A (0:04)

Welcome to the Tech Brew Ride home for Monday, August 18th, 2025. I'm Brian McCullough. Today, Meta is readying smart glasses with a display for as soon as next month. Is all the money in the world not actually going to get Zuck what he wants in AI? And is the aqui hair and all but name trend of recent months breaking Silicon Valley's fundamental business model? Here's what you missed today in the world of tech. Over 10 years ago, Mizzen and Mane invented and some might say perfected the performance fabric dress shirt. To this day, they continue to embrace that same entrepreneurial spirit by re engineering classic American styles with modern fabrics. The goal is to make it easier for guys to achieve and enjoy their version of success. So whether you're grinding away in an office in San Francisco or on site in Austin, you can, you can rest easy knowing you don't have to sweat about sweating. They're the brand you can turn to if you're looking for a machine washable and wrinkle resistant white dress shirt you can wear every day of the week, high stretch pants that are just as comfortable in the office as they are on the golf course, or a lightweight and breathable wool blazer that makes dressing up feel like a breeze. Go to mizzenandmain.com and use promo code brew15 to get 15% off your first purchase. That's mizzenandmane.com promo code brew15. We've been talking for a while now about how I feel like smart headwear will be the next big hardware story in tech for the next decade or so. Which, you know, at this point it's maybe Captain Obvious, Brian, but also I've been talking about how Meta is seemingly leading this new horse race. Well, a bunch of news on that front today. First off, Mark Gurman says that Meta may debut what they are calling Hypernova smart glasses, but finally with a display for around $800 later this year. As I've reported, Meta is readying its first smart glasses with a display dubbed Hypernova internally, a precursor to full blown augmented reality glasses. The device will be unveiled next month and I've already detailed how the technology works. There will be a small screen for many apps and alerts on the right lens and the spectacles can be controlled via a so called neural wrist accessory, the same one used with the Orion AR prototyp. During development of the product, Meta expected to charge at least $1,000 for Hypernova, with some people thinking the device could be as much as $1,400. That's far higher than the $200 to $400 meta ray ban glasses without displays, or even the new up to $500 Oakley smart glasses. In fact, it would have put the glasses on par with a high end iPhone. Well, here's some good news. Meta recently figured out a way to slash the price for consumers down to about $800. I'm told the move stems in part from the company accepting lower margins to boost demand, a common tactic for new products. One caveat the roughly $800 price point will be the starting price, meaning style variations and prescription lenses will quickly push up the cost. End quote. But wait, there's more. Smart glasses are on the low end in terms of tech. What about the high end? The Bleeding Edge over the weekend, Upload VR got hands on with Tira Ma Meta's Hyper Realistic VR Research prototype with beyond retinal resolution, high brightness and contrast, and a narrow field of view. Since Oculus first debuted and showcased the DK1 over a decade ago, the angular resolution of consumer headsets in VR has steadily advanced. Early devices offered just 6 pixels per degree ppd, a level comparable to legal blindness. While mainstream headsets today reach 25 ppd, high end devices achieve around 35 ppd, and Varjo's XR4 pushes 51 ppd in the center of vision. Yet the question remains how sharp must displays become before virtual worlds appear indistinguishable from reality? And are brightness, contrast and other specifications equally or more important than resolution? These questions lie at the heart of Meta's research, particularly through its Optics, Photonics and Light Systems Opals team. Their experimental headset, codenamed Tiramisu, represents a milestone in Meta's pursuit of the visual Turing test, a threshold where users cannot tell whether they are perceiving reality or simulation. Industry convention holds that 60 ppd approximates the limit of human visual acuity, but Meta's researchers have long doubted this figure. Earlier prototypes such as butterscotch at 56 ppd were used to probe whether so called retinal resolution was enough, but results tended to suggest otherwise. Now Tiramisu extends this exploration, achieving an unprecedented 90 ppd with micro OLED displays, a peak brightness of 1400 nits and triple the contrast of the Quest 3. Instead of pancake lenses which sacrifice brightness, Tiramisu employs a custom 3 element refractive lens stack, delivering far more light to the eye but at the cost of bulk and weight. The field of view is narrow, just 33 by 33 degrees, but within that window the visual fidelity is