Tech Brew Ride Home - (Omnibus) Week Of 03/03/2025

Episode Date: March 8, 2025

(Omnibus) Week Of 03/03/2025 Subscribe to get this every week AD-FREE at tech.supercast.tech Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. everybody, as I mentioned yesterday. This is another taste of the omnibus episodes that we do on the paid feed, where you can just listen to everything that happened this week in one single episode as you're walking your dog, going to the gym or whatever, in about an hour and 15 minutes.

If you subscribe at tech.comcast.com, you can get this without ads. Every week, there will be ads in what you're about to listen to. And also, you get ad-free versions of every daily episode. So if this is interesting to you, subscribe at tech. com.com. This is the Omnibus episode for the TechMeme Right Home podcast for the week of March 3rd, 2025. Well, might we get that U.S. Government National Crypto Reserve after all? President Trump has named Bitcoin, Ethereum, XRP, Salana, and Cardano as assets to be included in a strategic U.S. Crypto Reserve causing significant price surges across crypto, quoting CoinDesk. Notably, Trump did not initially mention Bitcoin or Ethereum, the two largest cryptocurrencies

by market capitalization in his statement, but he later clarified that the reserve would include these assets as well. Trump made the announcements on Truth Social, his social media platform. A U.S. Crypto Reserve will elevate this critical industry after years of corrupt attacks by the Biden administration, which is why my executive order on digital assets directed the presidential working group to move forward on a crypto strategic reserve that includes XRP, SOL, and ADA, Trump stated. I will make sure the U.S. is the crypto capital of the world. We are making America great again. Following Trump's initial announcement, XRP, Seoul, and ADA experienced significant price surges. The price of ADA stored by more than 63%

within two hours after the president's post, while Solana increased by 23% and XRP by 32%. While fans of XRP ADA and Solana celebrated the news, some Bitcoin and ETH fans initially responded with a mixture of disappointment and surprise. Around an hour after his initial post, the president clarified that BTC and ETH as other available cryptocurrencies will be at the heart of the reserve, end quote. And quoting CNBC, Trump is hosting the first White House crypto summit on Friday, and investors will be watching closely for more clues about the direction of the reserve plans. This is the first time Trump has specified his support for a crypto reserve versus a stockpile. While the former assumes actively buying crypto in regular installments, a stockpile would simply not sell any of the crypto currently held by the U.S. government. Much of the industry had a lukewarm response to the language, in addition to the order

specifying a stockpile over a reserve, use of the term digital assets suggested it could include other cryptocurrencies. However, many in the crypto community feel strongly that a crypto reserve should only hold Bitcoin, as it's the most battle-tested and decentralized of the crypto networks, the inclusion of other coins could also invite the government to pick winners and losers in the crypto market, they say, end quote. More presidential announcements, this is being announced as we speak, quoting the journal. Taiwan semiconductor manufacturing intends to invest $100 billion in chip manufacturing plants in the

U.S. over the next four years under a plan expected to be announced later Monday by President Trump, according to people familiar with the matter. The investment would be used to build out cutting-edge chip-making facilities. Such an expansion would advance a long-pursued U.S. goal to regrow the domestic semiconductor industry after manufacturing fled largely to Asian countries in recent decades. TSM, the world's largest contract chip manufacturer set down routes in Arizona in 2020 when it said it would build a chip factory there for $12 billion. Its ambitions for the site have expanded rapidly since, with two more factories on the same site and a total investment of $65 billion. The company's first factory began mass production late last year. TSM's announcement comes after years of deliberation regarding the future of semiconductor manufacturing in the U.S. and the global tech sector. The company currently builds its most advanced chip-making facilities only on its home soil, Taiwan. The chips it produces are critical for powering everything from the latest artificial intelligence systems to smartphones. Since the Biden administration, the U.S. has expressed concerns about TSM's near monopoly on advanced chip manufacturing

and has been urging the company to relocate more of its cutting-edge production, including advanced chip packaging facilities to the U.S. Advanced chip packaging facilities to the U.S. Chip packaging is particularly critical for AI-related chips as it enhances performance by integrating multiple semiconductor components, reducing size, improving power efficiency, and ensuring faster data transfer. Key factors for AI applications. The U.S. has supported TSM's growth through 2022's Chips Act, which earmarked tens of billions of dollars in grants to domestic chip manufacturing. TSM was awarded up to $6.6 billion of grants and has recently begun to receive federal money from the program. Generous tax credits have also boosted chip-making projects in the U.S. pleased to have an opportunity to meet with the president and look forward to discussing our shared

vision for innovation and growth in the semiconductor industry, as well as exploring ways to bolster the technology sector along with our customers. TSM said in a statement, end quote. Google has demoed some new Gemini features, including screen share to let users ask questions based on their phone screen and video search coming later this month. Quoting TechCrunch at the Mobile World Congress, MWC 2025 in Barcelona, the company showed off a new screen share feature which lets users share what's on their phone screen with Gemini and ask questions about the pictures. As an example, the company showed a video of a user shopping for a pair of baggy jeans and asking Gemini what other clothing would pair well with it.

As for the video search feature, Google had teased it at Google I.O. last year, the feature lets you take a video and ask Gemini questions about it as you're filming, end quote, and quoting Android Central. Project Astro was first shown off at Google I.O. last year where a woman is seen interacting with her surroundings through her phone, asking the assistant about various objects placed on a table, or even creating iterations about a golden retriever. It was fascinating to see how Gemini Live has pushed the boundaries of the mere Google search beyond just looking for something online. Now anyone could simply use Circle to Search on their devices to look up something they spotted

within a different app. With Project Astro, Google wants to make finding things easier, not just on your phone, but out in the world as well. There have been some rumors recently on how Gemini will be replacing Google Assistant in the future. A deep dive into the latest beta version of the Google app revealed that, hey, Google, could be replaced with Hey Gemini. Now with Google showing off Gemini's live video and screen sharing, this change could be coming sooner than we anticipated. The company stated it will start rolling out the above multimodal features to Gemini Advanced Subscribers as part of the Google One AI premium plan on Android devices. While the company didn't specify exactly when we can see it on our

devices, it did say later this month, so we're hoping it could arrive in the next monthly update, end quote. Lots of bosses are like, hey, you need to come back to the office. You need to work extra hours in order for this company or this project to succeed. But Sergey Brin took it a step forward, telling workers that Google could reach AGI if employees worked harder and were in the office more, saying 60 hours a week is the sweet spot of productivity. Quoting the Times, I recommend being in the office at least every weekday. He wrote in a memo posted internally on Wednesday evening that was viewed by the New York Times. The note highlighted Mr. Brin's belief that AGI, a long-sought goal in computing, could be within reach.

And it shed more light on how he believes Google could achieve that technological leap. Competition has accelerated immensely and the final race to AGI is afoot, he wrote. I think we have all the ingredients to win this race, but we are going to have to turbocharge our efforts. He highlighted the need for Google's employees to use more of its AI for coding, saying the AI's improving itself would lead to AGI. He also called on employees working on Gemini to be the most efficient coders and AI scientists in the world by using our own AI. More companies have ordered employees back to the office full-time to improve productivity. In September, Amazon said, its corporate employees must return to the office five days a week starting in 2025.

AT&T, JPMorgan Chase, and Goldman Sachs have also reversed hybrid work policies, end quote. And quoting the verge. While his note was intended for a small audience of AI researchers at DeepMind and not all of Google, it's still remarkable in a couple of ways. For one, it came from Bryn, who technically has no formal role these days besides being a board member, and not Demis Hasebis, who runs Google DeepMind. As some employees have pointed out to me, there's also irony in Brin recommend spending that employees work 12-hour days when he and co-founder Larry Page arguably left Google

rudderless when they retired in 2019 just before this AI boom cycle began in earnest. Finally, it's telling that Bryn, who attended President Donald Trump's inauguration with CEO Sundar Pichai, is now using his power to seemingly push for removing Gemini's guardrails, end quote. Green shoots, as folks like to say, green shoots in terms of hot startups getting back to their previously hot valuations. New York-based corporate payments startup ramp hit a 13, billion dollar valuation and a $150 million share sale to GIC, Stripes, Thrive, and others up from $7.65 billion in April 2024, quoting the FT. The leap in valuation to $13 billion puts Ramp among the most highly valued U.S. startups outside of a handful of artificial intelligence

companies such as Open AI. It follows rapid growth powered by an increase in spending on card transactions and bill payments. But Eric Gleiman, Ramp's co-founder and chief executive, emphasized that it had benefited from using AI across the company. It is not not possible to use Ramp without using AI, he said, adding that the technology had quickly moved from simple chatpots into being deeply integrated in every part of the business, expenses that do themselves, books that do themselves, money that finds higher yield, end quote. We're living in a world where computers can talk and think and reason and finance is really about reasoning, making sure your capital has more value each month, he said.

