faster than humans, especially at coding tasks:

1. The trend stalls, but today’s AI capabilities are widely diffused.
2. AI labs continue to see compounding efficiency gains.
3. AI systems themselves become capable of full recursive self-improvement, and begin building their successors.

They continue:

We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology. The Anthropic Institute will conduct research—in collaboration with many others—and take actions to help build the systems that a credible slowdown or pause would require. These systems would enable frontier AI developers to verify that others globally have actually stopped or slowed, and that a bad actor could not use the auspices of a coordinated slowdown to jump ahead in secret. If such systems existed, we expect that we would slow down or temporarily pause, if other developers at or near the frontier also did so in a verifiable manner.

A meaningful slowdown or pause would require multiple well-resourced labs at or near the frontier, in multiple countries, agreeing to stop under the same conditions. It would also require that each can verify that the others have actually stopped. Due to the unique characteristics of AI systems, the detectability (a lower standard than verifiability) element of this arms control problem is much more challenging than with other technologies. Training runs are far easier to conceal than missile silos, their inputs are general-purpose, and the incentive to defect quietly is enormous, because whoever continues while others pause could inherit the lead. A credible pause also has to specify what triggers it, what lifts it, and who adjudicates.

None of this is necessarily impossible in principle—the world has built verification regimes for other complex technologies (e.g., the Intermediate-Range Nuclear Forces Treaty)—but those regimes took decades to build both the infrastructure and the trust. We don’t have that long. A unilateral pause by one lab, by contrast, is achievable immediately, but accomplishes much less: it would change who the front-runner is, but it would not create the wider deliberative process that is currently missing.

The article closes with a promise to “organize conversations where policymakers, researchers, civil society, and other AI companies can help answer some of the questions this piece raises, especially around full recursive self-improvement and how to create better options for coordination and deliberation” in “the coming months.”

Meanwhile, Anthropic is barreling toward an IPO , even if there’s no clear return. Marina Temkin, Techcrunch:

Private investors have been falling over themselves to get a piece of Anthropic, given the AI model maker is growing at a dizzying pace. Multiple investors told TechCrunch that the company’s $65 billion fundraise at a $965 billion valuation, announced last week, was greatly oversubscribed. Now, with that private demand still strong, Anthropic has revealed that it’s taking steps toward a public listing by filing confidentially for an IPO.

Co-founder Daniela Amodei, speaking at the Bloomberg Tech conference on Thursday, said the decision comes down to capital. “It’s a really big upfront cost to train the models and to serve inference on them,” she said. “My guess is that over time, the sort of core set of companies that are working to advance the frontier are just going to need access to capital, and I think the public market is very well suited to that.”

Temkin continues:

That isn’t fazing Amodei, who believes businesses are still early in figuring out how to deploy AI effectively.

“The use cases today, I expect will continue to be the primary driver of efficiency or creativity, whether that’s coding, financial services, legal, [or] health care,” she said. “But as the business community gets more familiar with the tools, we’re all going to learn together. My hope is that over time it’ll be more incorporated into the day-to-day of how humans do our work, and there will actually be a lot more value realized.”

On one hand, Anthropic is worried about AI systems growing out of our control and wants everyone to take a breath. On the other, it wants a bunch of cash from investors to keep pushing the envelope.

Got it.

SpaceX has lined up another compute deal ahead of its historic IPO, this time with Google. The company announced the deal in a regulatory filing on Friday.

Under the terms of the deal, Google will pay SpaceX $920 million per month from October 2026 through June 2029 for access to “approximately 110,000 NVIDIA GPUs, CPUs, memory, and other related components.”

The deal is similar in length and scope to the one SpaceX announced with Anthropic in late May. As part of that deal, Anthropic agreed to pay SpaceX $1.25 billion per month through 2029 to rent all the available compute from its Colossus 1 data center near Memphis, Tennessee that xAI — now part of SpaceX — originally built for its own artificial intelligence efforts.

Lynn Doan, Bloomberg:

The cloud deal isn’t the only pact that Google and SpaceX have been engaged in talks over. The two companies had been discussing launching the search company’s test products for orbital data centers, a person familiar with the matter said in May. Google previously said it was exploring deals with other launch providers for what the company called Project Suncatcher.

The Anthropic deal was for access to “over 220,000 NVIDIA GPUs,” the entirety of Colossus I.

