OpenAI and SpaceX Are Racing to IPO First  🔗 **Source:** [OpenAI and SpaceX Are Racing to IPO First](https://share.snipd.com/episode/df80411c-6fa9-4d9b-9482-af8b5299b1ff) ## Highlights - **Prepare For Short Term Layoffs During AI Restructuring** - Expect short-term layoffs as companies restructure to integrate AI but plan for growth as AI investments scale over years. - Jaeden Schafer points to Intuit cutting 17% while reporting strong earnings, attributing the move to AI-focused simplification. Transcript: Jaeden Schafer We have some reports that Intuit is cutting their global workforce by 17%. A lot of this is due to what they're calling AI restructuring. 17% of their global workforce is about 3,000 people, and that's a bunch of office closures in Reno, Nevada and Woodland Hills, California. Their CEO, who is Susan Gudarzi, told all of the staff that the strategic reset and streamlined structure is going to help the TurboTax parent deliver better products. What a lot of people are pointing out, and I have been saying for my last two weeks of podcast episodes because I seem to cover job closures every single day almost, or layoffs, what people Are kind of pointing out here is that Intuit actually just had a stronger than expected third quarter earnings. They had 10% year-over growth, which lifted their forecast for the fiscal year. So financially, the company's actually doing really well. The reason why they're doing all this apparently is to focus on simplifying their operations and focus on AI integration. Now, to be fair, I've actually used a bunch of tools. I actually run tools directly onto my like Claude Cowork. I will tell it to do things inside of my QuickBooks account. And even inside of TurboTax, I had Claude Cowork do a ton of work helping me with TurboTax. I actually spoke with my accountant recently and told him not to bother doing all of the classification of all of our expenses for a few of my businesses because I'll just run Claude Cowork On that. It does a great job of going and deciding what category different expenses go into. If it doesn't know what a company does, it can go and research the company and then it will be able to figure it out. It does it really quickly. And I don't have to pay a person to manually do that. Now, the reason I bring that up is because I do think that there's so much happening with a company like Intuit where there are shifts going on, right? ([Time 0:01:11](https://share.snipd.com/snip/0e351c92-9b8e-4a35-90a3-2dd8b0dd21af)) - **Reallocate Staff Into AI Roles Not Just Lay Them Off** - Reallocate rather than only cut workers when pivoting to AI; retrain staff into AI-related roles to retain institutional knowledge. - Jaeden Schafer praises Meta reallocating 7,000 employees to AI roles while laying off 8,000. Transcript: Jaeden Schafer What I do appreciate from Meta is that there's an additional 7,000 employees that are going to be reallocated to AI related roles. This is what I would love to see from most startups that are saying, hey, look, we're doing layoffs to like, invest in AI or focus on AI. We've even seen companies like General Motors that is, you know, laying off 800 people in IT or 600 people in IT, but simultaneously they have open job roles for IT people that are quote Unquote AI native or something like that, right? So it's like they're hiring people that are AI skilled where at the end of the day, I love what Meta is doing here and they're just reallocating workers, like 7,000 workers to AI related Roles. Like there's no such thing as a person that, I mean, there is such thing as a person that, you know, their job is all about AI, but I think anyone can learn how to use AI in their role and can Learn how to use AI to accentuate their role. So I like this direction a lot more where we're just moving. If you really are having kind of a realignment of priorities and what you think is important to AI related roles, move some of your workers over there because I think that anyone can learn It. And this is, you know, a phenomenal time to do that. You also save yourself a lot of, you know, a company would save itself a lot of money. If it's not just laying off workers to save, you know, a literal dollar amount, which I feel like is basically what a lot of these companies are doing. They just don't want to say it out loud. But if that's not the case, then just move them there. You're not gonna have to pay severance and you're not gonna have to pay all of the money to recruiters to hire people. ([Time 0:04:03](https://share.snipd.com/snip/963ff20b-e629-45e1-a0e1-a89d6a537d84)) - **OpenAI