Ramp's valuation hit $8.1 billion in 2022, but dropped to $5.8 billion a year later as higher interest rates hit consumer spending, factors that also hit rival FinTech. tech companies such as Stripe and Klarna. Fintech obviously went through volatility given the wild swings in rates and spending across businesses and consumers, said Kareem Zakai, a partner at Thrive Capital who led the firm's investment into ramp. Businesses that took share before the downturn continued to take share, but customer spending was down. Now the market has turned around and they are accelerating, he added. According to a person with knowledge of the company's finances, ramps annualized revenue, a metric often used by fast-growing startups that multiplies the current

month's revenue by 12 is $700 million. That figures up from $300 million in, in August 2023. The company is processing $55 billion in payments on an annualized basis, compared with $10 billion at the beginning of 2023. Ramp's aim is to become a platform offering corporate customers a range of services rather than a single product, according to Glyman, end quote. Finally, today I save the Mark German Apple's scoop Monday to the end, because he has a really interesting look at what he calls Apple's AI crisis, as some in Apple's AI Division believe that a true conversational LLM Siri won't be ready until iOS 20. at best in 2027.

Quoting Mark in Bloomberg. With Apple Intelligence, the company was already trailing chat GPT, Google's Gemini, and Microsoft's co-pilot, but it still had the opportunity to respond with something fun, conversational, well-integrated, and useful. And Apple seemed to be moving in that direction when it unveiled the technology last June, describing it as AI for the rest of us.

But Apple couldn't pull it off. The system doesn't come close to matching the competition, and a miss in this area could have potentially devastating consequences for the company. AI promises to be a pervasive part of all our lives and change the way we interact with devices, communicate, and get stuff done. An AI-created emoji feature isn't going to cut it, not while Apple's rivals are showing off fully conversational and personalized assistants that are eerily similar to human beings. Case in point, Amazon's Alexa Plus, which was announced this past week. It's essentially a version of ChatGPT's voice mode with knowledge of who you are, who the people in your life are, your interests, and the context of your home and surrounding environment. The next version of Siri will be a test of whether Apple can mount a comeback.

The software will likely be released in May a full 11 months after it was introduced. The current iOS 18 version of Siri essentially has two brains, one that operates the legacy Siri commands like timers and making calls, and another that handles more advanced queries. The latter capability will be able to tap user data and already is used to not get confused when people change their request mid-command. In order to get Apple Intelligence out the door as part of iOS 18, the company didn't have time to meld the two systems together. That means the software doesn't work as smoothly as it could. For iOS 19, Apple's plan is to merge both systems together and roll out a new Siri architecture.

I expect this to be introduced as early as Apple's worldwide developers conference in June of this year, with a launch by spring 2026 as part of iOS 19.4. The new system, dubbed LLM Siri internally, was supposed to also introduce a more conversational approach in the same release. But that is now running behind as well and won't be unveiled. in June. Before Apple can go full-throttle on development of that Siri, which is supposed to finally work more like Chat VPT and the new Alexa, Apple will need to get the underlying system fixed, and that won't be easy. That's why people within Apple's AI Division now believe that a true, modernized, conversational version of Siri won't reach consumers until iOS 20 at best in

27. That would mean Apple is half a decade late to the game, an even bleaker timeline than many of us imagined. On top of that, Apple's rivals aren't standing still, given the speed at which they're operating now, imagine where OpenAI and Google will be in two years. That's not to mention the AI startups cropping up every few weeks. Everything I've heard suggests that iOS 19 will not include any significant consumer-facing changes to Apple intelligence. A big reason for that is the amount of time Apple is still spending working to get the features announced last year out the door. It's challenging to move on to next year's release before this year's operating system updates have rolled out to customers. That's left Apple at a make-or-break point. Clearly, the company

isn't moving fast enough internally to create the underlying AI technology it needs to keep up with the competition, and that suggests a change is required. People involved in Apple's AI work say its foundational and large language models, the basis for its homegrown AI features are reaching their limits. There also have been problems with rivals poaching talent and what they deem to be an effective leadership. The company is hoping its models ultimately get more capable, but getting its hands-on enough chips has been another challenge. There's been an industry-wide scramble to get AI accelerators, mostly from Nvidia, to train software. That's probably one reason the company is ramping up production of its own AI servers and has its chips team working on new solutions.

The rest of the industry has scooped up much of the capacity, according to multiple employees. Once Apple realized it needed more resources after being caught off guard by ChatGBT, it was too late to get enough. Moreover, the AI industry is advancing so quickly that the team's work is often obsolete by the time it's ready to unveil employees say. Apple management believes that despite being a laggard in AI, the company isn't in danger of losing its user base. Tight integration of hardware, software and services, its strength in-house chips, and software user interfaces still give Apple an edge, executives argue, while that's all true, AI is so disruptive that it's creating entirely new ecosystems and technology around it.

That puts Apple's model in jeopardy. AI is a once-in-a-generation technology. Apple probably still has time to turn things around, but that window is closing fast, end quote. We officially have it. the first IPO of the AI era, Corweave, which provides cloud-based GPUs, has filed for an IPO on the NASDAQ under the symbol CRWV and says it generated $1.9 billion in revenue in 2024, up 737% year-on-year. So to be clear, this isn't an AI company and the way that Open AI is. They basically help AI companies get the chips they need to make the AI.

Quoting Investors Business Daily. New Jersey-based Corweave rents cloud computing power with Nvidia chips to AI developers. The company has seen a big jump in sales as companies race to train AI models. Corweave's revenue grew more than 700% year-over-year to $1.9 billion in 2024, according to its IPO prospectus filing. The company posted a net loss of $863 million for 2024. The company has not yet listed the number of shares or price it will target in the listing. Bloomberg reported last week that Corweaves IPO filing was imminent, with the company likely to seek a valuation of more than $35 billion.

CoreWe've raised private funding in November at a reported valuation of $23 billion, end quote. Worth pointing out that 77% of their revenue is coming from just two customers, with 62% coming from just one of them, Microsoft, but they apparently have a pipeline of over $15 billion in coming contracts. Quoting CNBC, originally known as Atlantic Crypto, the company got its start in 2017 by offering infrastructure for mining the Ethereum cryptocurrency, After digital currency prices fell, the company bought up additional graphics processing units or GPUs and changed its name to Corweave with an increasing focus on graphics rendering and artificial intelligence.

At the end of 2024, Corweaves's 32 data centers housed over 250,000 Nvidia GPUs with a majority using the previous generation Hopper architecture, according to the filing. Running data centers full of GPUs requires considerable energy. CoreWeave had 360 megawatts in active power and a total of 1.3 gigawatts had been contracted, the filing said. Corweave gained popularity after OpenAI released ChatGPT in late 2022 because the company could quickly provide GPUs to businesses in need. Microsoft, whose Azure Cloud Unit has supplied computing power to OpenAI, started working with Corweave in 2023 to meet OpenAI demand, end quote. Also in AI news, Anthropic raised a $3.5 billion. Series E, led by Lightspeed at a $61.5 billion valuation. According to a source, its annual revenue run rate has grown by 30% in 2025 from around $1 billion in late 2024, quoting Bloomberg.

