Starting August 24, Walmart is expanding its payment options with tap-to-pay, after years of not allowing payment services like Apple Pay and Google Pay.

Walmart quietly announced the news in a press release on Friday, going out of its way not to name the services that the new tap-to-pay system will allow. We have confirmed that Apple Pay, Google Pay, and Samsung Wallet will all work with the new system.

In it, the company says that the initiative will be complete in Walmart and Sam’s Club stores by the end of 2026. Gas stations will get the service in 2027.

It’s been a long time coming

For years, Apple Pay and other competing services weren’t allowed at Walmart because the company had its own payment funnel it preferred. The company forced shoppers to scan Walmart Pay QR codes that it provided, so the company could better track consumer spending habits and schedules.

That system, which will still be functional after the tap-to-pay rollout, requires a customer to add a payment card to Walmart’s app. Then, a QR code is presented to the consumer to scan at checkout.

Walmart has accepted Apple Pay in Canada stores for six years.

Apple Pay launched on October 20, 2014. Walmart very briefly allowed it, before cutting off access and shifting to Walmart Pay exclusively.

Most major retailers now accept Apple Pay, as well as other forms of NFC-based payment. Even some notable holdouts like Midwest grocery retailer Kroger and home improvement chain The Home Depot have begun accepting Apple Pay in recent years, both after years of resistance.

Walmart is one of the last big holdouts.

Apple’s Irish tax arrangements were scrutinized for more than a decade, and has now, finally, culminated in one of the largest corporate tax recoveries ever ordered by the European Union.

Apple has been fighting a decade-long battle over how much it owed Ireland in corporate taxes. The company was accused by the European Union of exploiting a loophole to avoid reduce its tax liability, despite employing more than 5,000 people in the country.

Now, new filings, seen by The Financial Times, show that Apple paid $17 billion in taxes to Ireland in 2025. Globally, the company had paid $43 billion, which means Ireland accounted this one time for roughly 40% of Apple’s corporate tax bill for the year.

The payment was bolstered by the European Union’s back-tax ruling. The ruling found that Ireland had given Apple “unlawful aid,” resulting in a tax rate of less than 1%.

For years, Ireland opposed the EU’s conclusion.

Timeline of the Apple tax case

The case is the largest corporate tax recovery in EU history. It centers on the “Double Irish” strategy, which let Apple route profits through subsidiaries with no tax residency.

• 1991 and 2007: Ireland’s tax rulings allowed Apple’s Irish subsidiaries to attribute most profits to head offices that were not tax-resident anywhere.
• 2013-2014: EU began investigating.
• August 2016: Against Ireland’s wishes and testimony, the European Commission ruled Apple received illegal state aid, ordering repayment of about $14.2 billion (13.1 billion euros) plus interest.
• 2018: Ireland collected the full amount into escrow while appeals proceeded.
• July 2020: EU General Court annulled the Commission’s decision.
• September 2024: European Court of Justice overturned that ruling, siding with the Commission.
• 2025: Ireland confirmed it received nearly $15.5 billion (14.25 billion euros) from the account’s final closure.

Apple has consistently argued it paid the taxes it owed and complied with Irish and international law. The company has also emphasized that most of its profits were taxed in the United States when repatriated.

Ireland opposed the European Union on this matter for years. The country still markets itself as a low-tax destination but must navigate growing pressure to follow global tax reform rules as part of the EU.

The xTool M2 is priced at the entry level of the company’s fully enclosed machines, but as ever there’s no mistaking it for anything other than an xTool laser engraver. As I got into using the machine, I could see just how finely tuned it is. Whilst it’s lighter weight than many of its more expensive siblings, the laser engraving power, even from the 10W laser in my review sample, was exceptionally good.

Having reviewed these laser engravers for a few years now, there are several essentials I look for in any laser engraver for TechRadar Pro, and one top priority is that it’s fully enclosed. While I might’ve looked at fully open designs earlier on, these days there’s really no excuse for an open-frame design other than cost. When it comes to running a business, safety is obviously paramount, so it makes sense to look for a fully enclosed rather than open-frame design.

Inherent, a London AI lab founded by Google DeepMind alumni, says its AI agent just outperformed much larger models from Anthropic and OpenAI using a fraction of the size.

