Palmer Luckey, the founder of military tech startup Anduril (also the founder of Oculus), quietly teased a pair of unannounced AR glasses that are presumably in development at Anduril. Despite their incredibly compact appearance, Luckey says they are not merely smart glasses, but are indeed built for full AR capability.

Just days after Snap Inc revealed its latest Specs standalone AR glasses, a photo shared on X showed Anduril founder Palmer Luckey wearing what appeared to be some kind of clip-on attachment to his sunglasses. The caption of the photo, shared by Lulu Cheng Meservey, included the quote “Other AR glasses getting literally frame mogged.” This ostensibly referred to Snap’s Specs glasses, which drew significant commentary about whether their design was too bulky to be socially acceptable.

Regardless of where one stands on the size of Specs, the glasses shown in the photo of Luckey appear incredibly compact by comparison. This prompted Road to VR founder Ben Lang to question whether it was actually a fair comparison to make; Lang pointed out that surprisingly compact smart glasses—like the Even Realities G2—are already available on the market today, but they remain compact because they lack the full hardware stack required for true AR glasses.

Luckey responded directly to Lang’s scepticism, saying “[…] these are the real deal! No shortcuts, just really good tech in conjunction with deeply invested partners like Qualcomm, Meta Quest, and [Andrew “Boz” Bosworth, CTO at Meta].”

When asked by Lang if the device pictured is capable of true AR, Luckey added, “correct, [the glasses have] wide-FOV full-color stereo.”

Luckey’s confirmation that the device is capable of proper AR is indeed surprising considering the size. Significant questions, however, remain. Chief among them is whether the device is fully standalone or if it relies on a tethered puck (which would serve to offload battery and/or compute), allowing the device to be much smaller than a pair of standalone AR glasses like Specs.

Given the incredibly compact size of the glasses, a tethered architecture seems like the only possible explanation. However, neither the original photo published by Meservey, nor a follow-up photo published by Luckey, show any clear signs of a tether:

Another major question is whether the device is made as a modular add-on to existing off the shelf glasses. The design of the hardware suggests this as a significant possibility (indeed, the glasses pictured look like a regular pair of Ray-Ban Aviators). But making a simple modular add-on to existing glasses introduces significant complexities for the calibration of optics, displays, and tracking hardware.

Far more is unknown about this device than is known at this stage. But there’s reason to believe it’s legitimate, and possibly connected to Anduril’s EagleEye project, which aims to build an augmented reality helmet to bring superhuman perception to soldiers on the battlefield. In 2025, Anduril announced it had partnered with Meta on XR technology for the military, which aligns with Luckey’s statement that the pictured glasses were thanks, in part, to “deeply invested partners [like Meta].”

Meta is rolling out an update to its line of smart glasses that will disable the camera if the device’s capture LED has been tampered with or destroyed.

Since the rollout of Meta’s second-gen Ray-Ban Glasses in 2024, physically covering the white capture LED on any of the company’s current smart glasses shuts off the onboard camera, something the company did to curb casual misuse.

Before Meta’s mandatory v26 update—which is rolling out now to all Meta Ray-Ban, Meta Oakley and its new $300 Meta Glasses—some users got around the software check by simply drilling out the capture LED hardware itself. Granted, that was already against Meta’s terms of service, which state:

“You may not tamper with the Glasses or otherwise obscure or modify any of the features on the Glasses that signal to others that the Glasses are recording.”

Still, that wasn’t much of a deterrent, since the hardware check failed to notice when the capture LED was physically disabled, and not just covered up by a piece of tape.

Speaking to The Verge, Meta VP of Wearables Alex Himel says the privacy-focused update was meant to follow the release of the company’s cheaper Meta Glasses, which notably lack Ray-Ban or Oakley styling. At the time, Himel told The Verge Meta was aware of increasing misuse amid growing adoption.

Meta doesn’t seem to be outright bricking glasses with drilled-out capture LEDs, however not having the camera severely limits what people can do with them, since they don’t include any sort of display, which ought to be deterrent enough for now.

Lacking a display has essentially forced Meta to invest in use cases beyond the comparatively straight forward expectation of taking photos and video, as the company has released updates that bring camera features such as continuous Live AI capture, which lets Meta AI see what you’re seeing so it can help out with tasks or otherwise identify things directly in your filed-of-view.