and SpaceX Racing IPO Timing** - OpenAI and SpaceX appear to be in a timing race to file IPO paperwork, potentially compressing market windows and investor attention. - Jaeden Schafer links Elon Musk's lawsuit strategy to buying a cleaner runway for SpaceX's NASDAQ debut by delaying competing AI IPO noise. Transcript: Jaeden Schafer All right, let's get into what's going on with the OpenAI IPO. We know that they've been working with Goldman Sachs and Morgan Stanley. I've reported on this a couple of times. They're going to be filing this paperwork within the next week or weeks. This is going to put them in direct and basically a race with SpaceX because SpaceX now owns Elon Musk's XAI. And they are also trying to get this filed as soon as possible. I actually got a really great comment when I was posting about Elon Musk and SpaceX and Elon Musk's lawsuit of OpenAI and how he quote unquote like lost the lawsuit or whatever. I got a really great comment from someone named Raphael Strafanello Gisliani over on LinkedIn. And so I appreciate this comment, but he said talking about, you know, the quote unquote Musk losing. He said, the verdict is irrelevant. Musk already won. His goal was never to win the case. It was to delay opening eyes for profit conversion long enough to clear the runway for SpaceX's June IPO with NASDAQ's new fast tracked rule. SpaceX enters the NASDAQ 100 after 15 trading days, forcing index funds to buy regardless of valuation. That force passive inflow is worth far more if it isn't competing with another trillion dollar AI IPO in the same window. He said the lawsuit achieved its purpose the moment it was filed, the procedural knockout changes nothing. So I actually thought that was a really interesting perspective I hadn't thought of before as far as trying to get SpaceX and their IPO to be as successful as possible. And that seems to be perhaps the strategy that Elon Musk was playing at when he did the lawsuit against OpenAI, because at the end of the day, while some people would have defended his Lawsuit on moral grounds, it was outside of the statute of limitations. That's ultimately why it was struck down. And so it's just, you know, if he really cared about it that much, or if he really wanted that outcome, he should have filed the lawsuit much sooner when OpenAI made these kind of intentions Clear. And then potentially he would have been successful in doing that. But because he didn't, they're just kind of throwing it out. So bringing it back to OpenAI's IPO that they're looking at, apparently they have just crossed $3.8 billion in annualized revenue. They did that in Q4 of last year. Most of that was API sales and chat GPT subscriptions. They didn't say how profitable they were. So we'll find that out very soon. The SpaceX IPO filing are expected as soon as today. So we have a lot of AI news, a lot of stuff happening today and around this time. ([Time 0:05:55](https://share.snipd.com/snip/0f09f682-b81a-44f5-96fb-96814c0df048)) - **OpenAI Reported $3.8B Annualized Revenue** - OpenAI reached about $3.8 billion annualized revenue in Q4, mostly from API and ChatGPT subscriptions. - Jaeden Schafer notes revenue was disclosed but profitability details remain unrevealed ahead of the IPO filing. Transcript: Jaeden Schafer In doing that. But because he didn't, they're just kind of throwing it out. So bringing it back to OpenAI's IPO that they're looking at, apparently they have just crossed $3.8 billion in annualized revenue. They did that in Q4 of last year. Most of that was API sales and chat GPT subscriptions. They didn't say how profitable they were. So we'll find that out very soon. ([Time 0:07:53](https://share.snipd.com/snip/fc41740e-9ce3-4c23-aabe-28e18b5c3742)) - **Google AI Mode Reaches One Billion Users** - Google AI mode has hit 1 billion monthly users with usage doubling each quarter since launch. - Jaeden Schafer highlights Google overhauling search with redesigned search box, AI overviews, and agentic features replacing traditional links. Transcript: Jaeden Schafer And one of the big announcements we've also got today is that Google AI mode has just hit 1 billion monthly active users. Their usage has doubled quarterly, which is really impressive. As of this month, every quarter since they launched, their usage has actually doubled. So apparently right now they are overhauling search. I've seen our friend of the podcast, Logan Kilpatrick, posting about it on LinkedIn, showing what they've done. He said this is the most, you know, the biggest change they've made to search in 25 years. ([Time 