Tech Brew Ride Home - Tue. 05/07 – The Spring Apple Event

Episode Date: May 7, 2024

All the headlines from today’s Spring Apple event. New iPads? M4 chip? I don’t actually know cause I’m recording this bit beforehand. I can tell you Apple might make its own server chips tho. Nintendo has outlined plans to replace the Switch. And Microsoft is training its own high-end LLM, separate from OpenAI. Links: Apple Is Developing AI Chips for Data Centers, Seeking Edge in Arms Race (WSJ) Nintendo Teases Long-Awaited Switch Successor as Profit Slides (Bloomberg) Meet MAI-1: Microsoft Readies New AI Model to Compete With Google, OpenAI (The Information) Amazon’s Delivery Drones Won’t Fly in Arizona’s Summer Heat (Wired) China hacked Ministry of Defence, Sky News learns (SkyNews) 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. Welcome to the 10. Tech meme right home for Tuesday, May 7th. 2024, I'm Brian McCullough today.

All the headlines from today's spring Apple event. Did we get new iPads and M4 chip? I don't actually know because I'm recording this a bit beforehand. I can tell you Apple might make its own server chips, though. Nintendo has outlined plans to replace the switch, and Microsoft is training its own high-end LLM separate from OpenAI. Here's what you miss today in the world of tech. Apple held its let-loose event earlier than usual this morning.

It was at 7 a.m. Pacific 10 a.m. Eastern time, which was good for me. Here's the whole nine yards. Tim Apple showed up up top and said this was the biggest day of iPad news since the iPad was launched. He did say some things about how people are using the Apple Vision Pro. He swears, trust him. He moved on quickly, though. Then the first bit of hardware was the iPad air. For the first time ever, there are two sizes, a redesigned 11-inch air and a brand-new 13-inch, 30% more screen real estate on the bigger one they say. more space for multiple apps. There's new blue and purple colors. In terms of the internals, the air only gets the M2, but that is a bump up from the M1. 3X faster performance, they say. Both models start at 128 gigabytes of storage, but can go up to 1 terabyte. Price is 599 for the 11-inch. The 13-inch is 799. You can order today, available next week. They're basically the same. No mention of keyboard changes or anything, though they did move the camera to the long side of the bezel. Then it was on to the iPad Pro. We're not only going to push the limits of what you can do on iPad.

We're going to crush them, John Ternis said. A full redesign. Thinner, thinner than the iPod Nano, they said. The thinnest thing Apple has ever produced, they said. The 11-inch iPad Pro is 5.3 millimeters thin. The 13-inch is 5.1 millimeters thin. The 11 is 0.98 pounds, so less than a pound. The 13 inch is 1.28 pounds, and they both have OLED displays for the first time,

which gives you 1,000 nits of brightness and 1,600 nits of peak brightness. That all also gives you better highlights, deeper blacks. They call it the ultra-retna XDR display. There's also a new nanotexture glass option for real pro users out in the field. That's cool. They've never done that before. But the real headline came under the hood. the existing iPads Pro had M2 chips, but they skipped the M3 for these new pros and went all the way to the M4 chip. They say this is what allowed the thinner design and better display.

M4 has a second-generation three-nanometer process. There's a new CPU with four performance cores and six efficiency cores, all with ML accelerators. The CPU is 50% faster than the M2. It's got a 10-core GPU with ray tracing, up to 4x faster performance than the M2. Basically, they say the M4 delivers the same performance as the M2 using half the power. They also want you to know Wall Street that the M4 has the most advanced neural engine yet. Why, Wall Street, Wall Street, Wall Street, to accelerate AI workloads. Did you hear that Wall Street? AI.

Though, as Nelai from the Verge said, the problem with the iPads is not the hardware. Quote, kind of wild, we're talking about all this power that hardly any apps can use. What else? and Logic Pro for iPad got updates. There's a new live multi-cam mode. You can connect up to four cameras at a time. This requires a new app called Final Cut Camera for both iPad and iPhone to record additional video from additional iPhones and iPads. The camera moves from the portrait to the landscape side, just like on the Ares. But there's a brand new, fully redesigned magic keyboard for the pros, which has a function key row now. Feels just like using a MacBook keyboard, they say. There's also an entirely, new Apple Pencil. They're calling it the Apple Pencil Pro. There's a new sensor in the barrel of the

pencil to allow for more interactions. A gyroscope allows you to roll the pencil for other interactions. It also supports Find My for the first time and charges just by putting it on the side of the iPad Pro. Though apparently the Pencil Pro does work with the new airs as well. The new Magic keyboard is $299 for the 11-inch, $329 for the 13-inch Pro. The Pencil Pro costs $129. All of that you can order today, ships next week. For the pros themselves, the base storage for both new models is now 256 gigabytes. The 11-inch iPad Pro starts at $999, and the 13-inch starts at $1,29. So if you add the Magic keyboards, kidding out an iPad Pro costs the same as buying a MacBook. Order today, ships next week. The entry-level 10-inch iPad, by the way, got a price drop down to $329,

