Tech Brew Ride Home - Tue. 06/06 – The SEC Sues Coinbase

Episode Date: June 6, 2023

The SEC has officially sued Coinbase. Sequoia plans to split into 3 separate VC firms. Why did Apple announce a bunch of AI features yesterday but shy away from name dropping AI? And all anyone really cares about right now: two different hands on experiences with the Apple Vision Pro. What is this thing like to actually use? Sponsors: Collective.com Links: US Crypto Crackdown Reaches Fever Pitch as SEC Sues Coinbase (Bloomberg) Sequoia Is Splitting Into Three VC Firms (Forbes) Apple avoids “AI” hype at WWDC keynote by baking ML into products (ArsTechnica) First impressions: Yes, Apple Vision Pro works and yes, it’s good (TechCrunch) I wore the Apple Vision Pro. It’s the best headset demo ever. (The Verge) 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 TechMame Right Home for Tuesday, June 9th, 2020. I'm Brian McCullough today. The SECC has officially sued Coinbase.

Sequoia plans to split into three separate VC firms. Why did Apple announce a bunch of AI features yesterday but shy away from actually name-dropping AI? And all anyone cares about right now, two different hands-on experiences with the Apple Vision Pro. What is it like to actually use this thing? Here's what you miss today in the world of tech. There it is.

The SEC this morning officially sued Coinbase, alleging the exchange has acted as an unregistered broker since at least 2019 and broke various rules by offering staking services. Shares of the company are down more than 20% at the time of this writing this morning, quoting Bloomberg. The regulator said in a 101-page complaint that Coinbase, the largest US crypto platform, evaded regulations by letting users trade numerous crypto tokens that were actually unregistered securities. The move comes the day after the SEC sued Binance, the world's largest crypto platform for a wider set of violations as part of a growing crackdown that could push digital currencies back to the fringes of the U.S. financial system. We allege that Coinbase, despite being subject to the securities laws, commingled and unlawfully

offered exchange, broker-dealer, and clearinghouse functions, SEC chair, Gary Gensler said in a statement. The SEC is seeking an order that would require Coinbase to comply with securities laws and give up what the agency says were ill-gotten gains. The SEC also accused Coinbase on Tuesday of breaking the agency's rules with its staking service. That product lets customers turn over their crypto tokens to facilitate transactions on a blockchain, which then pays a return to the customer, end quote. So it's going to be interesting to find out which coins are actually named here. Like in the Binance lawsuit yesterday, the SEC cited 12 specific coins, among them, Binance's own BNB coin, the stable coin BUSD, Solana's Solana's Sol, Cardano's ADA, those are big names.

So like, does Coinbase have to delist those coins and give money back and then that's it? Maybe also shut down the staking service. Go back to just trading the big names like Bitcoin and ETH and stuff. It's unclear at the time of this writing. This is perhaps a bit insight. baseball, but Sequoia, thought by many, to be the gold standard for modern venture capital firms, has announced plans to split into three different firms. The Sequoia branding will remain for the firm centered in the U.S. and Europe, but there will be new branding in China, Hong Shan,

and peak 15 in India and Southeast Asia, at least I think it's 15, it's peak XV, I'm assuming the Roman numerals stand for 15, quoting Forbes. In separate interviews with Forbes, the three investment has, of the three firms said the decision to split up Sequoia's global brand was a gradual discussion that intensified over the last several months. They cited conflicts between the fund's respective startup portfolios, brand confusion as they diverged in strategies and increasing complexity of maintaining centralized regulatory compliance as factors, while acknowledging but attempting to downplay a frostyre geopolitical environment. Whereas Sequoia's U.S. business, which has expanded to include Europe and Israel, could more recently claim standouts such as

Airbnb, DoorDash, Snowflake, WhatsApp, and Zoom, Sequoia, China could boast of their own long list, including Alibaba and Maytwan, as well as TikTok parent bite dance. The India and Southeast Asia funds could point to Baiju's go-to and Zamato, among others. Across its funds, Sequoia regularly placed the most partners on Forbes's Midas list, the annual ranking of the world's top venture capitalists with 10 investors in 2023, led by Shen, who ranked number one for the fourth time. Sequoia investor has taken the top midas slot in half of its 22-year history. But from the start, Sequoia considered its regional funds largely independent with decentralized deal flow and portfolio decision-making. Partners from one geography would not review potential deals by another's.