And while Meta appears to be distancing itself from one of the least savory consumer segments by putting an end to surreptitious public recording, it’s also seemingly frontrunning increased legislation in a growing number of US states and cities, such as New York state and Philadelphia. Public court houses there have recently banned smart glasses of any type, even those with prescription lenses.

Scrutiny around privacy seems to be coming from all sides though, as a recent Wired report maintains Meta has essentially baked in facial recognition to its smart glasses, which is notably still unreleased at this time.

In March, it was revealed Meta was facing a class action lawsuit in the US over privacy concerns tied to its smart glasses, as the company is accused of sending private camera footage to a Kenya-based subcontractor for manual review to train its AI models.

HTC has been attempting a second act as an XR and AI hardware company for a few years now with mixed results. And while the once-dominant smartphone company is still the midst of a protracted revenue decline, HTC seems to have stabilized.

As reported by Taiwan’s Financial Express News, HTC revealed at its annual shareholders meeting a consolidated revenue of NT$292 million (~$10 million) for June 2026, showing a year-over-year decrease of 8.5% and first-half revenue decrease of 9.7% year-over-year.

Although that doesn’t sound particularly encouraging, the decline actually represents a rare degree of stability for the Taiwan-based company, which has battled significant revenue losses since as far back as 2013.

Granted, HTC is now a fraction of its previous size, which was before selling a significant portion of its smartphone engineering team and IP to Google in 2018, and a number of XR hardware talent to Google in 2025.

Still, HTC appears to be settling in at a much lower revenue base, as the company is ostensibly pinning its hopes on AI, smart glasses, its VIVERSE metaverse platform, and its usual smattering of enterprise hardware and services.

Notably, HTC Chairman Cher Wang stated at the company’s recent shareholders’ meeting that AI has become one of the most important trends it’s following, with the company’s first AI-powered smart glasses, the VIVE Eagle, expected to arrive in the US and Europe sometime in Q3 2026.

Wang also highlighted the company’s metaverse platform, VIVERSE, which has transformed into a user-generative platform since last year, having attracted over 1.7 million month active users in May, with more than 32,000 pieces of content and at least 14,000 creators joining.

However, HTC’s main challenge remains converting these user numbers and strategic initiatives into meaningful revenue growth, something the company hasn’t clearly demonstrated since making the pivot from smartphones to XR.

Outside of XR, HTC telecom subsidiary G REIGNS also made progress in industrial applications, as its partners Taiyang Technology and Alaska-based Microcom are working on an end-to-end open radio access network (RAN) solution, which is currently aiming at the US market and remote areas. G REIGNS is also introducing AI, 5G private networks, and edge computing technologies to Taiwan’s deep-sea fishing industry, which launched a smart fish-finding solution to help fisheries improve fish search efficiency and real-time decision-making capabilities.

One area not discussed was HTC’s XR headset strategy, which seems to have stalled somewhat following the 2024 release of its VIVE Focus Vision standalone in September 2024. As the successor to VIVE Focus 3, the $1,150 Focus Vision was primarily aimed at enterprise XR and enthusiasts.

That said, HTC has managed to survive the collapse of its smartphone business while staying afloat in XR amid Meta’s domination of the segment, which is no small feat. Still, its next big test will be to see whether it can efficiently build off these revenue streams in the coming years, and maybe, just maybe reclaim some of its former glory.

Dubai-based smart contact lenses startup XPANCEO announced it’s expanding a partnership with China’s JBD to co-develop a new micro-display designed specifically for smart contact lenses.

Previously, the companies co-developed a printed circuit board with integrated micro-display, as well as a unique optical system capable of forming a near-eye image for easy focus.

Now XPANCEO is deepening ties with JBD, the Shanghai-based micro display company, to co-develop the tiny displays that could one day be in consumer smart contacts of the future.

Notably, JBD is behind some positively miniscule XR displays; one such display we saw back at CES 2020 was smaller than a penny, but capable of a blinding 3,000,000 nits.

The companies say the partnership will focus on scalability and manufacturability, which could lead to their first mass-market production run of specialized contact lens microdisplays. That said, the pieces are still coming together.

Packing everything into a contact lens form factor is a major feat, with XPANCEO noting it should be around the thickness of a human hair to ensure optimal comfort and wearability.

Another critical challenge is brightness management, since light levels must be carefully balanced to ensure ocular safety while remaining strong enough to see clearly and comfortably. Since images are projected directly onto the user’s retina, prospective smart lenses can make use of lower brightness than smart of AR glasses, the companies say.