extraordinary. Apparently unreal Engine 5 demonstrations revealed crisp details, lifelike emissive lighting, and a realism beyond previous headsets. What did it actually look like? To, you know, look at it well, further, the increased brightness and contrast even seem to enhance the sense of depth, despite Tiramisu not being varifocal. And beyond that feeling of realism, Tiramisu's bright and vibrant image was downright pleasant to look at. As the head of DSR Douglas Landman, said, It was reminiscent of seeing a 4K HDR OLED TV for the very first time. Except even more striking, it's the most realistic image I've ever seen. And I don't just mean in VR, I mean from a display of any kind. Apparently. Though this is still impractical as a consumer product at this point, though Tiramisu does validate Meta skepticism apparently about that 60 ppd ceiling and might highlight how brightness and contrast are equal, equally vital to getting to that visual Turing test barrier. Its apparent successor still in development, Tiramisu 2 aims to balance resolution, field of view, compactness and comfort, bringing the vision of true to life XR displays closer to reality. But wait, there's more. This same author also got hands on with Boba 3 Meta's prototype PC VR headset that has a form factor. Crucially that is similar to the quest 3. Because the tiramisu and tiramisu 2 are so heavy, probably never going to be consumer devices, but with the Boba 3 featuring an ultra wide field of view of 180 degrees by 120 degrees, it is clearly something that they imagine releasing as a consumer product because it's smaller and lighter. And this is a winning endorsement again from the same author. Quote after trying boba3 and seeing a massive field of view in a viable form factor with my own eyes, I'm now more optimistic than ever about the future of VR. End quote. First disclosed in a paper abstract three weeks ago, Boba 2 was built last year after DSR optical scientist Yang Zhao saw that recent Meta driven advances in pancake lenses could be adapted to deliver an ultra wide field of view in a practical headset. Off the shelf VR LCDs had also reached the density needed to maintain Quest 3 class angular resolution at those widths. Using 3K panels, Boba 2 attains 25 pixels per degree, again PPD matching the Quest 3. This year's Boba 3 upgrades to 4K LCDs for 30 PPD and adds a refined lens production process that reduces edge artifacts seen on Boba 2. Both versions provide a sweeping 180 degree horizontal by 120 degree vertical field of view. For contrast, Quest 3 measures under 110 x 96 degrees, Quest 3s under 96 x 96 degrees and human vision is roughly 200 x 135 degrees. Meta estimates that Boba 3 covers 90% of the human field versus 46% for the Quest 3 in demos, Boba 3 apparently presents Quest 3 like clarity across its massive field with only mild edge distortion. Comparable to the Quest 3, a cameraed variant delivers full field of view pass through Mr. Via dual 20 megapixel sensors and a custom FPGA pipeline over fiber PCIe necessary because USB bandwidth falls short. However, the pass through is not depth correct and uses a static lens match projection producing geometric distortion on head motion. Researchers say dynamic reprojection could address this with further work, but according to this report, unlike many research headsets that trade one breakthrough for severe compromises, Boba 3 feels unusually complete. It's not a product yet, but its design relies on conventional components, dual element pancake lenses with a high curvature reflective polarizer and standard 4K LCD similar to those in Varjo XR4 and Pimax Crystal Super VR headsets. Trade offs do remain a smaller IBOX that dims rather than blurs when misaligned, adding weight from larger optics and an aluminum reinforced visor and external constellation tracking in at least this demo, the larger barrier to shipping is apparently compute Ultra wide field of view dramatically increases pixels and visible scene complexity per frame. Eye tracked, foveated rendering and neural upscaling can help, but only to a point. Again. Again, these are just research prototypes, but that last one technically could be manufacturable someday soon if the compute problems can be solved. So give it time I guess from the checking in on the health of the startup ecosystem file According to PitchBook, 15.9% of VC backed deals thus far this year have been down rounds. Now what's slightly more remarkable about this is that would mark a 10 year high for down rounds, and AI and ML startups accounted for 29.3% of the down rounds. Quoting Fortune, the soaring valuations of the early 2020s are finally coming back to earth. This week PitchBook data revealed that 15.9% of venture backed deals in 2025 so far have been down rounds, making a decade high. Additionally, almost every major IPO listing in Q2 hit the public markets its peak valuation the data from PitchBook Ads. Some examples include MNTN at IPO, its valuation was down from 2 billion to $1.1 billion. Circle, which