Instead, the funds shared back-office functions, including compliance, finance, and investor relations, basic infrastructure, and an online portal for limited partners. Investors in those different regional funds overlapped, and partners often invested personally in each other's funds. But the regions had already diverged in some respects, the partner said, with investor relations becoming more localized and funds setting up their own software. In their separate interviews, both the Shen and Singh all denied that geopolitical tensions were a specific catalyst for the move. Conflict between their broadening portfolios played more of a factor, they all said, end quote. Yeah, but the China-U-S pseudo-Cold War had to play some factor here, right?

Time to turn to analysis of yesterday's big Apple event. Over in Ars Technica, Benj Edwards, notes that Apple actually avoided mentioning AI specifically at WWDC in a lot of their announces, even while they were touting features clearly made possible by AI. Instead, they kept referring to machine learning and transformers while describing these features, which is interesting. Remember, the only major tech player yet to tout its AI. bonafides to the ceiling is Apple? I wonder why they're shy here. Quote. Notably, Apple mentioned the

AI term transformer in an Apple keynote. The company specifically talked about a transformer language model, which means its AI model uses the transformer architecture that has been powering many recent generative AI innovations, such as Dolly Image Generator and the ChatGPT chatbot. A transformer model, a concept first introduced in 2017, is a type of neural network architecture used in natural language processing that employs a self-attention mechanism, allowing it to prioritize different words or elements in a sequence. Its ability to process inputs in parallel has led to significant efficiency improvements and powered breakthroughs in NLP tasks, such as translation, summarization, and question answering. Apparently, Apple's new transformer model in iOS 17

allows sentence-level auto-corrections that can finish either a word or an entire sentence when you press the space bar. It learns from your writing style as well, which guides its suggestions. All this on-device AI processing is fairly easy for Apple because of a special portion of Apple silicon chips and earlier Apple chips, starting with the A-11 in 2017, called the neural engine, which is designed to accelerate machine learning applications. Apple also said that dictation, quote, gets a new transformer-based speech recognition model that leverages the neural engine to make dictation even more accurate, end quote. So clearly Apple is playing in these AI waters, they're just not getting all buzzworthy about it. An interesting choice in and of itself. But then also,

there was this about the new M2 chips, quoting again. During the WWDC keynote, Apple unveiled its most powerful Apple silicon chip yet, the M2 Ultra, which features up to 24 CPU cores, 76 GPU cores, and 32-core neural engine that reportedly delivers 31.6 trillion operations per second, which Apple says represents 40% faster performance than the the M1 Ultra. Interestingly, Apple directly said that this power might come in handy for training, quote, large transformer models, which, to our knowledge, is the most prominent mention of AI in an Apple keynote, albeit only in passing. This development has some AI experts excited. On Twitter, frequent AI pundit Perry E. Metzger wrote, quote, whether by accident or intent, the Apple Silicon

Unified Memory Architecture means high-end Macs are now really amazing machines for running big AI models and doing AI research. There really aren't many other systems at this price point that offer 192 gigabytes of GPU accessible RAM, end quote. Here, larger RAM means that bigger and ostensibly more capable AI models can fit in memory. The systems are the new Mac Studio starting at $1,99, and the new Mac Pro starting at $6,99, which could potentially put AI training within reach of many new people and in the form factor of desktop and tower size machines, end quote. All right, as promised, let's get some hands-on impressions that folks had of the Apple Vision Pro yesterday during the curated hands-on experience Apple allowed select people to have.

The responses were interesting. I'm going to give you two. First up, Matthew Panzerino at TechCrunch, quote, After a roughly 30-minute demo that ran through the major features that are yet ready to test, I came away convinced that Apple has delivered nothing less than a genuine leapfrog in capability and execution of XR, or mixed reality with its new Apple Vision Pro. To be super clear, I'm not saying it delivers on all promises, is a genuinely new paradigm in computing or any other high-powered claim that Apple hopes to deliver on once it ships.

