Tech Brew Ride Home - Tue. 05/30 – Nvidia Joins The $1T Club

Episode Date: May 30, 2023

As Nvidia joins the rarified trillion dollar market cap club, their recent announcements highlight why they’re truly the center of the current AI moment. ARM transitions to a comprehensive 64-bit platform. Hands-on with Meta's Quest 3 prototype. And those weird new CAPTCHA’s you’ve been seeing? Yes, AI is responsible. Sponsors: SwissAmerica.com/ride Bloomberg.com/careers Links: How Nvidia created the chip powering the generative AI boom (Financial Times) World’s Most Valuable Chipmaker Nvidia Unveils More AI Products After $184 Billion Rally (Bloomberg) Video of the NPC character demo Arm Unveils 2023 Mobile CPU Core Designs (AnandTech) A First Look at the Headset That Could Be Apple’s Biggest Competition (Bloomberg) Captcha Is Asking Users to Identify Objects That Don't Exist (Motherboard) 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 Tech meme right home for Tuesday, May 30th, 2023. I'm Brian McCullough today. As NVIDIA joins the Rarified Trillion Dollar Market Cap Club, their recent announcements highlight why they're truly at the center of the current AI moment. Arm transitions to a comprehensive 64-bit platform, hands-on with meta's Quest 3 prototype, and those weird new captures you've been seeing.

Yes, AI is responsible. Here's what you miss today in the world of tech. In the years of doing this show, there are some companies I never really thought would. matter very much to our conversation that have nonetheless subsequently become very important for us to cover. I'm thinking primarily of Disney, what with the streaming wars. But while Nvidia has always been an important tech company, it's impossible to describe the way that Nvidia has become central to what we've been talking about in this AI moment. It's maybe one of the most important tech companies full stop that we're covering these days. Invidia's stock opened this morning at a one trillion-dollar market cap for the first time, thereby joining Apple, Microsoft, Google, and Amazon

in that rarefied club after raising its Q2 revenue forecasts due to AI demand. InVIDIA shares are up 166.5% year-to-date, and that was before the opening this morning. Why? Here's the Financial Times. In 2022, U.S. chipmaker, Nvidia released the H-100, one of the most powerful processors it had ever built, and one of its most expensive, costing about $40,000 each. The launch seemed badly timed, just as businesses sought to cut spending amid rampant inflation. Then in November, chat GPT was launched. We went from a pretty tough year last year to an overnight

turnaround, said Jensen Huang. Invidia's chief executive, OpenAI's hit chatbot was an aha moment, he said. It created instant demand. ChatGPT's sudden popularity has triggered an arms race among the world's leading tech companies and startups that are rushing to obtain the H-100, which Wong describes as the world's first computer chip designed for generative AI, artificial intelligence systems that can quickly create human-like text, images, and content. The value of having the right product at the right time became apparent this week. Nvidia announced on Wednesday that its sales for the three months ending in July would be $11 billion, more than 50 percent ahead of Wall Street's previous estimates,

driven by a revival in data center spending by big tech and demand for its AI chips. investors' response to the forecast added $184 billion in a single day to NVIDIA's market capitalization on Thursday, taking what was already the world's most valuable chip company close to a trillion dollar valuation, end quote. Yes, saying you're going to grow revenue, revenue, not profits revenue 50% in a single quarter is a hell of a drug. Nvidia is central to this AI moment, which is why it's worth noting their big announces over the weekend. And NVIDIA announced various AI products and updates, including the DGXGH-200 supercomputer platform, also gaming features, data center networking tools, and a robotics platform.

Quoting Bloomberg. The wide-ranging lineup includes a new robotics design, gaming capabilities, advertising services, and a networking technology. Perhaps most central to his ambition CEO Jensen Huang took the wraps off an AI supercomputer platform called DGXGH-200 that will help tech companies create successors to chat GPT. Microsoft, Meta, and Alphabets, Google are expected to be among the first users. It's too much, Wong 60 said near the end of his keynote at the Computex show, I know it's too much.

The flurry of announcements underscores Nvidia's shift from a maker of computer graphics chips to a company at the center of the AI boom. Last week, Wang gave a stunning sales forecast for the current quarter, almost $4 billion above analyst estimates, fueled by demand for data center chips that handle AI tasks. that sent the stock to a record high and put NVIDIA on the brink of a trillion-dollar valuation, a first for the chip industry. In the presentation Monday, Wang argued the traditional architecture of the tech industry

is no longer improving fast enough to keep up with complex computing tasks. To realize the full potential of AI, customers are increasingly turning to accelerated computing and graphics processing units or GPUs like those made by NVIDIA. We have reached the tipping point of a new computing era, Wang said, as he paced the stage in a trademark leather jacket. Wang showed how Nvidia is teaming up with WPP PLC to use AI and the Metaverse to lower the cost of producing advertising. It's releasing a networking offering that's designed to turbocharge the speed of information within data centers, and the company is even looking to change how people interact with video games.