0:08:23](https://share.snipd.com/snip/13cf6ed2-79b8-470e-b3ad-8818fd45808c)) - **AI Overviews Will Reshape Search Clicks** - Google will surface an AI overview box that can push organic blue links further down, fundamentally changing click patterns. - Jaeden Schafer describes this as the biggest search change in 25 years, adding chat-like overviews and dashboards instead of classic results. Transcript: Jaeden Schafer This is a pretty big and controversial move. Essentially, what's going to be happening is you get the AI overview at the top and below that used to have the blue link. So it was kind of like they were there, you know, keeping the legacy Google results and they were had that. But now they're actually going to have an AI overview box that is going to sit right below the AI overview like a chat box which is basically going to make it so you can't see the top organic Results as much it's going to be steering people away from these kind of traditional web listings that we are so used to so things are really going to be shifting quite a lot we have some News that stability AI has just released Stable Audio 3.0. And what's interesting here is we had a really cool music release a while back from Eleven Labs. And when it came out, I was like, guys, it's basically nothing because Eleven Labs let out, you know, let you create music, but it could only be 30 seconds long. I'm like, this is basically useless. Stability has let, has come out and you can actually create songs that are up to six minutes and 20 seconds long. That's more than double the three minute limit that they had in last year's version and much higher than something like 11 labs. ([Time 0:09:10](https://share.snipd.com/snip/b44664de-f5eb-4af8-b6ef-419791a82132)) - **Stable Audio 3.0 Unlocks Longer Licensed Music** - Stability AI released Stable Audio 3.0 enabling songs up to 6 minutes 20 seconds and licensed training data from major labels. - Jaeden Schafer highlights multiple model sizes, open weights for small users under $1M revenue, and enterprise licenses above that. Transcript: Jaeden Schafer And what's interesting here is we had a really cool music release a while back from Eleven Labs. And when it came out, I was like, guys, it's basically nothing because Eleven Labs let out, you know, let you create music, but it could only be 30 seconds long. I'm like, this is basically useless. Stability has let, has come out and you can actually create songs that are up to six minutes and 20 seconds long. That's more than double the three minute limit that they had in last year's version and much higher than something like 11 labs. So apparently they've trained all of this on licensed data from Warner music group and universal music group. They basically sidestepped the entire lawsuit, uh, you know, fiasco that is currently hammering. I think Suno and UDO are both getting hit with this. And so they got around that by just going directly and getting the licensing. So big kudos to stability. They have four models released at 459 million, 1.4 billion and 2.7 billion parameters. The small medium versions are shipping with open weights for commercial use under a million dollars in revenue. So if you have under a million dollars in revenue, you could be using this with open weights for commercial use, which is really impressive. The large model is available only on their paid API or self-hosting. They have enterprise licenses that are going to be required for companies that are making above a million dollars in annual revenue. ([Time 0:09:46](https://share.snipd.com/snip/8e9ea655-d0f6-4beb-9c1e-98773243f42e)) - **Iris Go Learns By Watching One Demonstration** - Iris Go builds a desktop agent that watches one demonstration and automates the workflow, running much processing on-device for privacy. - Jaeden Schafer shares his wife's use case rebuilding an app with Claude Co-work and wanting a tool that watches and automates repetitive UI tasks. Transcript: Jaeden Schafer Okay, sounds like a big mouthful, but this is backed by Andrew Nagy's AI fund, and they're going to build a desktop agent that learns workflows by watching you perform a task once, and Then it's going to automate it forever, and you don't have to write a prompt. This is actually something my wife was telling me. She's vibe coding. She's completely redoing the first app we ever created. It's called SelfPause. It's an AI affirmation recorder. You can record yourself saying affirmations, and you put all like these sounds in the background, like meditation sounds. You can make like a mix and listen to them. Some