The latest deal for the maker of large language models best known for its chatbot named Claude was led by Lightspeed Venture Partners, which contributed $1 billion, said Anthropic. Other investors who participated in the Series E deal include General Catalyst, Jane Street, and Fidelity Management and Research Company, according to an Anthropic statement. Existing investors, including Menlo Ventures and Bessemer Venture Partners, also backed the company in the latest round. The company said, Anthropic, founded in 2021 by former employees of Open AI, has positioned itself as a reliable, safety-conscious firm that users can trust. The company was originally planning to raise $2 billion in the latest financing, but ended up raising more than anticipated in an oversubscribed

round. People familiar with the discussions previously told Bloomberg. The new funding will fuel Anthropics competition with Open AI, which is currently in talks for an even larger $300 billion valuation. Both deals underscore Silicon Valley investors continued enthusiasm for funneling big sums to leading AI firms despite the recent entrance of smaller companies like China's Deepseek, which claim to develop similar technology more cheaply. Anthropics says that it will use the capital to develop the next generation of its AI systems, expand its computing capacity, and further its research efforts, including in safety-related areas. The company also said it plans to accelerate its expansion in Asia and Europe. This investment fuels our development of

more intelligent and capable AI systems that expand what humans can achieve while deepening our understanding of how these systems work. Anthropic Chief Financial Officer Krishna Rao said in a statement, these capabilities are driving remarkable outcomes for our customers as our business and consumer usage continues to grow rapidly, end quote. The round follows additional investments from two major tech companies who had previously backed Anthropic. Amazon invested $4 billion in the company in November and Alphabet's Google invested $1 billion in January. Anthropic is also helping power Amazon's new AI version of its Alexa Assistant Alexa Plus, end quote. Apple has released updated 11-inch and 13-inch iPad Airs with an M3 chip and unveiled a new

$269 plus magic keyboard with a larger built-in trackpad, quoting the Verge. The new Air is primarily a spec bump over last year's device. It has an upgraded M3 chip, and of course, supports Apple intelligence. It's not Apple's most modern chip. It's not Apple's most modern chip, that would be the M4, but Apple says it's twice as fast as the M1 powered air and the A14 bionic powered air. Those are older devices and Apple's comparisons. Notably don't include last year's model. It comes in four colors, 11 and 13-inch models, and starts at $599 for the smaller model, and $7.99 for the larger. There's also a new magic keyboard attachment, $269 for the smaller model, and $319 for the larger, which includes the extra row of function keys and larger trackpad. You could,

only get on the pro, end quote. But wait, there's more. We now have an 11th generation iPad, the entry-level device, adding an A-16 bionic chip and now starting at 128 gigabytes of storage for the base model, with options up to 256 gigabytes and 512 gigabytes for the same $349 starting price. Quoting Mac Rumors, it is available in blue, pink, yellow, and silver, and continues to start at $349. customers can pre-order the new iPad starting today with launch scheduled for March 12th, quoting Apple itself. The A-16 chip provides a jump in performance for everyday tasks and experiences in iPad OS while still providing all-day battery life. Compared to the previous generation, the updated iPad with A-16 is nearly 30% faster. In fact, compared to iPad with A-13 Bionic, users will see up to a 50% improvement in overall performance,

and A-16 makes the updated iPad up to six times faster than the best-selling Android tablet, end quote. Uber has launched Waymo rides in Austin, Texas, operating within a 37-square-mile area, including Hyde Park and downtown. Quoting the Verge, the former rivals have joined forces in the hopes of accelerating the transition to autonomous vehicles, and as of today, any Austin resident with a desire to take a trip in a fully driverless robotaxie can open up their Uber app and hail away. But of course, there are some limitations. Waymo only operates within a 37-square-mile area in Austin, which includes Hyde Park downtown, Montopolis, and other popular destinations, according to the company, so the trip will need to originate and end within that service area in order to qualify for Waymo.

It also doesn't operate on highways yet, so the route will only comprise local streets. Another thing to consider, simply calling an Uber in Waymo's geofense doesn't guarantee a roba taxi will show up. Waymo spokesperson Chris Bonelli declined to share the size of the company's fleet, in Austin but acknowledged that it was small at launch. While smaller at start, the fleet will grow to hundreds of vehicles over time, Binelli said, end quote. Now, last they revealed these numbers in August of last year, Waymo had a fleet of about 700 vehicles to service their existing San Francisco, Phoenix, and Los Angeles areas, so maybe we can extrapolate several dozen vehicles for Austin. I don't know.

Quoting and Gadget. Unlike the Waymo One service offered to writers in San Francisco and Los Angeles, interested Austin riders will get a chance to ride in a Waymo just by ordering an UberX, Uber Green, Uber Comfort, or Uber Comfort Electric ride, similar to how things are handled in Phoenix. You're not guaranteed to receive a Waymo when you do order one of those rides, but Uber says you can increase your chances of receiving one by changing your ride preferences in the app. Up front prices remain the same, whether you get a Waymo or not, though if you do end up in a Robo taxi, you won't be prompted to tip. You can use the Uber app to unlock the Waymo, open its trunk and start your ride too, end quote. And quoting TechCrunch. The structure of the

Waymo on Uber service forecasts how each company envisions its role in the nascent autonomous vehicle industry. Under the arrangement, Waymo will be responsible for vehicle testing, roadside assistance, and certain aspects of rider support. Uber will manage the fleet services such as vehicle cleaning, maintenance, inspections, charging, and depot operations through a company called Move Cars, which is rebranded to VOMO. In 2021, Uber acquired a 30% minority interest, in the European company Move Cars, not to be mistaken with Move, which, coincidentally, is managing Waymo's AV Fleet and Phoenix. Uber has partnered with 14 autonomous vehicle companies that cover ride-hailing, delivery, and trucking, a handful of which are operating commercially.

In December, Uber launched Robotaxi rides with We Ride in Abu Dhabi. The company also has commercial deals with sidewalk delivery bot companies Avride, Car Taken, and Serve Robotics, end quote. Some other hardware for you today. Nothing. unveiled the $379 phone 3A and $459 3A Pro, both with 6.77-inch displays, Snapdragon 7S Gen 3, and new AI tools to help organize files shipping this month, quoting the Verge. The 3A and 3A Pro are mainly differentiated by their cameras, which you'll notice just by glancing at the two devices.

The 3A Pro's prominent round camera housing includes a 3X periscope telephoto lens. The 3A offers a standard 2X zoom, but both phones include a 50-migixel 1.8 main camera and an 8-machyxel ultra-wide. The telephoto cameras on each use a 50-mepixel sensor for a lossless, crop zoom, 4x for the 3A, and 6x for the 3A Pro. They're big phones, each with a 6.77-inch display, and the 3A Pro feels especially chunky with the protruding camera housing. Both use nothing's striking translucent back panel design for a bold look, which balances and awkward camera bump on the 3A Pro.

When I started using the phone, I felt like my fingers were constantly bumping against the housing, but I've adjusted to it after a few days and dig its popsocket-esque functionality. The phones come with Snapdragon 7S Gen 3 chipsets, 12 gigabytes of RAM and 256 gigabytes of storage, which is generous for the mid-range class. They ship with Android 15,

and nothing is promising three years of OS updates and six years of security patches, a decent, if not the best, software policy for a budget phone. They're being offered in the U.S. through nothing's beta program. The glyph interface and LED light strips are still present, but nothing seems to be shifting its efforts towards software features. Something called the essential space is a new place to save screenshots, voice memos, and images like Google's Pixel

screenshots app. It answers to the eternal question, what do I do with this thing? Is your photo gallery cluttered with pictures of stuff? You want to remember? Do you wish you had somewhere to keep all those inspiration photos for your bathroom makeover? Do you yearn for a place to put the information in an email you keep searching your inbox for every time you need it, then you get what the essential space is all about. You save stuff there, it uses AI to pull out relevant bits of information, and it helps organize what would otherwise be left floating around your phone somewhere. Using the essential key to add things to the essential space, took a little adjustment. It's right where I'm used to the power button sitting, so I kept pressing it unintentionally.

A single press will capture a screenshot and a double tap opens the app so you can browse through your collections. This feels backwards for reasons I can't quite explain, but I've mostly gotten used to it. Nitpicking aside, I think nothing is on to something. I added screenshots of travel information for an upcoming flight that are otherwise spread across emails and apps. The essential space keeps it in one tidy spot and is good at pulling key timing and dates from the screen grabs.