Of all the startups launched by Google DeepMind alumni, Inherent has gotten relatively little attention. But while better-funded rivals have yet to show the world anything concrete, the London-based team is starting to share what it’s been building.

Just weeks after emerging from stealth with a $50 million seed round, the British startup says its newly released AI agent, Faraday, has outperformed larger, better-known models at a specific task: independently reproducing the findings of published scientific papers without being told the answer in advance.

That may sound like a mere party trick given Inherent’s much loftier goal — building AI that can discover new scientific knowledge and not just verify old results. But paper replication is a standard training exercise for human scientists, too, cofounder and chief scientist Edward Hughes said. “Many PhD students actually start by doing this.”

Beating other AI systems at the task wasn’t the point, Hughes told TechCrunch; how they got there was. “What was most interesting to us about this was not so much the result of beating those frontier agents — which of course we liked — but was actually the way we went about building this.”

Here’s the part that should catch an investor’s eye: measured against Anthropic’s Claude Opus 4.8 and OpenAI’s GPT-5.5 — both much larger, frontier-scale systems — Faraday runs on a comparatively tiny model called Qwen 3.6 that has just 27 billion parameters. (Roughly speaking, “parameters” is a proxy for a model’s size and, typically, its training costs, as well.) Inherent’s bar for success was also higher than simply accuracy. Beyond replicating results, it wanted Faraday to demonstrate “research taste” — an instinct for what experiments are worth running and how to design them well.

Teaching something as intangible as taste is hard, which is where reinforcement learning comes in. It’s a training method that rewards an AI system for good outcomes rather than spelling out rules for it to follow. Rather than training its agents primarily on the study of how science itself is conducted, Inherent leans on this reward-based approach, betting it will generalize better to its longer-term goal of agents capable of contributing across many scientific fields.

“We’re always guided by that north star of building an AI scientist agent and imbuing our agents with taste,” Hughes said. That focus has also shaped what Inherent chooses not to build. Rather than developing its own coding tool, it had Faraday use OpenAI’s GPT-5.5 Codex instead, much the way human scientists lean on existing software rather than building everything themselves, according to the company.

Inherent is also trying to avoid building agents that simply tell users what they want to hear. Instead, Hughes said, the goal is modeled on his favorite kind of teammate — the kind who comes back and says: “I got curious about this, and I went off and I did these experiments. What do you think of these results?”

That collaborative instinct extends to how Inherent operates as a company. Its dozen employees all work in person out of an office in King’s Cross — the once-rundown London neighborhood that Google DeepMind’s presence helped turn into one of the world’s top AI hubs. “We believe that London is the place to be,” Hughes said.

Hughes is bullish on London’s density of AI talent, but he has also added his voice to calls to end “garden leave” — the practice, common in the U.K., of barring departing employees from joining or starting a rival company for months after they resign. It’s a restriction American researchers generally don’t face, giving U.S. startups a head start on hiring talent who’ve left a prior role. “This is a personal view rather than a company view, but I was affected by the garden leave problem,” he told TechCrunch.

Hughes eventually got around that constraint and started Inherent alongside two other DeepMind alumni and a fourth cofounder. The startup isn’t slowing down either. It plans to grow its headcount to “about 20 to 25” by the end of the year. Given its ambitions in world models as well, and with Demis Hassabis’s new role leaving some DeepMind staff unsettled, Inherent’s hiring push could make it an appealing landing spot for DeepMind employees weighing a move.

Pictured from left to right: Inherent co-founders Louis Kirsch, Kaloyan Aleksiev, Tantum Collins and Edward Hughes.

There’s a new publicity stunt tapping into widespread opposition to AI data centers, and it’s using bathroom humor to make a statement.

Former NFL player Jason Kelce, who is also the co-owner of Garage Beer and the co-host of the New Heights podcast with his brother Travis (Taylor Swift’s husband), is the unlikely face of a marketing campaign, launched Tuesday, that puts artificial intelligence on notice. And, well, it’s all about the power of pee.

You read that right, and I apologize in advance for the amount of urine jokes in this article. Tread lightly, I guess.

The now-viral ad opens with Kelce using the bathroom. It cuts to him holding a jar of yellow liquid, issuing a call to action to a happy mob of folks frolicking through grassy terrain, singing, “Let’s pee on computers together and save humanity.” It immediately conjures up that iconic 1970s Coke commercial where hippies harmonize a message of peace, love and unity on a sunny hillside. Except instead of teaching the world to sing, Kelce is asking you to bottle up your pee in a collectible mug and send it to the AI data center of your choice.