It’s unknown how many GPUs currently make up Colossus II, xAI’s second Memphis data center, but it seems like Google is leasing a good chunk of what’s available there.

Clearly SpaceX sees leasing hardware to its competitors as more valuable than using it to power Grok. Given that the company will be making north of $2 billion a month between the two deals, it’s hard to disagree.

Notably, Google is building a data center in West Memphis, Arkansas, just on the other side of the Mississippi River from Memphis. That project was estimated to cost $4 billion. It won’t be long before the company pays that amount to SpaceX for rent.

So far, the data it tracks is roughly similar to the Apple Watch, except for when it comes to sleep. If you wake up at any point, Google Health stops tracking sleep for the night and gives you a horrific sleep score:

At least at first. When I checked Google Health a little bit later, it had added additional time to my night:

There’s a lot I don’t like about Google Health, but this goes beyond personal taste. Showing me reliable information when I want it should be the priority for health apps.

We also changed the Bill of Rickies again.

Here’s a post I went to share with my brother:

It’s pretty straightforward — some text and a link with an image preview. You would think that the preview would contain at least some of that information. Well…

The white gradient obscures most of the content in the preview. Technically it does, but in a way that’s clear that Threads wants you to click through to its app or website. Gross.

Microsoft will prevent Office 2019 for Mac owners from editing their documents from July 13, a restriction the company is attributing to the productivity suite’s expiring digital certificate.

The Office 2019 apps affected include Word, Excel, PowerPoint, Outlook, and OneNote. Once the certificate used to confirm the suite’s license expires, these apps will drop into what Microsoft is calling “reduced functionality mode.” In other words, users will still be able to open, view, and print existing documents, but creating, editing and saving documents will be disabled. The same restriction will apply to iPhone and iPad apps that can’t be updated, according to Microsoft.

He goes on:

Microsoft has actually renewed the suite’s certificate, but the fix can only be delivered through a software update. That means users of Microsoft 365 and Office 2021 are in the clear – they’ll receive the update, so neither will be affected. However, Microsoft stopped offering support for Office 2019 on October 10, 2023, and the suite has received no updates since. As such, it won’t be updated to version 16.83, which is the release that includes the renewed certificate.

This is gross and Microsoft should find a way to let these users continue to use the software they purchased.

Nvidia has announced RTX Spark, its Apple silicon competitor for notebooks and small desktops. This puts Nvidia in the same market as Qualcomm building Arm chips for PCs. Here’s Sean Hollister:

“This is the most efficient PC chip ever built,” says Nvidia senior director of product management Mark Aevermann — without sharing so much as a single statistic or chart to back that up.

[…]

Like Apple and Qualcomm’s chips, this Nvidia chip is Arm-based silicon, meaning legacy Windows software made for Intel and AMD’s x86 processors needs to run through an emulation layer to work. That can mean lower performance. But Microsoft has now spent years getting Windows and its Prism emulator ready for Qualcomm and now Nvidia chips, and Nvidia claims its own graphics and AI chops will take the idea further than ever before.

Microsoft, Asus, Dell, HP, MSI, and Lenovo have all committed to building RTX Spark notebooks, but have been light on details. I think most of these will take on the likes of the MacBook Air and MacBook Pro. They’ll come with the same trade-offs as Apple silicon Macs: basically no upgradability, but great battery life, assuming Windows has its act together.

The Windows world is also taking aim at Apple at the low-end, with several OEMs shipping would-be MacBook Neo killers. One such machine is the Dell XPS 13, as covered by Antonio G. Di Benedetto:

Dell is making good on its tease from CES and finally announcing a new XPS 13. The XPS 13 returns as a budget-friendly option, launching in July at a promotional student price of $599 — though that introductory deal only runs until September for back-to-school shopping; it’ll start at $699 for everyone else. The $599 promo exactly matches up with the MacBook Neo’s starting price, but students can actually get Apple’s budget laptop for $100 less. That means Dell has its work cut out proving that the XPS 13 is worth the extra money.

LTT has a roundup of a bunch of these machines, some of which are getting around current RAM pricing by including an empty RAM slot for users to populate themselves, such as the Lenovo Ideapad Slim 5i.

(RIP folks running Windows 11 on 8 GB RAM.)

Much of this coverage misses the biggest point in my mind — the operating system. If I had three $599 laptops in front of me: a Chromebook, a PC, and a MacBook Neo, I’d choose the Neo because I want to run macOS.