It'll even make a little to-do list for you. It didn't quite get everything right about my connecting flight, but I think that's because the date wasn't visible in both screenshots. The software seems to do a decent job when it has complete information to work with, end quote. Finally today, remember the Xbox Connect, briefly around the year 2010 or so, it seemed to be the hot new technology. Then it drifted a bit and was actually discontinued in 2017. Well, The Guardian points out that the Connect technology has lived on,

actually finding applications beyond gaming, including interactive art, pornography, and ghost hunting. Quote, the Connect's technology was soon eclipsed by freely available open-source sensors and more advanced motion sensing devices, but since Microsoft's cease manufacture of the Connect line in 2017, the little camera has enjoyed a spirited and not entirely untroubled afterlife. It has watched over the Korean demilitarized zone and worked on topography and patient alignment in CT scanners. Reports have emerged of it being used in airport baggage halls as a security camera in Newark Liberty International Airport's Terminal C. United Airlines declined to comment on this, and even to gamify training for the U.S. military. It's been attached to drones,

rescue bots, and even found a brief application in pornography. I'm not sure anyone had a firm vision of what interactive sex involving the Connect would be, says Kyle Macalus, founder of Butplug.io, and another member of the OpenConnect team. The camera was deployed mostly as an over-complex controller for 3D sex games, fulfilling, quote, most of a futurist marketing role than anything the actual consumer uses, Michaelis says. In that role, it was successful. It attracted a flurry of attention and threats from Microsoft to somehow ban porn involving Connect. It was an interesting experiment, but it turned out that the addition to, of a novelty device was not a turn-on for many porn users. Besides, as mechalus says, when the camera

malfunctions, it looks pretty horrible. Unreliability is of less concern for ghost hunters who thrive on the ambiguity of aging technology and who have rebranded the Kinect as the SLS, structured light sensor camera. They deploy its body tracking to find figures the naked eye cannot see. Ghost hunters are thrilled by the Kinex habit of seeing bodies that aren't really there, believing that these skeletal stick figures are representations of disembodied spirits. The paranormal investigation industry doesn't care much about false positives, so long as those false positives can be perceived as paranormal, which is just as well, says John Wood, a freelance science performer who has a show devoted to examining ghost hunting technology. It's quite

normal for ghost hunters to be filming themselves in the dark with infrared cameras and torches. You're bathing the scene with IRR light while using a sensor that measures a specific pattern of infrared dots, he says. Given that connect is designed specifically to recognize the human body in any data it receives, it would be stranger if the connect didn't pick up anomalous figures in this context. There's a certain poetry in the connect living on among those searching for proof of life after death. In the right hands, the camera is still going strong. Dio Watson points me in the direction of Connected Worlds, an exhibit that has run in the New York Hall of Science since 2015. Of the many connect devices that power the installations only two have had to be replaced in the decade since it opened,

and one of those was only a few weeks ago. Watson started stockpiling the device when Microsoft ceased production. Half the projects on the website wouldn't exist without the connect, he says, if we had this camera for another decade, we would still not run out of things to do with it, end quote. A. U.S. judge has denied Elon Musk's effort to halt Open AI's for-profit shift, saying Elon failed to meet the burden of proof, but the judge did let other aspects of the suit proceed. Quoting Bloomberg, U.S. District Judge Yvonne Gonzalez-Rogers wrote in her order that she wants to resolve Musk's claims quickly, given, quote, the public interest at stake and potential for harm if a

conversion contrary to law occurred, Rogers said she will hold an expedited trial solely on the core claim that OpenAI's conversion plan is unlawful and, quote, potentially the interrelated contract-based claims, end quote. The timing is significant because Open AI is already in talks to win approval for its conversion from officials in Delaware and California, a process that's expected to be complex. We're pleased the court has offered an expedited trial on the core claims driving this case, which, in its words, present urgent issues in the public's interests. Mark Toberoff, Musk's lawyer said in an email, we intend to accept the court's offer, end quote. So this sort of seems like an initial setback for Elon's efforts.

Open AI is not immediately frozen in its tracks, but given that we have been told the earliest this trial would happen, would be late 2026 or 2027 even. The fact that this is getting expedited is probably positive for Elon. He'll have his day in court and sooner rather than later. While addressing Congress last night, President Trump said to get rid of the $52 billion CHIPS Act and use the leftover money to reduce debt or any other reason. So there goes that, I guess. Somebody check on Intel, quoting Bloomberg. Your Chips Act is a horrible, horrible thing. The president said in a primetime address to Congress on Tuesday. Trump implored U.S. House Speaker Mike Johnson to get rid of the legislation and use whatever is left over to reduce debt or any other reason. His remarks were

met with applause in a chamber that passed the Chips and Science Act less than three years ago. Vice President J.D. Vance, whose home state of Ohio won a massive Intel project thanks to the law, stood up to show his support for its revocation. The Chips Act is among the most significant U.S. forays into industrial policy in more than a generation. It's set aside $39 billion in grand. France plus loans and 25% tax breaks to revitalize American semiconductor manufacturing as well as $11 billion for chip research and development. The aim was to reduce reliance on Asia for electronic components that power everything from smartphones to massive data centers. Trump, however, has consistently derided a program he regards as a waste of government funds, arguing tariffs would achieve the same outcome while filling coffers. Republicans have also indicated that they want to repeal what they see as social provisions of the chips act.

that could involve eliminating laborer-friendly relations or environmental requirements. Officials on both sides of the aisle have touted the Chips Act as crucial to U.S. National and Economic Security, and Trump could have a hard time getting congressional support to repeal it. Dozens of GOP lawmakers voted for the measure, and many red districts have won factories or other projects supported by the law, end quote. More product unveilings from Apple this morning. Apple updated the Max Studio with an M4 Max Chip, starting at $1999 plus. and an M3 Ultra, starting at $3,99.99 plus with up to 32 cores, including 24 performance cores, shipping from March 12th. Quoting the verge.

Apple says the M4 Max Studio is up to 3.5 times faster than the original M1 Max version version, with a 14 to 16 core CPU and 32 to 40 core GPU. Like the M4 Max MacBook Pro, this version of the studio starts with 36 gigabytes of RAM, up from 32 gigabytes in the M2 model and can be had with as much as 128 gigabytes, a bump from the 96-gibite ceiling of the M2 Mac Studio. Like its outgoing predecessor, it starts with 512 gigabytes of SSD storage, but can go as high as 8 terabytes. Meanwhile, the Mac Studio with an M3 Ultra Chip sounds like it's going to scream.

It gets up to 32 cores, 24 of which are performance cores, something Apple Notes is 50% more than any previous ultra chip. The GPU has a base 60 core, configuration that maxes out at 80 cores, and Apple says it has a 32-core neural engine for machine learning and AI applications. The M3 Ultra version of the studio has 96 gigabytes of RAM to start, but it can go up to a gobsmacking 512 gigabytes of RAM, enough to run some of the largest local AI models locally. Finally, you can bump the base 1 terabyte internal storage to as high as 16 terabytes, end quote. But wait, there's more. Apple updated the $999.

$1.13 inch and $1,190.15-inch MacBook airs with an M4 chip, offering a 10-core CPU and an 8-core GPU, and a blue color option shipping from March 12th. But speaking of chips, again, say hello to the M3 Ultra, built from two three-nometer M-3 Max chips, offering Thunderbolt 5, a new Ultra Fusion packaging architecture, and 80-core GPU, and up to 512 gigabytes of memory. Quoting Tom's hardware. In a demo, I saw the M3 Ultra run Cinema 4D where an artist wanted to spread foliage around the landscape. Using LM Studio, they created a Python script to scatter the assets, pasted it into Cinema 4D, and it was done.

What could have taken a day took just minutes. From there, they were able to open Maxon Redshift to see a high-quality preview with hardware ray tracing. In addition, I saw, but did not play an early demo of Cyberpunk 2077, which is coming this year for Max, running on the hardware, locked at 60 frames per second, thanks to V-Sync on the monitor, with full ray tracing. Remember Dig, arguably the first social network of the Web 2.0 era at a breakthrough?

If you don't remember Dig, you might only know that it was the social network that Reddit dethroned. Reddit is what Dig was trying to be, but thanks to a bad redesign, its community decamped to Reddit en masse, and the rest is history. Well, Dig founder Kevin Rose has repurchased Dig, the site he founded in 2004, with plans to resurrect it as a competitor to Reddit, ironically with help from Reddit co-founder Alexis O'Hanian. The two former rivals have joined forces to reimagine what a community-focused platform could be in today's fractured social media landscape. The partnership represents a surprising twist in their shared history back in 2005. O'Hanian had identified Dig as the enemy when he and Steve Huffman launched Reddit.

While Dig initially thrived, raising millions in funding and fielding acquisition offers from companies like Google, it eventually faltered following an unpopular redesign. Users abandoned the platform, and in 2012, Dig was sold off in pieces while Rose departed, but quoting the verge. Rose and a group of what he calls brainstorming partners, which included Ohanian, design and product exec, Justin Mazzle, and even folks like blogger and Twitter co-founder Ev Williams started to talk about whether AI might be able to help them build a better social platform. I would call Alexis up and we would chat, Rose says, and we'd be like, hey, what if, what if? And a lot of those things started giving us both butterflies in the stomach situation where you're like, oh, this could be cool, this could be really cool, end quote.