In May, XPANCEO revealed it struck a similar partnership with France-based solid-state battery startup ITEN, which aims to solve another big challenge in ocular wearables; conventional batteries are thick, not durable enough, and aren’t suitable to be used in in devices worn directly on the human eye.

And while XPANCEO has its work cut out for it, the company seems to be in the best possible position, as the company secured its Series A financing round in July 2025 to the tune of $250 million, garnering it a $1.35 billion valuation.

Even Realities, the China-based smart glasses company, has officially secured unicorn status with its most recent Pre-B financing round, bringing the company $150 million and a $1 billion valuation.

The round, which was led by Meituan and Tencent, is said to be primarily directed toward the development of Even Realities’ next-gen smart glasses platform, the deepening of AI integration, and the scaling of global operations.

The latest round brings the company’s overall funding to $159 million; its initial $9 million Series A round included investment by CDH Investments, Cyanhill Capital, and China Growth Capital among others.

Unlike Meta’s slate of camera-clad smart glasses, Even Realities builds frames that incorporate binocular, monochrome green displays, critically omitting any sort of camera. And its flagship smart glasses, Even G2, is one of the leaders in the quickly growing ‘display glasses’ segment, incorporating user input through its health-tracking Even R1 smart ring.

“Every generation of computing has brought information closer to people, from the desktop to the laptop, from the laptop to the smartphone,” said Will Wang, founder and CEO of Even Realities. “We believe smart glasses are the next step in that progression. The future isn’t about pulling out a device every time you need information. It’s about having the right information available exactly when you need it, while remaining fully present in the world around you. This investment has shown us that our investors share the same belief that good technology should extend life, not interrupt it. We will continue building toward that future.”

The financing comes as smart glasses enter a new phase of growth, with companies such as Google, Meta, Apple, and Samsung investing heavily in the consumer segment.

Even Realties says its new funding will support continued investment in hardware and software development, deepen AI integrations, and expand global operations to meet growing demand. The company says it also plans to accelerate product innovation as it continues building “a new generation of smart glasses designed to make information more accessible in everyday life.”

Elon Musk’s brain-computer interface (BCI) startup Neuralink announced the company has successfully performed its first transdural brain implant surgery, which promises to increase scalability while significantly reducing surgical trauma.

The News

In a new video (seen below), Neuralink reveals it has implanted electrode threads through the dura—the brain’s thick, leather-like membrane that sits just below the skull—without cutting or removing it like in the company’s previous human clinical trials.

The procedure was carried out in May 2026 during a clinical trial at University Health Network’s Toronto Western Hospital, which is intended to make implantation safer, less invasive, and faster, the company says.

To do this, Neuralink redesigned its insertion needle to better penetrate the dura, which can be over 10 times thicker than the company’s electrode threads, which are thinner than a human hair. The key issue though is the brain beneath is constantly moving, and the dura obscures a dangerous network of blood vessels that need to be avoided.

Neuralink also developed synthetic dura models for extensive testing, allowing researchers to test new imaging systems to better view the brain below the dura.

As seen in the video, Neuralink’s new technique uses indocyanine green (ICG) video angiography, an imaging technique that uses a fluorescent dye to visualize blood flow in real-time, and optical coherence tomography (OCT), which can precisely measure the distance to the brain’s surface as it throbs and moves beneath the dura.

Neuralink says eliminating the step of removing the dura simplifies the procedure and is an important step toward more automated, scalable brain implant surgeries.

“We like to say ‘the best step is no step’ and deleting the durectomy takes one of the most delicate manual steps out of the procedure,” the company says. “This potentially means a safer, more repeatable surgery, and a real path to scaling Neuralink to the many people who could benefit.”

The result: the participant was controlling a cursor with their thoughts within an hour of surgery, and recovery is progressing as expected, the company says.

My Take

Neuralink’s latest breakthrough comes in stark contrast to Meta’s recent Brain2Qwerty v2 research. In short, Meta thinks implants are difficult to scale, which is why they’re investigating the limits of what non-invasive imaging methods can achieve when combined with AI-assisted signal decoding.

For now, both Meta and Neuralink are currently hoping to help people with neurological injuries, which impair speech or movement to some extent, although I can’t help but wonder at how both companies are really talking about scaling. At this stage, they’re most definitely talking about scaling to other similarly affected people, although the real target seems fairly implicit : BCI will eventually be a consumer product, and companies need to do the heavy research now to reserve a spot at the table.

