Tech Brew Ride Home - Thu. 10/10 – OpenAI Won’t Be Profitable Until When?

Episode Date: October 10, 2024

Apple TV+ finally comes to Prime Video Channels. OpenAI says they’ve been disrupting AI election influence campaigns. But they won’t be profitable until when? How Google plans to beat the regulation rap. And a review of basically the highest end smartwatch you can buy. Sponsors: Notion.com/ride Links: Amazon Adds Apple TV+ to Its Channels Store for Streaming (Bloomberg) OpenAI says bad actors are using its platform to disrupt elections, but with little ‘viral engagement’ (CNBC) OpenAI Projections Imply Losses Tripling to $14 Billion in 2026 (The Information) Apple’s Dan Riccio, Key Executive in Both the Jobs and Cook Eras, to Retire (Bloomberg) Competition and the clock: how Google plans to deflect and delay a historic break-up threat (Financial Times) Garmin Fenix 8 review: only kind of smart (The Verge) RAD! 80s90s History on Apple Podcasts 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 Thursday, October 10th, 2024. I'm Brian McCullough today. Apple TV Plus finally comes to prime video channels. OpenAI says they've been disrupting AI election influence campaigns, but they won't be profitable until when? How Google plans to beat the regulation wrap and a review of basically the highest and smartwatch you can buy. Here's what you miss today in the world of tech. Amazon has announced that Apple TV Plus will soon be available as a $9.99 per month subscription. add-on and Prime Video channels. That's coming later this month in the U.S., joining more than 100 streaming services already there, quoting Bloomberg. What we offer channel partners is hundreds

of millions of subscribers around the world. Amazon said having this engagement around the world with Prime members, accessing Prime Video is a real great platform for other partners to be able to reach them, end quote. Amazon markets more than 100 streaming services via its channels store, enabling customers to watch shows from Max, Paramount Plus, and Hallmark. With it, Prime Video. Those services rely on Amazon to expand their audience. Apple TV Plus was one of the only major streaming services not available in the store, along with Netflix and Walt Disney's main properties. Those two are among the few with streaming services that have more viewers than Prime Video. The agreement expands an existing relationship between the companies. In 2017, Amazon started

offering a Prime Video app on the Apple TV set-top box, while Prime Video is integrated into the Apple TV app across the iPhone maker's device lineup. The two companies are also retail partners with Apple offering an official store via Amazon.com, end quote. OpenAI says it has disrupted more than 20 operations and deceptive networks so far this year that tried to use its models for disinformation purposes, including four that had election-related content, quoting CNBC. In a 54-page report published Wednesday, the chat GPT creator said that it's disrupted, quote, more than 20 operations and deceptive networks from around the world that attempted to use our models. The threats ranged from AI-generated website articles to social media posts by fake accounts. The company said its update on influence and

cyber operations was intended to provide a snapshot of what it's seeing and to identify, quote, an initial set of trends that we believe can inform debate on how AI fits into the broader threat landscape, end quote. Open AI's report lands less than a month before the U.S. presidential election, but beyond the U.S., it's a significant year for elections worldwide, with contests taking place that affect upward of 4 billion people in more than 40 countries. The rise of AI-generated content has led to serious election-related misinformation concerns, with the number of deepfakes that have been created increasing 900% year-over-year, according to data from clarity, a machine learning firm. OpenAI wrote in its report that election-related uses of AI, quote, ranged in complexity from simple requests for content generation, to complex multi-stage efforts to analyze and reply to social media posts.

The social media content related mostly to elections in the U.S. and Rwanda, and to a lesser extent, elections in India and the EU, Open AI said. In late August, an Iranian operation used Open AIs products to generate, quote, long-form articles and social media comments about the U.S. election as well as other topics, but the company said the majority of identified posts received few or no likes, shares, or comments. In July, the company banned ChatGPT accounts in Rwanda that were posting election-related comments on X, and in May, an Israeli company used ChatGPT to generate social media comments about elections in India. OpenAI wrote that it was able to address the case within less than 24 hours.

In June, OpenAI addressed a covert operation that used its products to generate comments about the European Parliament elections in France and politics in the U.S., Germany, Italy, and Poland. The company said that while most social media posted identified, received few likes, or shares, some real people did reply to the AI-generated posts. None of the election-related operations were able to attract viral engagement or build sustained audiences via the use of chatGBT and OpenAIs other tools the company wrote, end quote. Sticking with OpenAI for one more second, the information has seen internal documents from OpenAI with projections suggesting that the company won't turn a profit until 2029 when its revenue would hit $100 billion. Losses could hit $14 billion.

in 2026, up nearly 3x on this year. So if true, number one, they need to hit $100 billion in revenue just to turn a profit that does suggest that margins aren't great. But number two, that level of cash burn suggests they have a hell of a needle to thread. Quote, this estimate doesn't include stock compensation, which is one of Open AI's biggest expenses, although not when it pays in cash. Open AI is emphasizing to investors a metric of profitability that excludes some major expenses such as the billions it is spending annually on training its large language models. The documents show by that metric, Open AI projects it will turn profitable in 2026. The company last week completed a $6.6 billion funding round, valuing it at