The new dig will initially launch as an invitation-only mobile focus platform in the coming weeks. Rose and O'Hanian have raised funding from their respective venture firms, true ventures and 776, and assembled a small team of engineers and designers. Mezzell will serve as CEO with Rose as board chair and O'Hanian joining the board. The revamped platform will leverage AI to enhance user experience and moderation, potentially offering features like translating discussions into fictional languages and reducing spam and harassment. Most crucially, they plan to prioritize empowering, moderators with better tools, an area both entrepreneurs believe has been neglected in social media.

What we never focused on is the back end, Ohanian reflected, but it's the back end that really, really matters. Rose acknowledges the initial reaction may be muted with some viewing it as merely nostalgic. However, he believes dig's nimbleness will be advantageous against larger, slower-moving competitors. We won't have everything we want dig to be on day one, Rose said, but a year from now, we will be having a very different conversation, end quote. There have been so many stories recently. of this crypto company that was being investigated, but the investigation has been called off, that crypto company facing regulatory scrutiny, but it's been called off. So I'm just going to use this as a placeholder for all of that. Yuga Labs, creator of the board ape yacht club NFTs,

says the U.S. Securities and Exchange Commission has closed its investigation into the company without taking enforcement action, adding NFTs are not securities. So, officially now, meme coins are not securities, and NFTs are not securities either, quoting to Crypt. Uga Labs had been under SEC scrutiny since 2022 over whether it's NFT offerings and ape coin token distribution violated federal securities laws. The company welcomed the decision calling it a huge win for the NFT industry. The company also labeled the move as a boon for all creators pushing our ecosystem forward. NFTs are not securities, it said. The floor price for board apes on the NFT marketplace OpenC jumped 3.8% on Monday from 13.39-Eth to a local peak of 13.9Eth.

Still, bored apes, once championed by celebrities as status symbols and part of a broader cultural movement, have seen their value decline sharply. The floor price, which peaked at 153.7Eth nearly three years ago during the last crypto bull run, has dropped by about 91% according to Coin Gecko data, end quote. So given this change, regulatory environment, can bigger fish get off the hook? I guess it's worth a try, isn't it? Sources tell Bloomberg that Google representatives asked DOJ officials in a meeting last week to back a way from a push to break up the company citing national security concerns. Quote, the Biden administration in November had called for Google to sell its Chrome

web browser and make other changes to its business, including an end to billions of dollars in exclusivity payments to companies, including Apple. Although Google has previously pushed back on the Biden-era plan, the recent discussions may preview aspects of the company's approach to the case as it continues under the Trump administration. A federal judge is set to rule on how Google must change its practices following hearings scheduled for next month. Both sides are due to file their final proposals to the judge by Friday. Google's argument isn't new, and it has previously raised these concerns in public in response to antitrust pressure from regulators and lawmakers, but the company is re-upping the issue in discussions with officials at the department under Trump

because the case is in its second stage, known as the remedy phase, during which the court can impose sweeping changes on Google's business. After the Justice Department proposed its remedy in November, Kent Walker, Google's chief legal officer, said in a blog post that the department was pursuing a, quote, radical interventionist agenda that would harm Americans and America's global technology leadership. In a separate post in January, unrelated to the department's case, Walker highlighted the company's work with the government in using artificial intelligence to police cybersecurity threats. During their meeting last week, company representatives argued that Google's critical importance to the U.S. economy and national security requires a softer touch,

according to the people. They didn't raise specific threats from the DOJ proposed changes. The people said, end quote. Finally today, let me tell you about Sesame AI, because apparently their speech model demo has many users expressing astonishment at its realism, with some testers reporting emotional connections to the voice assistant. Quoting Ars Technica, I tried the demo, and it was genuinely startling how human it felt, wrote one Hacker News user who tested the system. I'm almost a bit worried I will start feeling emotionally attached to a voice assistant with this level of human-like sound.

In late February, Sesame released a demo for the company's new conversational speech model, or CSM that appears to cross over what many consider the uncanny valley of AI-generated speech, with some testers reporting emotional connections to the male or female voice assistant Miles and Maya. In our own evaluation, we spoke with the male voice for about 28 minutes, talking about life in general and how it decides what is right or wrong based on its training data. The synthesized voice was expressive and dynamic, imitating breath, sounds, chuckles, interruptions, and even sometimes stumbling over words and correcting itself. These imperfections are intentional. At Sesame, our goal is to achieve voice presence, the magical quality that makes spoken interactions feel real, understood, and valued, writes the company in a blog post. We are creating conversational partners that do not just process requests they engage in genuine dialogue that builds confidence and trust over time. In doing so, we hope to realize the untapped potential of voice as the ultimate interface for instruction and understanding, end quote.

Another example, a demonstration video on Reddit shows a simulated argument between a boss and an embezzling employee where it's apparently difficult to distinguish the human from the AI. The system achieves its realism through two collaborative AI models based on meta's Lama architecture. Rather than using the traditional two-stage approach of many text-to-speech systems, Sesame's CSM integrates processing into a single-stage multimodal transformer that jointly handles text and audio tokens. Their largest model uses 8.3 billion parameters and was trained on a process. approximately one million hours of primarily English audio. In blind tests without context, human evaluators showed no clear preference between CSM-generated speech and actual human recordings. However, with conversational context, evaluators still preferred real human speech,

indicating remaining gaps in fully contextual speech generation. Sesame AI acknowledged current limitations on hacker news, noting the system is still too eager and often inappropriate in its tone, prosody, and pacing with issues in conversational flow. Today, we're firmly in the valley, but we're optimistic we can climb out, they wrote. The technology has evoked strong responses. One Reddit user wrote, this is the first time I've experienced something that made me definitely feel like we had arrived. The first time I've had a real genuine conversation with something I felt was real. Many describe the experience as jaw-dropping or mind-blowing. Others find it unsettling. Mark Hackman, senior editor at PC World reported feeling freaked out after his interaction

with Sesame, noting how the AI's voice eerily resembled someone from his past. Some users have compared it favorably to OpenAI's advanced voice mode for ChatGBTGBT, GBT, particularly appreciating Sesame's willingness to role play confrontational scenarios that chat GPT refuses. Despite its technical impressiveness, such advancements carry significant risks for deception and fraud. The ability to generate convincingly human-like interactive speech could supercharge voice fishing scams, allowing criminals to impersonate family members, colleagues, or authority figures with unprecedented realism. Sesame plans to open-source key components under an Apache 2.0 license. Their roadmap includes scaling up model size,

increasing dataset volume, expanding language support to over 20 languages, and developing fully duplex models for more natural conversations. On the tech IPO watch, things are starting to shake loose. Case number one, Klarna is set to file for a $1 billion-plus IPO as soon as next week, quoting Bloomberg. The payments business aims to price the APO in early April, the people said, asking not to be identified as the information isn't public. Klarna is targeting a valuation of more than $15 billion in the New York Stock Exchange listing, the people said. Klarna is best known for its offering of buy-now pay-later loans, which typically allow

users to make a purchase and split their payments into equal interest-free installments. The company has amassed 85 million customers around the world and 600,000 retail partners. Klarna has seen estimates of its valuation fluctuate wildly in recent years. The figure reached a high of $45.6 billion in a funding. round in 2021, only to plunge to $6.7 billion in around the following year. Analyst's last year gave it an implied valuation of around $14.6 billion based on shareholder Chrysalis' investment's estimate of the worth of its stake. Chief Executive Officer Sebastian Simatowski said last month the firm, known for its buy-now pay-later offerings, is exploring an expansion into the cryptocurrency

market. He posted, I give up on social media platform X on February 8th, saying Klarna will embrace crypto. Klarna was founded 20 years ago in Stockholm, Sweden, but in 2023, it began setting up a new UK holding company as part of its preparatory work for an eventual public offering. The company has since refocused ahead of the planned IPO shedding businesses focusing on payment partners and investing in artificial intelligence. In recent months, Klarna agreed to divest its checkout payments business for about $520 million, while it also snapped up Laybuy, a provider of buy-now pay-pay services in New Zealand, end quote. And now case number two, the Times says Discord is in early talks with bankers about a potential IPO that could come as soon as later this year. Discord was last

valued by investors in 2021 at around $15 billion. Quote, Discord was founded by Jason Citron and Stanislav Vishnevsky, two tech workers who originally came together to build a video games studio. After releasing their first title, the two hit upon the idea of Discord after realizing they had difficulty communicating with each other online while playing a game. They released the first version of Discord on desktop computers and smartphones in May 2015. Gamers embraced the platform, which allows people to chat over text, voice, and video calls inside the app for free. Discord also saw a boom in use during the early days of the pandemic. The company held acquisition talks with Microsoft in 2021 for a reported $10 billion, though no deal materialized.