I will need a lot more time with the device than a guided demo. But I've used essentially every major VR headset and AR device since 2013's Oculus-DK-1, right up through the latest generations of Quest and Vive headsets, none of them had the advantages that Apple brings to the table with Apple Vision Pro, namely 5,000 patents filed over the past few years and an enormous base of talent and capital to work with. Every bit of this thing shows Apple-level ambition. I don't know whether it will be the next computing mode, but you can see the conviction behind each of the choices made here.

No corners cut. Full-tilt engineering on display. The hardware is good. very good. With 24 million pixels across the two panels, orders of magnitude more than any headsets most consumers have come into contact with. The optics are better. The headband is comfortable and quickly adjustable, and there's a top strap for weight relief. Apple says it is still working on which light seal, the cloth shroud, options to ship with it when it releases officially, but the default one was comfortable for me. They claim to ship them with varying sizes and

shapes to fit different faces. The power connector has a great little design as well, that interconnects using internal pin-type power linkages with an external twist lock. There is also a magnetic solution for some, but not all optical adjustments people with differences in vision may need. The onboarding experience features an automatic eye relief calibration matching the lenses to the center of your eyes. No manual wheels adjusting that here. The mainframe and glass piece look fine, though it's worth mentioning that they are very substantial in size, not heavy per se, but definitely present. If you have experience with VR at all, then you know that the two big barriers most people hit are either latency-driven nausea or the isolation

that long sessions wearing something over your eyes can deliver. Apple has mitigated both of these head-on. The R1 chip that sits alongside the M2 chip has a system-wide polling rate of 12 milliseconds, and I notice no jutter or frame drops. There was a slight motion blur effect used in the pass-through mode, but it wasn't distracting. The windows themselves render crisply and moved around snappily. Of course, Apple was able to mitigate those issues due to a lot of completely new and original hardware. Everywhere you look, there's a new idea, a new technology, or a new implementation. All of that new comes at a price. $3,500 is on the high end of expectations and firmly places the device in the power user category for early adopters.

The eye tracking and gesture control is near perfect. The hand gestures are picked up anywhere around the headset. That includes on your lap or low and away, resting on a chair or couch. Many other hand tracking interfaces force you to keep your hands up in front of you, which is tiring. Apple has high-resolution cameras dedicated to the bottom of the device just to keep track of your hands. Similarly, an eye-tracking array inside means that, after calibration, nearly everything you look at is precisely highlighted. A simple, low-effort tap of your fingers, and boom, it works. Pass-through is a major key.

Having a real-time 4K view of the world around you, that includes any humans in your personal space, is so important for long-session VR or AR-WARE. There is a deep animal brain thing in most humans that makes us really, really uncomfortable if we can't see our surroundings for a length of time. Eliminating that worry by passing through an image should improve the chance of long-use times. There's also a clever breakthrough mechanism that automatically passes a person who comes near you through your content, alerting you to the fact that they're approaching. The eyes on the outside, which change appearance depending on what you're doing,

also provide a nice context clue for those outside. The resolution means that text is actually readable. Apple's positioning of this as a fool-on computing device only makes sense if you can actually read text in it. All of the previous iterations of virtual desktop setups have relied on panels and lenses that present too blurry a view to reliably read fine text at length. In many cases, it literally hurt to do so. Not with the Apple Vision Pro. Text is super crisp and legible at all sizes and at far distances within your space. The personas play. I was highly doubtful that Apple could pull off a workable digital avatar based off of just a scan of your face using the Vision Pro headset itself. Doubts crushed. I'd say that if you're measuring the digital version of you that it creates to be your avatar

and FaceTime calls in other areas, it has a solid set of toes on the other side of the uncanny valley. It's not totally perfect, but they got skin tension and muscle work right. The expressions they have you make are used to interpolate out a full range of facial contortions using machine learning models, and the brief interactions I had with a live person on a call, and it was live, I checked by asking off-script stuff, did not feel creepy or odd. It worked. It's also crisp. I'm sort of stating this again, but really, it's crisp as hell, running right up to demos like the 3D dinosaur you got right down to the texture level and beyond. 3D movies are actually good in it. Jim Cameron probably had a moment where he saw Avatar Way of Water on the

Apple Vision Pro, this thing was absolutely born to make the 3D format sing, and it can display them pretty much right away. So there's going to be a decent library of shot on 3D movies that will bring new life to them all. The 3D photos and videos you can take with Apple Vision Pro directly also looks super great, but I wasn't able to test capturing any myself, so I don't know how that will feel yet. Awkward, hard to say. The setup is smooth and simple, a couple of minutes, and you're good to go. Very Apple, end quote. And then here's NELI's impressions from The Verge. Quote, the headset itself weighs a little less than a pound.