If Apple can continue to educate the public about how tightly integrated the Mac is with the iPhone, and why macOS is a superior choice over Windows, the MacBook Neo will continue to be just fine.

Photos of the blast’s aftermath reveal that the accident caused significant damage to the launchpad infrastructure in Cape Canaveral, Florida, destroying a key apparatus used to support and transport the rocket for flights and causing one of two lightning towers to collapse. Other hardware nearby also appeared considerably scorched. The explosion was so destructive that it generated seismic waves.

Rebuilding New Glenn’s sole operational launchpad will likely pose the biggest hurdle to resuming flights. While the company is building a launch site in California, the vast majority of the missions were slated to occur in Florida.

How long it will take to rebuild is anyone’s guess, but even with a billionaire owner, Blue Origin is going to need some time. The before and after photos are pretty rough.

That said, Blue Origin seems hopeful that things won’t be that bad, as Eric Berger reports:

Writing on the social media site X, Blue Origin’s Dave Limp said the company had been able to complete a preliminary survey of the LC-36A launch site.

“Now that we’ve had access to the pad and integration facility, we can share a bit of good news,” Limp said. “The propellant farm, oxygen, liquid hydrogen and LNG tanks are all in good shape. This is good luck because these are very long lead items. The water tower is also good.”

Berger continues:

Limp also confirmed that the company would press ahead with a rebuild of the LC-36A site, which is designed for the 7×2 variant of the New Glenn rocket. One option had been to focus on building a larger pad next door, at LC-36B, capable of supporting the larger 9×4 variant of the rocket (the nine and four, respectively, refer to the number of engines in the first and second stage of the rocket).

Notably, Limp also said Blue Origin had a plan to replace the massive transporter-erector that moves the New Glenn rocket from its nearby integration hangar out to the launch pad. This was damaged beyond repair during the test failure on Thursday, May 28.

My pals Myke and Jason are launching a new project:

The two of us, Myke Hurley and Jason Snell, have been discussing Apple in depth every week for more than a decade on the Upgrade podcast. We want to go further, creating a podcast series that’s dedicated to telling stories from across all five decades of Apple history. This is a big reach for us, requiring a significant outlay of our time and the time of our audio editor. We have the knowledge and skills—but we need your help to make it happen.

Designed in California will consist of at least 30 episodes dropping over 12 months, looking at a wide range of Apple history, drawing from the best sources available.

Jason Snell at Six Colors:

We’re using Kickstarter for this project because researching and writing these scripts is quite labor-intensive, and I was hesitant to make that time commitment in the hope we would eventually build up enough of an audience to justify the large workload. We’ve set a goal that would allow us to generate thirty 30-to-45-minute episodes over the course of a year, with our first stretch goal to raise that number to a full fifty episodes in a year.

This is going to be very, very good.

Stephen Hackett returns to the show to help me make sense of a problem that is hard even to describe, let alone solve, regarding Apple One, family sharing, a separate Media & Purchasing account, and iCloud storage space. Also: what we’re expecting from WWDC 2026.

These tests are designed to test the rocket’s fuel system and motors by firing them when the rocket is clamped down to the launchpad. Thankfully, it seems that no one was hurt.

Eric Berger has more at Ars:

The company has launched New Glenn three times, during each of which the first stage performed well. The company had already demonstrated the ability to land the New Glenn first stage, and impressively reused it in April for the first time.

During that third flight, carrying the Blue Bird 7 satellite, an upper stage issue caused the mission to fail. However the company responded rapidly to the in-flight failure and returned to the launch pad in less than two months this week. The first stage for this mission, nicknamed No, It’s Necessary, was making its debut launch.

Prior to this launch attempt Blue Origin had in its inventory two first stages and about six New Glenn upper stages completed, and it was poised to break into a monthly launch cadence. By all accounts, the rocket was viewed as a major success for a company which, for so long, had seemed to plod along. New Glenn’s success catapulted the company to the upper echelon of spaceflight enterprises in the world. That Blue Origin was on the precipice of accelerating further makes this setback all the more painful.

It’ll take some time to learn what happened to this New Glenn rocket, but the impact of its failure may be wide-reaching.

This is a potential setback to NASA, which just this week announced Blue Origin will play a strategic role in its every-changing plans for a lunar return.

This is also a potential setback for United Launch Alliance’s Vulcan rocket, which uses the same BE-4 engines that may be at the center of last night’s disaster.