$157 billion. The documents which include financial statements and forecasts contain a number of revelations about Open AI that may change perceptions of the company's financial prospects. Open AI appears to be burning far less cash than previously thought. The company burned through about $340 million in the first half of this year, leaving it with $1 billion in cash on the balance sheet before the fundraising effort. But the cash burn could accelerate sharply in the next couple of years, the documents suggest. There's a big gap between OpenAI's cash flow and its bottom line, reflecting standard accounting differences in how it treats its major expenses, such as stock compensation and compute credits in different financial statements. OpenAI calculated its net

loss to be $3 billion for the first half of the year. Open AI expects to spend spend more than $200 billion through the end of the decade, excluding stock compensation costs, between 60 to 80 percent of its spending each year would go toward either training or running the models. An analysis of the documents suggests that OpenAI projects its total losses between 2023 and 2028, excluding stock compensation, will be $44 billion. The same analysis suggests the company anticipates it will make $14 billion in profit on that basis in 2029. In the first half of this year, the company reported stock compensation of $1.5 billion, likely equivalent to its revenue in the period. The documents imply that Microsoft gets a cut of 20% of Open AI's revenues,

which is higher than previously thought. Open AI is projecting that computing costs for model training could rise sharply in the next couple of years to as high as $9.5 billion a year in 2026. That's in addition to the amoratized upfront costs of training for large language model research, which its financial documents report as research compute costs spread out over a number of years. That figure is also rising sharply to more than $5 billion in 2026 from a projected $1 billion this year. Some of OpenAI's computing costs aren't paid in cash. Microsoft advanced open AI computing credits as part of its $10 billion investment last year, and in the first half of this year, Open AI recorded about $500 million of data center lease expenses covered by Microsoft,

according to the documents and a person familiar with the matter. It's unclear how many computing credits OpenAI has left. It's likely, however, that OpenAI will have to spend more of its own cash if it increases compute spending as much as projected in the documents. OpenAI also has been discussing borrowing money to try to set up data centers more quickly than Microsoft can, the information reported this week. Still, OpenAI could scale back the compute spending. For instance, if upcoming models have more staying power than prior ones because rivals

aren't able to catch up quickly, or if the models are less costly to train because of future breakthroughs. That would reduce the demands on its cash resources, end quote. This makes me a tiny bit concerned about the Vision Pro. Sources say Dan Riescio, Apple's Vision Pro chief and former senior vice president of hardware engineering is leaving Apple in October. The Vision Pro team will now report to John Ternis, quoting Bloomberg. Mike Rockwell, Riccio's current lieutenant will continue to the Vision Products Group on a day-to-day basis, said the people who asked not to be identified because the changes aren't public. A spokesperson for the Cupertino-California-based company declined to comment. Recio's exit marks one of the most significant departures from Apple's management team,

though it's been in the making for some time. Three years ago, he switched from being a senior vice president to a vice president and left the 12-person executive team a step toward his eventual retirement. Recio spoke at an event at the Massachusetts Institute of Technology on Wednesday, telling students he is winding down his career at Apple. He said he wants to get more involved at the engineering program named for him at the university, including by potentially teaching or developing curriculum material. He also spoke about the birth of the original IMac and said the iPhone 10 was the product he was most proud to create. The executive spent 26 years at the company and was known at times for an abrasive style and controversial moves, but he also made key hires and

encouraged acquisitions that resulted in major new products. His tenure included overseas, seeing some of Apple's biggest bets, including development of the Vision Pro headset and the failed bid to make a car, end quote. The Financial Times takes a look at how Google plans to deflect and delay that U.S. breakup threat that we spoke about yesterday. A remedies ruling is likely in mid-20205, but legal timelines may let it put off any impact of that ruling for years. Quote, Google's initial response to the DOJ's proposals that competition is thriving in search ads and fierce in artificial intelligence would have been less convincing even two years ago before Open AI's launch of the breakthrough chat GPT chatbot. Spinning out its arguments through the appeals courts will

be crucial to Google's strategy as it looks to deflect or delay the effects of August landmark ruling by a federal judge that it maintained an illegal monopoly by paying billions of dollars to device makers, mobile carriers, and browser developers. The legal timelines involved in such a complex and high-stakes case are likely to allow Google to put off any impact on its business for years. It plans to appeal the liability decision when the judge rules on remedies, which is likely to be in mid-20205, and may then also contest the remedies themselves. Broadly, Google's argument focuses on what it describes as regulatory overreach following a case about the impact of its distribution agreements. Forcing it to divest assets or share data with competitors would, quote, go far beyond the specific legal issues in this case,