It's connected by a braided white power cable to a silver battery pack that offers about two hours of use. The cable detaches from the headset with a mechanical latch, but it's permanently connected to the battery pack. If you want to plug into the wall, you plug a USBC adapter into the battery pack. When you put on the headset, there's a quick automatic eye adjustment that's much quicker and more seamless than on something like the Quest Pro. There are no manual dials or sliders for eye settings at all. Apple wouldn't say anything specific about its field of view this long before launch, but I definitely saw black in my peripheral vision. The Vision Pro is not as totally immersive as the marketing videos would have you believe. The display itself is absolutely bonkers, a 4K display for each eye with pixels

just 23 microns in size. In the short time I tried it, it was totally workable for reading text in Safari, looking at photos and watching movies. It is easily the highest resolution VR display I have ever seen. There was some green and purple fringing around the edges of the lenses, but I can't say for certain if that was down to the quick fitment or early demo nature of the device or something else entirely. We'll have to see when it actually ships. The video pass-through was similarly impressive. It appeared with zero latency and was sharp, crisp, and clear. I happily talked to others, walked around the room, and even took notes on my phone while wearing the headset, something I would never be able to do with something like the

MetaQuest Pro. That said, it still video passed through. I could see pretty intense compression at times and loss of detail when people's faces moved into shadows. I could see the IR light on the front of my iPhone, futilely blinking as it attempted to unlock with face ID to no avail, and the display was dimmer than the room itself, so when I took the headset off, my eyes had to adjust to how much brighter the room was in reality. Similarly, Apple's ability to do mixed reality is seriously impressive. At one point in a full VR avatar demo, I raised my hands to gesture at something, and the headset automatically detected my hands and overlaid them on the screen. Then noticed I was talking to someone and had them appear as well. Reader, I gasped. Apple's also gotten a lot farther with

eye tracking and gesture control. Eye tracking was pretty solid, and those IRLuminators and side cameras mean you can tap your thumb and index finger together to select things while they're down in your lap or at your sides. You don't need to be pointing at anything. It's pretty cool. Apple has clearly solved a bunch of big hardware interaction problems with VR headsets, mostly by out-engineering and out-spending everyone else that's tried. But it has emphatically not really answered the question of what these things are really for yet. The main interface is very much a grid of icons, and most of the demos were basically projections of a giant screen with very familiar apps on them, Safari, photos, movies, the freeform collaboration app, FaceTime video calls. There was one demo with 3D dinosaurs where a

butterfly landed on my outstretched hand, but that was as much augmented reality as I really experienced. Was all this made better by the wildly superior Vision Pro hardware without question? But was it made more compelling? I don't know. And I'm not sure I can know with just a short time wearing the headset. I do know that wearing this thing felt oddly lonely. How do you watch a movie with other people in a Vision Pro? What if you want to collaborate with people in the room with you and people on FaceTime? What does it mean that Apple wants you to wear a headset at your child's birthday party? There are just more questions than answers here, and some of those questions get at the very nature of what it means for our lives to be literally mediated by screens. I also know that Apple

still has a long list of things it wants to refine between now and next year when the Vision Pro ships. That's part of the reason it's being announced at WWDC, to let developers react to it, figure out what kinds of apps they might build, and get started on them. But that's the same promise we've been hearing about VR headsets for years now from meta and others. Apple can clearly outpace everyone in the industry when it comes to hardware, especially when cost is apparently no object, but the most perfect headset demo reel of all time is still just a headset demo reel. Whether Apple's famed developer community can generate a killer app for the Vision Pro is still up in the air, end quote. So I need to research the warranty on this thing, like, for the purposes of the show,

could I buy one, try it out for a week, and as long as I return it within 30 days or something, get my money back, because $3,500 is a bit far for me, even for, you know, doing this job. But Apple, if you wanted to send a review unit, I'd be more than happy to record, like, an hour at least episode just based on my impressions of using the thing. You know where to find me.