dropped from 7.7 to $5.8 billion. Hinge value at IPO was $6.2 billion, down from the $23 billion high and Chime going public at $9.1 billion, down from a $25 billion peak valuation. AI does continue to be a bright spot in many ways, but isn't entirely exempt either, as 29.3% of down rounds were in PitchBook's broad AI and machine learning vertical. Of course, the biggest names in AI like OpenAI reportedly heading toward a $500 billion valuation, and Anthropic, reportedly raising at a $170 billion valuation, continue to hit eye popping levels and lower on the food chain. AI is still consistently valued at a premium, with PitchBook reporting that median Series B step up for AI startups is 2.1%. Well, AB median of 1.4% that all other categories fetch. The IPO market is some would say back. I think it pretty much is, but also has been for a while. For those with the stomach for it in Q2, venture backed startups in the US generated $67 billion in exit value, PitchBook said, the highest since the last quarter of 2021. But here's a sobering fact. There are still lots of unicorns out there, so distributions back to VC firms and by extension their LPs are limited. An unsurprising but ice cold and very down to earth number to leave. The unicorns that have made their public debuts this year comprise a mere 1% of all US unicorns out there. End quote. We've all been there. Too many SaaS tools, not enough visibility at all, and way too much access for you to keep track of. It's the stuff security nightmares are made of. That's where Trelica by1Password comes in. They inventory every app your company uses and create app profiles to help you easily assess risks, manage access, and make sure your password security is locked down tight. With 1Password's extended access management, you can control your company's many, many SaaS tools, securely onboard and offboard your people and actually hit your compliance goals. I've been telling you about 1Password Extended Access Management all year and now Trelica comes along to make things even better. Sleep Easy with Trelica by 1Password learn more @1Password.com Ride. That's 1Password.com Ride. You think that Mark Zuckerberg throwing all that money around to hire AI talent has him thinking, okay, we're good now? Well, not so fast because the Information is reporting that Meta plans its fourth AI restructuring in just the past six months. Quote Meta's new AI organization Meta superintelligence Labs is expected to be divided into four groups a new lab that has been going by TBD Lab, short for To Be Determined, a team focused on products that include the Meta AI Assistant, a team focused on infrastructure, and the company's fundamental AI Research Lab, which works on longer term research, two of the people said. TBD Lab, which is working on the newest version of Meta's flagship large language model Llama, is expected to have several leads, the two people said. Jack Ray, formerly of Google, is expected to lead pre training, according to two of the people familiar with the matter. Roman Pang, who previously led model development at Apple, is expected to lead infrastructure for TBD Lab, the two people said. Among the leaders of the post training is expected to be Hong Yu ren, formerly of OpenAI and Pisun, formerly of Google, according to one of the people familiar with the matter. Nat Friedman is expected to continue overseeing products for Meta superintelligence Labs, the person said. Robert Fergus, who co founded fair, is expected to continue leading that lab. Meta most recently restructured its efforts in AI in June, shortly after it hired Wang and Friedman. In May, the company cleaved its Generative AI Group, previously led by Al Dahl, into teams overseeing AI research and another working on AI products. In February, Meta moved two engineering leaders from the group and installed Laura Dana Krisan, then head of messenger, to lead product in an early sign of the shakeup to come. End quote. So why is this chaos still happening over there? Well, Forbes has sources saying that Meta's chaotic culture and lack of overall vision have led to what it terms an AI brain drainage. They already had the best people and lost them to OpenAI. This is Mark trying to undo the loss of talent, one ex Meta AI employee said. And even as Zuckerberg makes jaw dropping offers for top tier AI researchers, the social media giant continues to lose those that are left. Today, when it comes to recruiting high caliber AI researchers, Meta is often an afterthought. Insiders at some of Silicon Valley's biggest AI companies said that prior to the fresh hiring of the last few months, Meta's talent largely didn't meet their hiring bar. We might be interested in hiring some of the new people Mark is hiring now, but it's been a while since we were particularly interested in the people who were already there, a senior executive at one of the major frontier AI companies told Forbes. Google has hired less than two dozen AI employees from Meta since last fall, according