If you’re not yet up to speed on the matter, AI data centers are large facilities that store the massive amount of hardware needed to train and run AI models. These warehouses are stocked with high-performance graphics processing units, which provide the power for you to use AI for researching, streamlining video production and, well, creating all the slop your heart desires.

Despite the unserious angle of the marketing stunt taking the piss out of AI (I’m sorry), Kelce’s ad carries a deeper message about the threat data centers pose to our neighborhoods, our resources and our future.

AI data centers guzzle resources

There are nearly 4,000 AI data centers in the country, currently. A database created by environmental activist Erin Brockovich shows where they’re located, with the majority being built in rural areas.

To keep data centers running, an excessive amount of fresh water is required to cool servers so they don’t melt down or explode. Evaporative cooling towers use water to absorb heat generated by servers.

The Environmental and Energy Study Institute published a report in 2025 that says large data centers can consume up to 5 million gallons each day, which is equivalent to the amount of water used by a town populated by 10,000 to 50,000 people.

As massive amounts of potable water are being redirected, people who live near these data centers have growing concerns about how this water extraction will affect their daily lives, farming and basic household needs. Another fear: Who will be stuck paying these ginormous utility bills? Consumers are left holding the bag.

The other resource to discuss is electricity. New facilities, like the recently announced data center from OpenAI and Nvidia that would be built in Pike County, Ohio, are projected to use a lot of compute. The Ports-Pike Technology Campus is projected to consume roughly 8 gigawatts of power, equivalent to the annual energy use of 6 million average US households.

Then there’s Meta’s Hyperion, the 4-million-square-foot AI campus being constructed in Louisiana. It’s flattened an area known for its soybean crops and river cane, and when it’s all built and operational, the Institute on Taxation and Economic Policy says the gargantuan data center will consume three times the power New Orleans uses on any given day.

In other words, more than a piss load of energy.

AI backlash across party lines

Resource consumption, uncertainty over costs and environmental concerns have helped fuel a growing AI backlash.

So, antics aside, Kelce’s trolling ad touches a hot-button issue. The collective sentiment against AI data centers is trending downward. A recent Gallup poll indicated 71% of Americans oppose the construction of new AI data centers in their community.

With the midterm elections a few months away, this negative sentiment has blossomed into a major talking point for both Republicans and Democrats in their campaigns. Another major issue with these data centers is a glaring lack of transparency.

As reported by Brockovich in a blog post, a trend of backdoor deals, elusiveness and secrecy has accompanied the facilities as they pop up all over the country, with the trust of community members tanking. “They’re watching their utility bills climb, finding sick animals they can’t explain, and worrying about the long-term impacts on their health and property values,” Brockovich wrote.

This brings me back to Ohio, an important state for the GOP during the midterms if the party wants to maintain a slim Senate majority. A memo circulated by the National Republican Senatorial Committee warned of a potential defeat for incumbent Sen. Jon Husted unless there’s a fix to the public relations strategy toward AI data centers.

“If voters’ perceptions of data centers are not fixed quickly, the campaign against them will expand far beyond Ohio,” the NRSC stated in the memo, as reported by Axios.

Real-life action versus promotional gimmicks

I now return you to the matter of, ahem, the general public sending urine to an AI data center to cool the machinery of the future.

According to the Cleveland Clinic, urine is 91% to 96% excess water. Heck, there’s a reason Saul Goodman survived the desert in that one Better Call Saul episode. Watch any number of programs featuring survivalist Bear Grylls, and he, too, will tell you this. But I digress.

The marketing ploy here is to equate pee with water and, considering the massive water consumption that keeps AI data centers running, put the public disdain into the type of prank (which you’d normally see in a Jackass episode) to make its point loud and clear.

But you could also argue that it makes no point at all. Ordinary people (those without Super Bowl rings) who have protested data centers IRL have gotten arrested for it. There is real work to be done policy-wise. A trolling ad like this fits into the performative and parasocial attention economy, not into genuine activism.

Remember, this is a commercial for Liquid Death and Garage Beer — a product is being sold here. A collectible mason jar, which is currently sold out, is listed on the Liquid Death website for $18. The product is currently rated five stars, with a single review saying, “the perfect product for shipping your locally sourced urine to your friendly neighborhood data center.”