Granted, it’s still very early days. Invasive methods, like those used by Neuralink, provide a much higher signal-to-noise ratio when recording and decoding action potentials, although they still require a surgeon to saw out a quart-sized hole in your skull and put needles in your brain. For consumers at least, it’s the sort of elective surgery that will need to approach LASIK levels of safety and precision before it comes anywhere near a ‘neurotypical’ brain. Actually, probably even more than LASIK, as it would need to be sufficiently reversible and potentially upgradable too.

On the flipside, non-invasive methods like Brain2Qwert v2 make use of magnetoencephalography (MEG) imaging equipment strapped to your head, and rely more heavily on AI to decode signals from the respectively higher noise, since it has to infer action potentials from outside the skull. Meta says it can obtain better results than the more common electroencephalography (EEG) methods of imaging, although current MEG devices need to be used in a magnetically-shielded room to work.

I will continue to be impressed with the technical progress on both ends of the BCI spectrum, although I think it’s safe to say most people won’t have to worry about renting a piece of their brains to trillion-dollar market cap company for a little while longer.

Meta has announced the next version of its AI system for non-invasively decoding brain activity into text. Called Brain2Qwerty v2, the company hopes its latest method will help people with neurological injuries or diseases that impair speech.

Meta’s latest brain-computer interface (BCI) builds off last year’s Brain2Qwerty v1, which initially showed that non-invasive brain recordings could be decoded into text with surprisingly high character-level accuracy. It used both electroencephalography (EEG) and magnetoencephalography (MEG)—two non-invasive methods that measure the magnetic and electric fields elicited by neuronal activity—although it was only capable of decoding individual characters.

Now, the company has shown off its v2 model, which is said to improve nearly every aspect of the system by using an end-to-end architecture, large language models (LLMs), real-time decoding, and vastly improved pattern recognition.

Note: The findings was presented in a recent paper, which involved researchers from Meta and a host of universities and institutes, including Université PSL (incl. École Normale Supérieure), University of Lille, Paris Cité University, Université Paris-Saclay, CNRS, Inria, CEA (NeuroSpin), Basque Center on Cognition, Brain and Language (BCBL), and Hospital Foundation Adolphe de Rothschild.

According to the paper, Brain2Qwerty v2 was trained on approximately 22,000 sentences from nine volunteer participants, each of which were recorded for 10 hours wearing an MEG device while actively typing.

Meta says that instead of relying on hand-crafted pipelines to detect neural events, they used end-to-end deep learning to decode directly from raw brain signals—essentially meaning they could not only decode single letters like in v1, but also full words and sentences.

While v2 represents a pretty significant leap forward, it hasn’t approached 100 percent accuracy yet:

“Brain2Qwerty v2 recovers sentences coherently from noisy neural inputs, achieving a word accuracy rate of 61%, significantly improving upon the 8% word accuracy from other non-invasive methods,” Meta says.

Meta says its most performant participant achieved a 78% word accuracy, where “more than half of all sentences are decoded with one word error or less,” the company says.

“We also find that decoding accuracy improves log-linearly with data volume, suggesting that the remaining performance gap with surgical approaches could be further narrowed through data scaling alone.”

The project’s long-term goal is to develop communication technologies for people with neurological injuries or diseases that impair speech, notably without requiring invasive surgery, like Elon Musk’s Neuralink BCI startup, which expanded human clinical trials earlier this year.

Researchers highlight in Nature that while invasive methods are more efficient at thought-to-text, they expose patients to “nonnegligible risks of brain hemorrhage and infection.” Very real challenges in maintaining cortical implant function over extended time periods is also a risk, making invasive methods less scalable overall.

That said, there’s still a long way to go before we see anything approaching consumer-grade MEG devices. Many of the classical MEG devices of today are still very much “helmet in a hospital room” levels of massive, although there are smaller devices now that can operate at room temperatures, like Cerca’s optically-pumped magnetometers (OPMs).

The key limiter holding MEG back though for eventual consumer adoption is background magnetic interference, which requires even these much smaller systems to work in a magnetically-shielded environment; the magnetic fields generated by the brain are much weaker than the Earth’s magnetic field and the host of everyday tech like smartphones, Wi-Fi, and power lines which are all millions of times stronger.

Whatever the case, it’s heartening to see that patients who can’t qualify for invasive BCI could get a significant boost in quality of life someday, hopefully soon.