and that was it. A big kickoff to the AI era at Apple. This was not. Unrelated Apple news, the journal is reporting that Apple has been working on its own chip designed to run AI software in data center servers. The project is internally codenamed Project ACDC. Now that would be a new revenue line, potentially a big one. Quote, over the past decade, Apple has emerged as a leading player designing chips for iPhones, iPads, Apple Watch, and Mac computers, the server project, which is internally codenamed project ACDC for Apple chips in data centers, will bring this talent to bear for the company's servers, according to people familiar with the matter. Apple has been closely working with its chip-making partner, TSMC, to design and initiate

production of such chips, yet it remains uncertain whether they have yielded a definitive result, some of the people said. For Apple's server chip, the component will likely be focused on running AI models, what's known as inference, rather than on training AI models, where chipmaker Nvidia will likely continue to dominate, according to some of the people. An Apple server chip could come at an opportune time as rivals such as Microsoft and meta have spent billions and moved quickly to reorient their businesses around so-called generative AI after the explosive emergence of OpenAI's chat GPT service. Most of the largest tech companies have developed their own AI server chips or are exploring the possibility to untether themselves from Nvidia, which

has a market share estimated at over 80% in such chips. Google has perhaps the most mature internal AI server chip product with its tensor processing units having started on that project in 2013. OpenAI is also in discussions to fund an ambitious internal chip project. The Wall Street Journal previously reported, Apple's existing chip prowess might be one avenue where the company can find a critical advantage over other tech giants in AI. Among its rivals, Apple has had the longest history and most success in designing custom silicon. The company first began incorporating its own processing chips into iPhones and iPads in 2010, end quote. Nintendo reported earnings yesterday, which we won't go into, but I will note this news.

Nintendo President Shuntaro Furukawa said the company plans to announce its switch successor in the coming year, quoting Bloomberg. Fans and investors have long-awaited word on a succession plan for the 2017 gadget, which has ratcheted up more than 141 million sales over its lifetime. Nintendo has resisted discussing any plans and kept the original console going with marquee game releases. But it lost significant ground to a resurgent Sony with its PlayStation 5 of the past year and will contest the key holiday season against newer and more powerful machines, including in Microsoft's Xbox range. Nintendo told suppliers and software partners earlier this year that it was delaying the launch of the next-generation console until March 2025 or later,

scrapping an initial plan to release it by this year's end. Bloomberg News has previously reported, the creator of the Super Mario Bros. Games is also likely to hold back major releases of its most popular franchises to help boost initial sales of the next machine. Now that this announcement is out in the world, I think Nintendo has no choice but to cut prices or offer attractive hardware and software bundles to get to the 13.5 million switch units they want to sell this year, said industry analysts Sercon Toto. They will get through the switches final year somehow, but I would not expect too many blockbusters until the new hardware arrived.

and quote. Something, something, diversification, not having all your eggs in one basket. The information is reporting that Microsoft is training My One, a new in-house AI model with around 500 billion parameters, which would be large enough to compete with top models from Google Anthropic and OpenAI. Quote, the new model internally referred to as My One is being overseen by Mustafa Suleiman, the ex-Google AI leader who most recently served as CEO of the AI startup inflection before Microsoft hired the majority of the startup staff and paid $650 million for the rights to its intellectual property in March. But this is a Microsoft model, not one carried over from inflection, although it may build on training data and other tech from the startup. It is separate

from the models that Inflection previously released, according to two Microsoft employees with knowledge of the effort. My One will be far larger than any of the smaller open source models that Microsoft has previously trained, meaning it will require more computing power and training data and will therefore be more expensive, according to the people. My One will have roughly 500 billion parameters or settings that can be adjusted to determine what models learn during training. By comparison, OpenAI's GPT4 has more than 1 trillion parameters, while smaller open source models released by firms like meta-platforms and Mistral have 70 billion parameters. That means Microsoft is now pursuing a dual trajectory of sorts in AI, aiming to develop both

small language models that are inexpensive to build into apps, and that could run on mobile devices alongside larger state-of-the-art AI models. It also shows Microsoft's willingness to chart a new path in AI separate from the technology developed by OpenAI, which currently underlies all of the AI co-pilot chatbots in Microsoft's products that can automatically spin up emails or quickly summarize documents. But the exact purpose that the new model will serve hasn't been determined yet and will depend on how well it performs, one of the people said. To trade the new model, Microsoft has been setting aside a large cluster of servers equipped with NVIDIA-made graphics processing units and has been compiling a corpus of training data to improve

the model. Some of that includes data drawn from various data sets that it previously used to train smaller models, including text created by OpenA's GPT4, as well as from other sources such as public data across the internet, one of the people said. Microsoft could preview the new model as soon as its build developer conference later this month, depending on how well development goes in the coming weeks one of the people said, end quote. Weird, quick little thing here. Amazon's new hexacopter drones planned for use delivering items in Tulison, Arizona, can't operate when temperatures top 104 degrees Fahrenheit. I say that's weird, because that is less than the average summer highs in that city. Quoting wired, Amazon plans to start flying delivery drones in