More than 200 million people use Discord every month with 93% of them using the app to play games. The company makes money through a freemium model in which people can use the apps for free and pay subscription fees for upgraded features like custom emojis and higher quality voice and video chat. Discord does not rely on traditional advertising to generate revenue. In recent years, Discord has tried to expand its appeal embracing influencers, musicians, and other figures who have created their own Discord chat rooms for their fans to gather and form online communities, end quote. Here's another one.

Alibaba has released QQ32B, an open source reasoning model on hugging face and model scope, claiming performance similar to Deepseek R1 with lower compute needs. Quoting Venture Beat, Quen team, a division of Chinese e-commerce giant Alibaba, developing its growing family of open source Quinn large language models, has introduced QWQ32B, a new 32 billion parameter reasoning model designed to improve performance

on complex problem-solving tasks through reinforcement learning, or RL. QWQ, short for Quen with Questions, was first introduced by Alibaba in November 2024 as an open-source reasoning model aimed at competing with OpenAIs O1 preview. Traditional instruction-tuned models often struggle with difficult reasoning tasks, but the Quinn team's research suggests that RL can significantly improve a model's ability to solve complex problems. QWQ32B builds on this idea by implementing a multi-stage RL training approach to enhance. enhance mathematical reasoning, coding proficiency, and general problem solving. For example,

while DeepSeek R1 operates with 671 billion parameters with 37 billion activated, QWQ32B achieves comparable performance with a much smaller footprint, typically requiring 24 gigabytes of V-RAM on a GPU, NVIDIA's H-100s have 80 gigabytes, compared to more than 1,500 gigabytes of V-RAM for running the full Deep-Seek R1-16 NVIDIA-100 GPUs, highlighting the efficiency of Quinn's RL approach. For enterprise leaders, including CEOs, CTOs, IT leaders, team managers, and AI application developers, QWQ32B represents a potential shift in how AI can support business decision-making and technical innovation.

With its RL-driven reasoning capabilities, the model can provide more accurate, structured, and context-aware insights making it valuable for use cases such as automated data analysis, strategic planning, software development, and intelligent automation. Companies looking to deploy AI solutions for complex problems, solving, coding assistance, financial modeling, or customer service automation may find QWQ32B's efficiency and attractive option. Additionally, its open weight availability allows organizations to fine-tune and customize the model for domain-specific applications without proprietary restrictions, making it a flexible choice for enterprise AI strategies. The fact that it comes

from a Chinese e-commerce giant may raise some security and bias concerns for some non-Chinese users, especially when using the Kwanat chat interface. But as with Deepseek R1, the fact that the model is available on Hugging Face for download and offline usage and fine-tuning or retraining suggests that these can be overcome fairly easily, and it is a viable alternative to DeepSeek R1, end quote. In a while, talk about going upmarket. The information says OpenAI told some investors about plans for new agents that would be serious level in lots of different ways, including price. quote, Open AI executives said it plans to sell low-end agents at a cost of $2,000 per month to high-income

knowledge workers, mid-tier agents for software development costing possibly $10,000 a month, and high-end agents acting as Ph.D-level research agents, which could cost $20,000 per month, according to the person who's spoken with executives. In the long run, OpenAI expects 20 to 25 percent of the company's revenue to come from agent products, the person said. Our past reporting has given hints about each of these types of agents. OpenAI recently demonstrated ChatGPT's ability to sort through and rank sales leads, an example of the sort of high-income knowledge worker work its models can handle. In January, we broke news that OpenAI has been working on a coding assistant product

targeting senior software engineers and explained the types of tasks it might do. That sounds like the mid-tier agent OpenAI discussed with investors. In November, we reported how researchers at Lawrence Livermore National Laboratory have been using OpenAI's O1 model to solve questions related to nuclear fusion. Presumably, that reflects a Ph.D. level of intelligence or something close to it. How to price AI applications is still a work in progress across the industry. Some companies are stuffing AI features into app bundles like workspace and Office 365 and increasing the price, while others are charging for AI features only when they solve a task.

Meanwhile, Devin, the coding tool created by startup cognition, costs around $500 a month for a coding agent similar to what OpenAI hopes to launch. As a reminder, OpenAI's most expensive subscription tier today, ChatGPT Pro costs $200 a month and was growing nicely as of several weeks ago. But ChatGPT Pro isn't technically an agent the way the Ph.D-level research agent or coding agent would be, though buying the subscription does give you access to OpenAI's web browsing operator agent. It stands to reason that customers would be willing to pay a lot more for AI that effectively does the work of a $200,000 per year software engineer,

to speed up important medical research. Such potential customers are probably getting a great deal with pro, which could be the reason that tier is still unprofitable, end quote. Google is rolling out in experimental AI mode in Google search designed to let users ask complex multi-part questions to Google One AI premium subscribers right now, quoting TechCrunch. Google is launching a new AI mode, experimental feature in search that looks to take on popular services like Perplexity AI and OpenAIs chat GPT search. The tech giant, announced on Wednesday that the new mode is designed to allow users to ask complex multipar questions and follow-ups to dig deeper on a topic directly within Google search.

The feature uses a custom version of Gemini 2.0 and is particularly helpful for questions that need further exploration and comparisons, thanks to advanced reasoning, thinking, and multimodal capabilities. For instance, you could ask, what's the difference in sleep tracking features between a smart ring, smartwatch, and tracking mat? AI mode can then give you a detailed comparison of what each product offers, along with links to articles that it's pulling the information from. You could then ask a follow-up question, such as what happens to your heart rate during deep sleep to continue your search. Google says

that in the past, it would have taken multiple queries to compare detailed options or explore a new concept through traditional searches. With AI mode, you can access web content, but also tap into real-time sources like the knowledge graph info about the real world and shopping data for billions of products, end quote. Now, related to that, A new toll-bit analysis of 160 websites on their content licensing platform has found that AI search engines drive 96% less referral traffic to websites than traditional Google search. All right, for the umpteenth time, we've got rumors of a foldable iPhone. But this one is from Ming Chi Quo, and this has an actual timeline associated with it. So we have the opportunity for some actual verifiable claim chowder, as Gruber likes to call it.

Quoting Mac rumors. Apple's first foldable iPhone should arrive around the end of 2026 or early 27 with a book-style design and a premium price tag of over $2,000, according to analysts Ming Chi Quo. In a report today, Quo outlines his expectations for the device, noting that it will have an approximately 7.8 inch crease-free inner display and a 5.5-inch outer display matching a rumor from last month. Quo says that the foldable iPhone will measure between 9 and 9.5 millimeters when folded and around 4.5 to 4.8 millimeters when unfolded. It will feature a titanium alloy casing with a hinge made of stainless steel and titanium alloy, and it will have a dual lens rear camera with a front-facing camera for both folded and unfolded states. Notably,

Quo says that the phone will forego face ID authentication instead using a touch ID side button as a means to save precious internal space. It will also be positioned as a true AI-driven phone with the larger screen allowing for a better multitasking AI experience. With premium pricing above $2,000 and perhaps even above $2,500, Quo believes the foldable iPhone will tap into Apple's loyal fan base, some of whom will view it as a must-have device if its quality meets expectations. Quo says Apple is aiming to finalize the specs of its first foldable iPhone during the second quarter of this year and officially kick off the project in the third quarter.