A service called Nvidia Ace for games will use AI to enliven background characters and give them more personality. Wong also unveiled a new robotics platform that he said, is aimed at helping Nvidia expand to industries beyond tech. It's a reference design that will help other companies build their own robots for use in a range of activities. In heavy industry, for example, he sees opportunities in using robots in factories and warehouses. The DGX computer is another attempt to keep data center operators hooked on Nvidia's products. Microsoft, Google, and their peers are all racing to develop similar services to OpenAI's chat GPT chatbot, and that requires plenty of computing horsepower. To satisfy this appetite,

that Nvidia is both offering equipment for data centers and building its own supercomputers that customers can use. That includes two new supercomputers in Taiwan, the company said. One of the biggest AI bottlenecks is the speed at which data moves within data centers. Invidio's Spectrum X, a networking system that uses technology acquired in the 2020 purchase of Melanox technologies will address that issue, and the company is building a data center in Israel to demonstrate how effective it is. The WPP partnership, meanwhile, will streamline the creation of advertising content. The UK advertising Titan will use Nvidia's Omniverse technology to create virtual twins of products that can be manipulated to customize ads and reduce the need

for costly reshoots. InVidia's original business was selling graphics cards to gamers and it's returning to that world with the ace offering. The service will address the problem of NPCs or non-player characters, the background figures that populate video games. MPCs typically give repetitive responses with scripted dialogue, and that limited range has made them the subject of ridicule and memes, and even the Ryan Reynolds movie Free Guy. Nvidia Ace will listen to what the gamer says to a character, convert into text, and then dump that into a generative AI program to create a more natural off-the-cuff response. The Santa Clara, California-based company is currently testing the service and will add guardrails

to ensure that responses aren't inappropriate or offensive, end quote. I believe we said this NPC revolution was coming to Gaming Tootsweet, and here it is. By the way, if you check the show notes, I think it'll be the, third link, I have a link to a video of the demo of this NPC gaming interaction in action. Bit of a chip-heavy newsday, Arm has announced its transition to a comprehensive 64-bit platform with new mobile CPU core designs, Cortex4, Cortex870, and Cortex A-520 built on its Arm version 9.2 Architecture, coding a Nantec. For their 2023 IP, Arm is rolling out a wave of minor micro-architectural.

improvements through its cortex line of cores with subtle changes designed to push efficiency and performance throughout, all while moving entirely to the A-Arch 64-64-bit instruction set. The latest CPU designs from Arm are also designed to align with the ongoing industry-wide drive towards improved security, and while these features aren't strictly end-user-facing, it does underscore how Arm's generational improvements are more than just performance and power efficiency. In addition to refining its CPU cores, Arm has undertaken a comprehensive upgrade of its dynamic shared unit core complex block with the DSU 120. Although the modifications introduced are subtle, they hold substantial significance in terms of improving the efficiency of the fabric,

holding arm CPU cores together, along with extending arms reach even further in terms of performance scalability with support for up to 14 CPU cores in a single block, a move designed to make Cortex A slash X even better suited for laptops. Expanding on the enhancements, introduced in the ARM v9.1 architecture last year. Arm is progressing through its scheduled development cycle with the latest Arm 9.2 architecture. The primary objective of this cycle is to eliminate support for 32-bit applications and transition to a comprehensive 64-bit platform. Underpinning this transition is Arm's strategic framework, total compute solutions, or TCS, which revolves around three core principles, compute performance, security, and developer

access. This approach forms the foundation for arms methodology, and guides its efforts in delivering optimal performance, robust security measures, and streamlined developer capabilities. Arms focus on phasing out the 32-bit instruction set has been one it has been working towards for several years. For their latest TCS 23, they have finally created a fully 64-bit cluster to capitalize on the benefit of a complete 64-bit mobile ecosystem, excising a arch 32-32-bit instruction support entirely. So whether it's a big, mid, or little core, for Arms latest generation of IP there is only AARCH 64. Arms latest dynamic shared unit DSU-120 offers support for up to 14 CPU cores and a cluster,

which opens the door to a significant number of different CPU core combinations. We'll see what system-on-a-chip vendors have opted for later this year, end quote. Mark Gurman got a hands-on with Meta's Quest 3 VR headset prototype, although given the features he saw, maybe we should start calling it XR, He said that the Quest 3 is lighter, thinner, faster, with better video pass-through than the Quest 2, but the VR displays have similar clarity. So how does this position the Quest if Apple's headset comes out as a tangible leap forward, even if it's an expensive one?

Quoting Bloomberg. The device, codenamed Eureka, feels far lighter and thinner than the existing Quest 2 from 2020. The strap to place it on your head seems a bit stronger, and it uses fabric on the sides instead of the Quest 2's plastic. The front of the device is new. Instead of a bland-looking gray face, it has three vertical pill-shaped sensor areas across the front. The left and right pills each include one color video pass-through camera, the part that lets you see the world around you, as well as a standard camera. That means it has two color cameras versus only non-color cameras on the Quest 2. The middle pill includes a depth sensor, a first for the quest. The front

lower sides of the headset each include one tracking camera, while the bottom has a volume rocker and a wheel to adjust the headset's IPD, the distance between your pupils. That's a big improvement over the Quest 2, which requires you to take off the headset and manually move the VR displays inside of the device. The power button and the USBC port remain on the side of the headset. The actual clarity and VR displays within the Quest 3 feel similar to those in the Quest 2, despite the resolution being rumored to be slightly higher. But there are two areas where I saw major improvements.