Arizona this year, but don't count on them to bring you a refreshing drink on a hot day. The hexacopter can't operate when temperatures top 104 degrees Fahrenheit or 40 degrees Celsius, the company says, and average daily highs exceed that for three months of the year in Tulison, the city outside of Phoenix, where Amazon is preparing to offer aerial deliveries from inside a seven-and-a-half-mile radius. The drones can't help with midnight snacks either because they'll be grounded after sunset. Potentially being inoperable for a quarter of the year might make launching drone deliveries in Tulison and neighboring desert communities seem like an odd choice. It's far from the first challenge, faced by Amazon's much-delayed drone project. The unit is years behind its goals of flying items to

customers and under an hour on a regular basis, and a one-time target of 500 million deliveries by 2030 seems distant. Amazon Prime Air has completed just thousands of deliveries falling behind rivals. Alphabet subsidiary Wing has notched hundreds of thousands of delivery flights and Walmart, more than 20,000. In the California wine country town of Lockford, where Amazon initially launched drone deliveries, some residents told Wired last year that they ordered. only because Amazon lured them with gift cards. In Arizona, it could be discouraging not being able to rely on drones during those hours when one might not want to venture too far from the comfort of air conditioning. Before its drones start making deliveries, Amazon has more than summer heat to contend with. The company needs to obtain some local permits to renovate its tollison warehouse to host and operate the drones. The U.S. Federal Aviation Administration has to sign off on Amazon's operating plans for its new drone, known as the MK30, which the online shopping giant wants to use in Arizona. And after that, Amazon must persuade users to sign up to have an 80-pound six-rotor drone flying into their yard to drop a box from several feet up onto the giant

QR code mat that the craft uses to identify a customer's drop-off point, end quote. And finally today, this seems bad. Sources say that China has allegedly hacked a payroll system used by the United Kingdom's Ministry of Defense targeting service personnel, military personnel, names, and bank details have apparently been exposed, quoting Sky News. Members of Parliament will be told today of a massive data breach involving the Ministry of Defense targeting service personnel. The government will not name the country involved, but Sky News understands this to be China. The Chinese state is to be accused of two or three attempts at hacking MOD employees, including personnel. The cyber attack was on a payroll system with current service personnel and some veterans.

It is largely names and bank details that have been exposed. All salaries will be paid this month. China's foreign ministry said it, quote, firmly opposes and fights all forms of cyber attacks and rejects the use of this issue politically to smear other countries, end quote. Tobias Elwood, a conservative MP and former soldier told Sky News that China, quote, was probably looking at the financially vulnerable with a view that they may be coerced in exchange. for cash. Defense Secretary Grant Shaps will make a statement to the Commons today with the BBC reporting that he will set out a multi-point plan, which will include action to protect affected servicemen and women. The MOD hopes serving personnel will not be concerned about their safety.

They will be provided with advice and support. The contractor system is not connected to the main MOD computer system and has been taken down with a review launched. The MOD has been working at speed over the last 72 hours to understand the scale of the hack after it was discovered in recent days. It is understood investigations have not so far shown any data has been taken. This could raise questions about whether other countries with challenging relationships with China will want to share sensitive intelligence with the UK. This comes fewer than two months after China's state-affiliated actors have been blamed by the government for two malicious cyber campaigns in the UK.

Government Minister Mel Stride told Sky News on Tuesday morning, China is, quote, an epic defining challenge and, quote, our eyes are wide open, quoting the integrated review of the UK's national security and international policy completed in 2021, end quote. Busy, busy day today. I got up early and pre-recorded the second half of the show. I've got a rush now to get this Apple event segment recorded and then run into the city for that AI demo day at Betawks. Not sure this will come out in time, but if anybody listening to me right now is at BetoWorks in New York City today, feel free to come up to me and say hi.

And tomorrow, just note, the show will be significantly late. I've got a doctor's appointment in the city at 9.30. I couldn't schedule any other time. So until that is done, and until I can get back to Brooklyn, I basically can't work on the show at all. So I wouldn't expect tomorrow's show to drop it before, I don't know, 3 p.m. 4 is a possibility. 3 is maybe the best case scenario.

Talk to tomorrow whenever I actually can.