it said in a blog post on Tuesday. Requiring Google to split off its Chrome browser or Android operating system or other structural remedies would, quote, tilt the field at the precise moment that competition is thriving, the company said. With new search advertising competitors such as Amazon and TikTok emerging and widespread disruption to its core business from AI startups, including open AI and perplexity, Google can argue that it is facing the stiffest competition since Microsoft's being launched 15 years ago. On Tuesday, for example, Google pointed to an e-marketer forecast that its share of U.S. search advertising spending would fall below 50 percent next year. for the first time since the research group started tracking the market in 2008,

primarily due to rapid growth in Amazon's marketing business. The company is likely to appeal its antitrust cases all the way up to the U.S. Supreme Court. This is the start of a long process, it said in Tuesday's blog post. Yet Jason Kent, a big tech critic who leads the digital content next trade group of online publishers, said it was not a given that the Supreme Court would take up the case. He estimated that it could take two or three years for any remedies to be enforced. if the case proceeds through the courts, adding, quote,

the reality is Google is racking up legal losses. They have a difficult set of facts along with spoliation from purging evidence, and they may try to settle or proactively make moves to control the outcome, end quote. Finally today, a review. The Verge says that the Garmin Phoenix 8, the $1,000 smartwatch, has excellent battery life and fitness tracking and an OLED display, but it costs $350 more than the Phoenix 7 and lacks LTE, which apparently is key, quote. The big additions to the Phoenix 8 are a speaker and microphone. They mean you can now interact with

the watch in more ways than fitness tracking and training. You can take calls from the wrist, use an on-device voice assistant for setting timers and starting activities, and connect to your phones built-in assistant for more complex queries. If you can convince your loved ones to download the new Garmin Messenger app, replying to messages on the wrist, also has gotten easier. Theoretically, these are all good updates. In practice, it means the Phoenix 8 doesn't play to garment strengths. The bulk of the smart updates zero in on the addition of the microphone and speaker,

namely voice assistants and calling from the wrist. But there's a big factor holding back the Phoenix 8's smart features, the lack of LTE. Yes, you can make and receive calls with the Phoenix 8 so long as your phone is within Bluetooth range. Yes, you can talk to a voice assistant, but it's limited unless your phone is nearby. Ah, yes, you can reply to text, but you get my drift. If you get a cellular Apple Watch, Galaxy Watch, or Pixel Watch, you can leave your phone at home. That's not really an option with the Phoenix 8.

The friction is particularly noticeable with the on-device voice assistant. The assistant works pretty well, but it's mostly limited to commands that don't require an internet connection. You can start activities, set timers, and tweak settings. It can tell you the weather, provided it's loaded the day's weather from Wi-Fi or your phone via Bluetooth. But say you want to send a text message. This nameless assistant can't do that. Instead, you'll have to scroll to a different menu and launch your phone's assistant. It's good to have that option, but at that point, you're probably just going to whip out your phone anyway. Part of the problem is that Garmin has always been staunchly platform agnostic. I happen to think that's a good thing, but the price you pay is never having your watch work quite as seamlessly with your phone. For example, it's great that you can dictate voice memos now, but their usefulness is limited because there's no easy way to get them off your wrist and onto another device. That's unlike the Pixel Watch 3, where I can just record something, and boom, it's magically on my Pixel 8 Pro. Everyone I've ever met with a Phoenix watch has been a triathlete, ultra-marathoner, or someone who

casually qualifies for the Boston Marathon with a sub-three-hour marathon time. These folks pick something like the Phoenix because they need a tracker with Mondo Battery Life, stellar GPS accuracy, in-depth training programs, and maps for traversing trails. Those are the features that people would leave Apple or Samsung for, and they don't get a major upgrade with the Phoenix 8. To make matters worse, I almost choked when I saw the Phoenix 8's starting price was a whole $350 more than the Phoenix 7. That's not a $50 to $100 hike, where you grumble and then begrudgingly admit that inflation is a bitch. For $999, you could buy a laptop. Hell, the Apple Watch Ultra 2 is $800, while the Galaxy Watch Ultra is $650.

Those do come with LTE. And if you're determined to get a Garmin, there are literally dozens of cheaper options. Might I suggest the $450 forerunner 265 or a discounted Phoenix 7 Pro or Epic's pro. It's frustrating because Garmin is right to shore up its smart features. There's not much to improve on the fitness front, so making its watches more useful outside of training makes a lot of sense. It's also crystal clear that Garmin has some catching up to do, and progress isn't going to happen overnight. The Phoenix 8 might be on the right track, but it doesn't quite nail the price or execution, end quote. Okay, everybody. Rad. 80. 90s history is currently number 34 on the history chart. Please, if you haven't done so,

subscribe to the show, especially if you use Apple Podcasts, please follow the show on Apple Podcasts. And also, write a review. Give it five stars and write a quick sentence or two review. Even if you've already subscribed, go back and do that. It'll really help. We can do this, people. We can get to number one. The history category is not as competitive as some others, so we can make my dream come true of having a number one podcast. Link in the show notes to the Apple Podcast link itself. Again, please follow, rate and review. Thanks in advance.