people like to listen to their, you know, personally recorded affirmations in the morning or listen to them all night. Anyways, this is the first product we ever made out of college. We grew it to about 150,000 users. And then I kind of have been working on AI box and this podcast for the last three years and haven't really put any time or attention on that, uh, startup, but it continues to chug along. And, you know, I think every month we get over 300,000 listens to audio on that, um, on that platform, on that, on that app. And in any case, my wife has been recently, um, using Claude code and Claude Cowork to completely, basically redo the whole thing, add all the features she ever wanted, build it native And Swift. So it's like a native iOS app because before we had it on React, which is kind of both. Anyways, she's rebuilding the whole thing. It took us years to make it and we spent hundreds of thousands of dollars and now she's, you know, being able to recreate it in a couple weeks, vibe coding. She did study computer science. So she, you know, she's pretty good at this kind of stuff, but still it's phenomenal. And just recently she was telling me, she was like, man, I wish I could just do, do this task and have cowork, watch me do it and then do it for me. So I don't have to tell it to do X, Y, Z, like a lot of the things I think she was recently like redesigning all of the eight icons inside of the app and she was using ChatGPT to help her redesign All the icons, but she got to like crap out the backgrounds and put them on a black box and whatever. She has a whole workflow, right? She's a hundred icons and she was getting co-work to do it, but she had to do these really descriptive prompts. Anyways, this is exactly what Iris Go is going to solve. And I love this. The co-founder is Jeffrey Lai. He built the Chinese version of Siri at Apple. And then he basically demonstrated the system ordering a coffee by recording one transaction. And then it repeated it autonomously. I think a lot of processing for this is going to be happening on device for privacy, which is amazing. The cloud tasks are going to require the user to actually authorize it. There's going to be end-to encryption. There's going to be, I mean, basically I think this is a big differentiator against a lot of cloud heavy competitors. They shipped their beta macOS and Windows apps and they secured a pre-install agreement with Acer to distribute the agent on new devices. So they're not going to have to spend any money acquiring users with that particular deal. So all in all, I think this is a really cool tool. They already have some good distribution. They've raised some money. ([Time 0:11:40](https://share.snipd.com/snip/c91fd6ae-524a-481a-8b4f-5aae34d920c3)) - **Google Embeds Ads Inside AI Responses** - Google will embed AI-generated product descriptions and sponsored items inside conversational AI responses to monetize the expanded AI real estate. - Jaeden Schafer compares this to ChatGPT ads and notes paid tiers can remove ads for users who prefer that. Transcript: Jaeden Schafer Okay, Google has embedded AI generated product descriptions inside of the search ad. This is kind of one of the big things that came out of Google IO they've been talking about. I just wanted to cover it quick on the podcast, but basically Google's now using Gemini to write custom product descriptions for ads inside search results. They're going to be embedding sponsor items directly into the conversational AI mode responses. So I think right now the AI mode is taking over more of Google's screen. It's going to be kind of blocking out the traditional search results. And because of that, they still need to make money. So they're putting ads directly inside of their responses, which to be fair is exactly what ChatGPT is doing. So, you know, Google is the ad ad king of the entire market. If they weren't putting ads in their AI model and ChatGPT was, I think that was pretty strange. And basically all that expanded ad real estate is going to, you know, get them more views beyond just whatever the traditional sidebar ads that are now appearing. I think this is good. I think the ads are going to be more specific. I know maybe some people will complain like, oh no, I don't want ads inside of my AI results. Well, of course you could go and pay $20 a month for Claude or ChatGPT or probably Gemini 2 and not get those ads. So that's one option. Or if you want to use all this stuff for free, you're just going to get ads. ([Time 0:15:26](https://share.snipd.com/snip/a8807dca-872f-40fe-9c17-046f12911afb))