Mass production is reportedly planned for the fourth quarter of 2026, so we could perhaps see a launch late next year or early 2027. A second generation model is said to be in the pipeline for a year later, end quote. And actually, Apple folks, I've got some good news for you. Those just announced M4 MacBook airs support two monitors of up to 6K resolution each, plus the built-in display simultaneously, thereby addressing a noticeable limitation of previous models. Quoting Apple Insider, while the previous M3 model featured support for, two monitors, it was only possible to use both when the MacBook Air's lid was closed. So while users

gained the wider screen real estate of two external monitors, they lost the use of their MacBook Air internal display. Now the M4 MacBook Air supports two monitors of up to 6K resolution each, plus the built-in 3.6 inch or 15.3 inch screen. That 6K resolution is also an improvement. The previous model did support two external monitors, but only one could be 6K, the other could be driven at up to 5K resolution at 60 hertz. Either screen could be run at 4K resolution at 144 hertz, end quote. Finally today, I miss this, but you YouTube heads out there, you've got a new cheaper option that is actually pretty compelling if you spend even a half dozen hours or so a month watching YouTube, quoting TechCrunch. YouTube on Wednesday announced a new premium subscription tier

called Premium Light, which will allow users to watch most videos ad-free for $7.99 per month. YouTube Light, customers will be able to watch videos across verticals like gaming, fashion, beauty, cooking, news, and more ad-free, but will be shown ads on music content and music videos. As an entry-level subscription, the new tier won't offer several benefits of the full YouTube premium 1399 per month subscription like downloads, background play, or the ability to watch music videos ad-free. The expansion serves as a way for YouTube to generate more revenue from subscriptions which could help fuel creators' businesses. The new Light Tier may make the most sense for younger viewers who have largely replaced TV time with watching online

videos on sites like YouTube and TikTok. It will also make watching videos on the big screen in the living room more appealing, as there will be fewer interrupting ads. YouTube recently said that the majority of its viewing in the U.S. is now taking place on televisions, which has surpassed mobile devices and desktops, end quote. Yes, speaking of the subscription play, YouTube says YouTube music and YouTube premium now have more than 125 million subscribers global. quoting music business worldwide. The platform, which also operates the TikTok rival Shorts, has surpassed the significant milestone of 125 million paid YouTube music and premium subscribers worldwide, including trials. The milestone means that YouTube has added more than 25 million subscribers over

the past 12 months, which works out to slightly over 2 million subs per month on average since its last subscriber announcement of 100 million in February 2024. A question the entire music business will likely be asking today is how does YouTube's subscriber growth compared to that of the world's largest subscription music streaming platform Spotify. In the 12 months to the end of December, Spotify added 27 million net-paying subscribers to reach 263 million paying users. YouTube's latest subscriber update was revealed by YouTube's global head of music, Lyre Cohen, in a letter to the music industry today, March 5th, celebrating the subscriber milestone and the platform's 20th anniversary. This year, YouTube celebrates its

20th anniversary, wrote Cohen from a simple video, we've grown into the largest music stage in the world reaching over 100 countries in 80 languages. We've accomplished so much together over the last 20 years, but if you think we're slowing down, you're dead wrong. In the latter, Cohen said that 125 million paid subscribers is an incredible milestone that many laughed off as impossible when we first launched. He added, this momentum is critical to our goal of becoming the number one contributor of revenue to the industry, and we won't stop until we get there. YouTube last announced in September 2022 that it had paid music rights holders over $6 billion in the 12 months to the end of June 2022. Elsewhere in today's letter, Cohen explained how YouTube is feeding its subscriptions

funnel by expanding the platform's premium light pilot to U.S. users, end quote. A Brazilian judge has given Apple 90 days to allow side-loading on iOS in Brazil, saying the limitations imposed by Apple on developers could deter new competition. As ever with these things, the question you have to ask is, is this a big enough market for Apple to feel compelled to comply? Quoting 9 to 5 Mac. In November 2024, the Brazilian antitrust regulator Cade ruled that Apple can no longer prevent developers from selling content and distributing apps outside the App Store in Brazil. The company would have 20 days to comply with Brazil's antitrust legislation,

otherwise it would be fined more than $40,000 a day. Unsurprisingly, Apple appealed the decision and a judge ruled that the injunction was unnecessary, giving the company more time to discuss the case. At the time, Apple said that the request to implement changes to the app store wasn't urgent and that it would also affect the company's business. Last month, Apple had to face a public hearing in Brazil concerning the case. But now, Judge Pablo Zwinga has ordered that Apple will have to implement the required changes in Brazil within the next three months. The judge states that despite Apple's claim, the company has already complied with similar obligations in other countries without demonstrating a significant impact or irreparable damage to its

business model. The antitrust investigation against Apple began after Latin American e-commerce giant Mercado Libre accused Apple of forcing developers offering digital goods or services in their apps to use Apple's own payment system. Other companies such as Match, which owns Tinder and Epic Games, have also filed complaints against Apple with the Brazilian regulator. An Apple spokesperson told Valor Economico that the company believes in vibrant and competitive markets and that it faces competition in all segments and jurisdictions where the company operates reinforcing its commitment to its users. Apple believes the changes will harm the privacy and security of iOS users and says it will appeal the decision, end quote. It has happened. President Trump has signed an executive order to

establish a strategic Bitcoin reserve and a digital asset stockpile using BTC and other digital assets forfeited in criminal or civil cases, quoting Coin Telegraph. Additionally, a March 6 fact sheet from the White House said the order also establishes a U.S. digital asset stockpile, which crypto-Zar David Sacks said would be made up of cryptocurrencies other than Bitcoin. Sacks said the Bitcoin Reserve was a digital Fort Knox for the cryptocurrency, and the U.S. wouldn't sell any Bitcoin put in the reserve. It will be kept as a store of value, he added. The White House said that Bitcoin owned by the Treasury Department would cede the reserve, and other federal agencies will evaluate their legal authority to transfer any Bitcoin they own into the reserve. It added,

the Treasury and Commerce Secretaries would make budget-neutral strategies for buying more Bitcoin for the reserve provided that those strategies impose no incremental costs on American taxpayers. As for the digital asset stockpile, Sachs said its purpose is the responsible stewardship of the government's digital assets under the Treasury Department. Notably, he added that the government wouldn't buy additional cryptocurrencies for the stockpile beyond those obtained through forfeiture proceedings, and the Treasury Secretary may determine strategies for responsible stewardship, including potential sales from the stockpile. Arkham Intelligence data shows the U.S. government doesn't hold any XRP, sole, or ADA among the $18.28 billion worth of crypto under its control.

Its largest holding is $198,109 Bitcoin worth $17.87 billion. Its ETH holdings are worth $119 million, the third largest holding behind $122 million worth of the stable coin tether, end quote. Interesting raise, because it continues to highlight how defense startups are all the rage. San Diego-based Shield AI raised $240 million from L3 Harris, A16Z, and others at a $5.3 billion valuation to develop its hive-mined autonomous vehicle platform. Quoting Bloomberg, variations of Shield AI's hive-mine software can pilot autonomous vehicles as well as help companies build their own autonomous drones, robots, and other systems.

Shield AI is most famous for its drone called V-BAT, capable of vertical takeoff and landing. One is prominently displayed in its San Diego offices. But going forward, the company wants to be known for its software. This next phase is really about working with the small and medium businesses that want to operate autonomous hardware across air, land, or sea. Shield AI co-founder Brandon Seng said, we spent a decade and $1 billion plus building this. Shield AI was founded by Seng, a former Navy SEAL and his brother Ryan Seng, an engineer and former technical lead at Qualcomm, who is now the startup's chief executive officer.

Shield AI is one of dozens of defense startups to emerge in recent years with enthusiastic backing from venture capitalists. Investors have poured record amounts into AI, space weapons, and other defense technologies hoping the Pentagon will increase its tech spending as modern warfare evolves. Shield AI's hive-mine software aims to make it easier for more developers and businesses to create autonomous hardware. The company's tools can also help pilot autonomous vehicles ranging from one-way attack drones to F-16s and support swarming operations. Shield says its software can allow vehicles to execute complex missions autonomously, including when GPS and and communications are jammed. Brandon Seng said interest from investors, including defense contractors

and VCs, has recently surged, adding that the company turned down massive checks and that the current round was oversubscribed. Physical AI is the next thing, and we are mobilized against it, he said. Shield AI aspires to service the autonomous needs for the defense sector like Palantir services its intelligence needs, end quote. This is technically not an interesting raise, because he doesn't really need outside money, but sources say, Google founder Larry Page has formed a new company, Dynatomics, to use LLMs to create highly optimized designs for various objects and then have a factory build them. Quote, the stealth company is run by Chris Anderson.

These people said Anderson was previously the chief technology officer of another pageback company, Kitty Hawk, an ambitious project to build small electric airplanes, potentially revolutionizing how people get around cities. The company shut down in 2022 amid failed prototypes and regulatory concerns. Anderson did not respond to a request for comment, and Page could not be reached for comment. Page has remained largely outside day-to-day operations at Google's parent alphabet, where he is still on the board and the most powerful shareholder. His Google co-founder, Sergey Brin, on the other hand, has been going into the office to work on the development of Gemini, Google's large language models. Silicon Valley entrepreneurs are increasingly interested in using AI models to build physical things.