So, are people actually sending their excreta through the mail? A spokesperson from the United States Postal Service didn’t immediately respond to our request for comment.

That said, USPS rules say urine specimens can be sent by mail, but only for medical purposes such as drug and alcohol testing. To do this right, you’d need to carefully package the fluid in a leakproof container, then place it in appropriate packaging to prevent a messy spill.

Truth is it was never meant to be an action at all. Small print at the bottom of the ad says, “The suits want us to tell you to please don’t actually send your pee.” So yeah, Liquid Death and Garage Beer are really just pissing in the wind.

One thing is certain: Jason Kelce’s reach is strong, and his blue-collar persona can help give this campaign — and the message behind it — legs in rural America, the exact place battling against the spread of water-guzzling AI data centers.

Aaron is a writer for the streaming entertainment team. Previously, he wrote about TV and movies for places like Rotten Tomatoes, Inverse, TheWrap, Washington Post and The Hollywood Reporter. Aaron is also an actor and played a nerd on TV a lot in the ’90s. Seriously.

See full bio

Advertisement

• Bluecore’s floating reactor could power about 15,000 homes near the port
• California has banned new nuclear reactor construction since 1976
• The reactor would sit on a barge leased at Berth 48

The Port of Long Beach is exploring small modular reactors that could eventually supply electricity for port operations and ships.

Bluecore Energy, a startup building compact nuclear reactors meant to operate from floating barges, is leading the proposal.

SYSTEMS

Mercury Research blames costly memory and scarce GPUs for 20% processor shipment slide

The processor market is sending mixed signals, with server and mobile shipments rising while desktop CPU volumes decline amid higher system prices. Meanwhile, AMD’s House of Zen has taken market share from Intel across every category.

For Q2 2026, Mercury Research says total processor shipments were lower than in the same period a year ago, attributing this to much lower system-on-chip (SoC) and embedded volumes due to AMD’s declining games console business, plus a large drop in desktop CPU volumes.

Mercury associates the decline in desktop chips with weaker demand for high-end gaming PCs. Although the second quarter is not typically strong for consumer sales, it adds: “We believe that higher PC prices and limited GPU supplies are having a significant impact on end demand for desktop PCs, and thus desktop CPUs, as well.”

Those higher PC prices are the result of increases in the cost of memory components due to a shortage caused by chipmakers prioritizing output of more profitable high-bandwidth memory chips used in AI servers, as The Register has been covering for some time. A shortfall in the availability of consumer GPUs appears to have much the same cause.

According to Mercury, desktop CPU shipments fell by more than 20 percent year on year, although AMD’s decline was smaller than Intel’s. As a result, AMD gained market share, taking nearly 35 percent of desktop chips compared with about 32 percent a year ago.

In contrast, shipments of mobile processors for laptops and tablets were up strongly on the previous quarter, running counter to Mercury’s earlier expectations, although there was only a modest increase compared with a year ago.

The growth followed a sharp increase in Intel’s output, particularly of mobile chips, after two heavily supply-constrained quarters. Intel added millions of units of mobile CPU capacity during Q2, significantly narrowing the gap between supply and demand.

However, AMD’s share of this mobile segment is now up to nearly 29 percent, a significant increase from the 20.6 percent it stood at in the same quarter a year ago.

Server processor shipments also rose, increasing 20 percent year on year and more modestly from Q1. According to Mercury, demand was higher for both datacenter-class CPUs (Xeon SP and AMD Epyc) and chips aimed at networking and storage applications.

Once again, AMD gained market share, accounting for 34.5 percent of server processors compared with 27.3 percent a year ago. Mercury adds that if the calculation included only Intel Xeon SP and AMD EPYC chips, AMD’s share would reach 46.4 percent.

The research firm also keeps an eye on the Arm-based CPU market for PCs and servers, with the usual caveat that its estimates have significant uncertainty as there is no centralized reporting of Arm server or client CPU revenues.

It recorded significant growth in Apple’s Mac products, including the new lower-cost Neo line, while Arm-based Chromebooks also posted strong gains.

Mercury estimates that Arm-based systems captured 15.3 percent of the client market in Q2 2026, up 0.9 percentage points to a record high. In servers, it estimates that Arm’s share reached a record 13.6 percent, up 0.5 percentage points. ®