Video Pass-Thru for Mixed Reality and the device's speedier performance. Video pass-through is the heart of mixed reality. It relies on external cameras to let headsetware see a live video feed of the real world, creating an augmented reality effect without the use of clear lenses. While I don't believe the Quest 3's video pass-through performance will come close to that of Apple's device, which will have about a dozen cameras, it is a night and day improvement over the Quest 2. Due to the dual RGB color cameras, video pass-through on the Quest 3 presented colors more accurately and offered an almost lifelike rendering of the real world. was even able to use my phone while wearing the headset, something that often feels impossible on a Quest 2. Overall, I came away impressed with the mixed reality focus of the Quest 3, the much improved video pass-through capabilities, the faster performance, and the large content library. Assuming the device costs about $500, it would be about a fifth as much as the Apple headset

while being more than a fifth as compelling. With that in mind, I think the Quest 3 could make some noise this holiday season, especially as XR is top of mind for consumers, end quote. Finally today, I look at something that you might have started to encounter in various corners of the interwebs. It's called H-capsia, a recapture alternative used by Discord and others that makes use of AI-generated CAPTCHAs to train machine learning systems and generative adversarial networks, quoting motherboard. People trying to use Discord are being asked to identify an object that does not exist. The object in question is a yoko, which appears to be a kind of mix between a snail and a yo-yo. Multiple people have reported seeing a prompt to identify a Yoko when asked to solve a simple

capture prompt while trying to use Discord. Discord's captias are run by a company called H-capsha. The technology that generates these prompts is proprietary to our third-party partner, and Discord does not directly determine what is presented to users, Discord told motherboard. H-capsia bills itself as a privacy-focused alternative to the ubiquitous recapture. According to a blog from 2018 about how it works, H-capscha's prompts are generated by clients looking for, quote, high-quality human- annotations for their machine learning needs. H-CAPHA makes money from both clients like Discord,

who purchase pro and enterprise plans to run CAPTCHA services and from clients who generated the prompts. H-CAPTCHA is using the CAPTCHAs to help train machine learning systems and generative adversarial networks. This isn't the first time people have noticed weird AI images popping up in H-CAPHA services, and it won't be the last. Two months ago, a Redditor noticed Discord was asking it to distinguish AI-generated soccer players amid a group of pictures of people playing hockey and golf. In March, another redditor complained that the Discord captions had become unintelligible. Like, choose the robot where none of the images look human to begin with. Funner than usual said on the Discord subreddit, then I never get the email to validate my account even after jumping through hoops. Seriously, blocking bots is fine. Trying to train AI with

SC rendered capture images is another thing entirely, end quote. The issue with H-capture's strange AI-generated prompts highlights two issues with machine learning systems. The first is that the AI systems require an enormous amount of human input to not be terrible. Typically, image labeling is outsourced to foreign workers who do it for pennies on the dollar. The other is the issue of data drift. The longer these machine learning systems run, the more input they require. Inevitably, they begin to use data they've generated to train themselves. Systems that train on themselves long enough become AI Hapsburgs churning out requests to identify incomprehensible objects like Yocos. End quote.

Quick Zelda report for you from the weekend. We've got two big bosses down, the Gerudo one, and the Birdie Place one. And this weekend, I fully got the power of all those YouTube videos we've been seeing of gameplay. I decided it was time to clean out that Pirates Nest in Laurelund Village. It's a side mission where you have to battle like 30 enemies all at once. So what I did was I crafted an actual battle tank using Zonai stuff. Armor on the sides, big wheels to run over the terrain. and of course lasers.

It took me a good half an hour to piece it all together. Then I rolled my war rig into the village and took out 25 of the bad guys without even taking one hit of damage myself. I now have a whole slew of robot tools that I use when I head to the depths to do some zonite harvesting. Now, instead of battling a slew of bocoblins and risking my precious weapons, not to mention my health,

I just spin up a quick drone with lasers. Roll that up to the bad guys, run off to a safe distance, and it clears out the whole nest. Then I just waltz back in to clean up any survivors and harvest the loot. If all of that sounded like gibberish to you, look, this is an open-world game

where you can solve any problem any way you dream it up. And if that dream is to build a war rig, you can do it. If you don't think that's cool, I don't know what to tell you. It's great fun for someone like me who sucks at gaming and can never get the controls right

to do proper combat. And building this stuff never feels unfair or like you're nerfing the game because if I build a proper tank and it works, I feel like I earned my victory. Tears of the kingdom. Hell of a drug. Talk to you tomorrow.