Arcade is one such company using AI to design jewelry, end quote. And confirmation of what Chris and I discussed on the 2000th episode spectacular. Yes, a lot of the hot AI startups right now fit what we called our AI varietals thesis. Quoting Bloomberg, not long ago, Silicon Valley was dismissive of startups like Harvey, while OpenAI developed cutting-edge artificial intelligence models with the potential to shake up almost every industry. Harvey had a more modest goal, building software that makes OpenAI's technology more useful for lawyers. The market's perception of companies like us, that their GPT rappers, said Harvey Chief Executive Officer Winston Weinberg, referencing a derisive

term used to suggest the repackaging of OpenAI's models. If investors were going to put money into something he added, it needed to be into OpenAI or Anthropic. Today, so-called AI rappers are all the rage. Step into any venture capital office in Silicon Valley, and you'll hear investors buzzing about startups that offer AI chatbots, research tools, and other software applications for coding, clinicians, and customer service, all built at least in. in part on the backs of large language models created by other leading AI developers. These startups are seeing revenue and valuations grow at a fast clip, often while spending a fraction of the amount that top AI model developers do on chips,

data centers, and talent. Harvey, founded in 2022, surpassed $50 million in annual recurring revenue in December, Weinberg said. Likewise, any sphere, the startup behind the popular code editing tool cursor, has hit $100 million in annual recurring revenue, according to people familiar with the matter, who spoke on condition of anonymity to discuss private information. Any Sphere did not respond to a request for comment. Michael Magnano, a partner at Lightspeed Venture Partners, likens this moment in AI to the original smartphone app, boom,

nearly two decades ago. Just like after the iPhone launch, there were millions of new mobile apps, said Mignano, an investor in the AI note-taking service Granola, which uses technology from OpenAI and Anthropic. Now, with AI and LLMs, there will be millions of new AI products. Investors are eager to put their money into these services,

Harvey raised a $300 million round earlier this year, led by Sequoia at a $3 billion valuation. AnySphere raised a $105 million round led by Thrive Capital and Andreessen Horowitz in January, valuing it at $2.5 billion. And VC demand is so high that founders like Varan Moan say the conventional fundraising process isn't something we have to do. Mowen's company, Codium, which also offers an AI tool for coders, is currently raising money at a valuation near $3 billion in a round led by Kleiner Perkins, according to a person familiar with the matter. Certainly, VC appetite remains strong for OpenAI and Anthropic as evidenced by their recent mega-rounds, but backing these richly

valued companies is getting too costly for some firms. Investors have also begun to question the wisdom of AI companies pouring billions into developing models in the wake of Deepseek, a Chinese upstart that claims to have built a competitive option for just $5.6 million. And several notable model makers have shifted away from the race to build more advanced AI systems or been partial absorbed by big tech firms, raising fears of more consolidation to come. At this point, it's very clear that the apps are definitely the best place to invest because that is where the revenue is, that is where the customers are, said Jesse Zhang, CEO of Deca Gong, a startup that builds AI customer support agents, in part by using models from OpenAI and Anthropic. The models will get

better and better and cheaper, he added, and the apps will benefit the most from these improvements. Perhaps no company embodies the promise of the AI app era like any sphere, founded in 2022 by four friends who met at the Massachusetts Institute of Technology and had little more than intern-level job experience, Ennisphere has seen soaring growth for its AI-powered code editor, cursor. AnySphere's investors claim it's the fastest-growing software startup of all time, a title previously held by Wiz, a cloud security company that reached $100 million in annual recurring revenue in only 18 months. AnySphere hit that in just 12 months, according to people familiar with the matter. Leaning on a mix of LLMs from other providers,

EniSphere has built a coding tool that developers say saves them time, and it's one a cult following of engineers and tech elites willing to pay for the service. Coinbase Global CEO Brian Armstrong wrote on X that 100% of his company's coders use cursor, end quote. That was a bit of a long read, so let's do this properly. In this week's weekend, Long Read's suggestions, first up a look at McDonald's new tech initiatives, including the use of edge computing to predict equipment breakdowns and computer vision to ensure order accuracy. Quoting the journal, McDonald's tapped Google Cloud in late 2023 to bring more computing power to

each of its restaurants, giving them the ability to process and analyze data on site. The setup known as edge computing can be a faster, cheaper option than sending data to the cloud, especially in more far-flung locations with less reliable cloud connections, said Brian Rice, the Burger Giants' chief information officer. Edge computing will enable applications like predicting when kitchen equipment, such as friars and its notorious McFlurry ice cream machines, is likely to break down, Rice said. The Burger Chain said its suppliers have begun installing sensors on kitchen equipment that will feed data to the edge computing system and give franchises a real-time view into how their restaurants are operating. AI can then analyze that data for early signs of maintenance problems. McDonald's is also exploring the use of computer vision, the form of AI behind facial recognition in store-mounted cameras,

to determine whether orders are accurate before they're handed to customers, he said. If we can proactively address those issues before they occur, that's going to mean smoother operations in the future, Rice added. Additionally, the ability to tap edge computing will power voice AI at the drive-thru, a capability McDonald's is also working with Google's cloud computing arm to explore, Rice said. The company has been experimenting with voice-activated drive-thrus and robotic deepfriars since 2019 and ended its partnership with IBM to test automated order tracking at the drive-thru in 2024.

Edge computing will also help McDonald's restaurant managers oversee their in-store operations. The Burger Giant is looking to create a generative AI virtual manager, Rice said, which handles administrative tasks such as shift scheduling on managers' behalf, fast food giant Yum brands, Pizza Hut, and Taco Bell have also explored similar capabilities, end quote. And we mentioned it earlier this week, so from Veronaughey, A look at YouTube's origins, its growth, and its domination as the platform turns 20 years old, including an interview with CEO Neil Mowen, who is nearing 10 years at the company, quote, When YouTube was incorporated in February 2005 by three former PayPal employees, Chad Hurley, Steve Chen, and Jowad Karam,

the startup was housed in Hurley's Garage in Manlo Park, California. The very first video was uploaded April 23, 2005, by Karam entitled Me at the Zoo. It was a 19-second clip. shot in front of the Elephant Exhibit at the San Diego Zoo. The founders actually launched the platform as a site for video dating, but after one week, not a single person had uploaded any videos. So they pivoted to promoting YouTube as a general purpose video sharing platform. It was a long-shot idea, as it wasn't clear at the time whether the internet could support video delivery at scale, said Rolf Botha, managing partner at Sequoia Capital, the Menlo Park-based venture capital firm

that has backed some of Silicon Valley's most world-beating startups, both that led Sequoia's total investment of $8.5 million in YouTube. The three entrepreneurs were scrappy and smart, he says, explaining why he made the bet on the fledgling service. The site was usable in a way that others were not. Two decades later, the platform is a massive fire hose that spraise out every kind of content imaginable, watched by an estimated 2.44 billion monthly users in 2024, according to researcher e-marketer. There are vlogs, how-to videos, comedy sketches, music videos, news segments, talks shows, product reviews, video game playthrus, life hacks, animated shows, podcasts, TV clips, full movies, and movie trailers, stunts, pranks, and challenges.

More than 500 hours of video are uploaded to YouTube every minute. It's the number two most visited site in the world behind only, Google.com, and the second biggest search engine after Google. Worldwide, users watch on average more than one billion hours of YouTube content on TV's every day, according to the company. In the U.S., TVs recently surpass phones and tablets as YouTube's number one viewing device. According to Nielsen, for two years running, Americans have spent more time watching YouTube on their TV sets than any other streamer, including Netflix, Disney Plus, and Amazon Prime video. In 2024, YouTube generated global ad revenue of $36.15 billion up an impressive 15%.

For the 12 months that ended September 2024, YouTube's subscription revenue topped $15 billion. That was generated by YouTube TV, the biggest internet delivered live TV service in the U.S. with 8 million-plus customers, YouTube Premium, which provides ad-free videos and other perks, and YouTube Music-Music Premium, a music-only streaming service. This week it announced that it now has more than 125 million subscribers for YouTube music and premium services up from 100 million a year ago. As a standalone entity, apart from Google, YouTube would be worth more than $400 billion. Wall Street analyst firm Moffat Nathanson has estimated more than Disney, Comcast,

Warner Brothers Discovery, and Paramount Global combined, end quote.