to a person familiar with Google's hiring compared to the hundreds of AI researchers and engineers it hired in that timeframe. Overall, that person told Forbes, the prevailing belief is that Meta didn't have much talent left to poach from. Google declined to comment. This has lent an air of desperation to Zuckerberg's attempts to raid the likes of OpenAI and Thinking Machine Labs, the fledgling startup helmed by former OpenAI CTO Mira Muradi with nine figure offers and promises of near unlimited compute. In at least two cases, the Meta CEO has offered pay packages worth over $1 billion spread across multiple years, according to the Wall Street Journal. He reportedly poached at least 18 OpenAI researchers, but many have also turned him down, betting on bigger impact and better returns on their Meta is the Washington commanders of tech companies, one AI founder told Forbes, referring to the NFL team and its pursuit of free agents. They massively overpay for okayish AI scientists, and then civilians think those are the best AI scientists in the world because they are paid so much. Anthropic CEO Dario Amodai has said he's spoken to Anthropic employees who have gotten offers from Meta who didn't take them, adding that his company wouldn't renegotiate employee salaries based on those offers. If Mark Zuckerberg throws a dart at a dartboard and hits your name, that doesn't mean you should be paid 10 times more than the guy next to you who's just as skilled, just as talented, he said last month on the big technology podcast. Anthropic declined to comment. Anthropic has an 80% retention rate, the strongest among the frontier labs, according to a May report by VC firm Signalfire. The findings are based on data collected for all full time roles, including engineering, sales and HR, and not AI researchers specifically. In comparison, DeepMind has a 78% retention rate, OpenAI 67%, and Meta Trails with with 64%, end quote. Finally today from the Journal, Asa Fitch makes the argument that Big Tech's reverse aqua hires for AI talent that have been all the rage are hollowing out startups and eroding the culture that has made Silicon Valley an unparalleled source of innovation. The problem is that the moves challenge Silicon Valley's cultural foundations. Silicon Valley's basic bargain has always been rooted in taking enormous risk in the hope for an equally enormous reward. Most startups fail, but those that succeed can be wildly successful, generating hundredfold or higher returns for their venture capital backers and making employees many lured with the promise of equity rich. It is an especially risky business for rank and file employees of venture backed startups who are tied to the success of one company at a time, not a diversified portfolio of them. But for many employees of startups hollowed out in reverse aqua hires or passed over in big tech hiring sprees, the rewards haven't been very handsome. A bunch of tech employees losing out on a payday might not seem all that significant, but Silicon Valley's innovation machine only works if it has an army of people who aren't founders or leading researchers moving its gears. The employees who handle sales, marketing, human resources, or are part of large engineering teams. That's who's getting the short end of the stick. There are a ton of employees who are bought into this system and the history and the tradition is you come out here, you try to make something of value and if it works out, everybody wins, said John Sakoda, the founding partner of Decibel, a venture capital firm. If you thought you had a share of a company and you actually didn't have a share of a company, there's a loss of trust. If the reverse aquihire trend persists, there is a reasonable chance. Many people who would have been bold enough to join a risky startup will give more more weight to their other options. They might instead go directly to big tech companies in what might be a safer route for themselves, but one that makes the pool of available startup talent shallower. Long time listeners will remember many years ago now when I finally dumped the Chrome web browser for the Brave browser as my daily driver browser. Mostly to get away from Google if you'll recall. Well, what if I get away from Google entirely? I'm thinking about it. The last link in the show notes is to an Ars Technica article I read this weekend that is making me consider making the move. All in has anyone out there used the Kaji search engine K A G I and by that I mean actually paid for a kaji subscription. I'm considering, as I said, going all in paying for a subscription but also crucially using their Orion browser too. So hivemind before I pay up to get totally ad free tracking free actually not shitty search. I need to know I can trust using Orion as my daily web browser. Anyone out there used it for like more than a month or so. Any and all feedback are welcome. Before I make the leap, hit me up on the socials or email me@brianidehome.in fox. Thanks in advance.