AI Bubble? Layoffs Hit AI Companies

0:00Today on the AI Breakdown, we're looking at what it means that two AI companies have recently done layoffs. Before that, on the brief, a guild of authors has thrown in their lot with the SAG and WGA. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to breakdown.network for more information about our newsletter, Discord, and YouTube channel. Welcome back to the AI Breakdown Brief, all the AI headline news you need in five minutes or less. We begin today with an update in the AI data revolt. Over the last few weeks, we've seen the first coordinated labor movement that is at least in part about AI.

0:36It started with the Writers Guild of America, a union that represents TV, film, and entertainment writers. Now, there are many different parts of the dispute, but one of them is certainly a fear of being replaced by AI. The WGA had wanted to be able to negotiate about how AI could or couldn't be used in the writing and production of TV shows and films. And the entertainment industry on the other side basically initially at least said, nah. Right away, people understood that there was a significance for other types of jobs and employment, not just WGA writers. Polygon Peace on May 31st wrote, AI can't replace humans yet, but if the WGA writers don't win, it might not matter.

1:11The WGA strike is only the first battle in an oncoming labor war. Then, of course, last week, the Screen Actors Guild joined the fight. This is the first time that the Screen Actors Guild and the Writers Guild of America have been on strike at the same time since 1960. and again, while the issues are more numerous than just artificial intelligence, AI sits right at the center. Certainly AI has been at the center of a lot of the discourse around this particular strike, given that last week one of the statements made by studio executives said that SAG had rejected what they called a groundbreaking AI offer.

1:42Now SAG, meanwhile, said that that groundbreaking offer involved scanning background actors with AI and being able to use their likeness in perpetuity without permission or payment, effectively relegating all background actors to getting one day of pay for the rest of their lives. Perhaps expectedly, studio executives hit back and said that wasn't exactly the case. But wherever the truth lies, the point is that at that point you had SAG and the Writers Guild of America who were actively protesting against, in part, AI in the workplace. Today, a group of nearly 8 ,000 authors published an open letter asking Meta, OpenAI, Microsoft, and more to stop training their AI models on their works without permission or compensation.

2:20The signatories include a who's who of modern authors, including Nora Roberts, Margaret Atwood, and more. But in the same way that the SAG strike isn't really about the big actors and their destinies, and is instead about all the working actors and all the people in the industry who don't make bank from that work, the author's letter is also really about the average writer who has seen a significant decline in their income over the last decade. According to a report from the Authors Guild, the median income for a full-time writer last year was just$23 ,000. Between 2009 and 2019, writers' incomes declined by 42%.

2:51So of course, the signatories' letter fear that this will further marginalize authors, and basically they ask for what. Alexander Chee, who's written best-selling novels like Edinburgh and The Queen of the Night, wrote, there's no urgent need for AI to write a novel. The only people who might need that are the people who object to paying writers what they're worth. Mary Rassenberger, the CEO of the Authors Guild, said, it's not fair to use our stuff in your AI without permission or payment. Please start compensating us and talking to us. Now, for those of you asking, why not just sue? Rassenberger said, lawsuits are a tremendous amount of money.

3:21They take a really long time. Because of that, the Authors Guild is trying this public pressure approach first. Of course, some other authors, including Sarah Silverman, have recently filed class action lawsuits against those companies for training AI on their works. And it appears that there may be more legal action coming from the publishing world, as IEC's Barry Diller says that he's planning on taking legal action against these companies around the exact issues. using published proprietary works and the training of AI. In an appearance on CBS's Face the Nation on Sunday, Diller said, It will be long-term catastrophic if there is not a business model that allows people professionally to produce content.

3:54That would be, I think everybody agrees, catastrophic. Diller went on, saying, Of course we're open to commercial agreements. The only way you get to the point is to protect fair use, protect the copyright. Comparing it to previous battles, he said, It took 15 years to get back paywalls that protected publishers. I don't think that same thing is going to happen. Now, Diller also did say that he thought that generative AI was overhyped. When asked about whether he thought it presented a real threat to these Hollywood jobs, he said, I think the one to three year period, not much is going to happen.

4:23But post that, there are, of course, all these issues. Next up on the brief, moving to the policy side of things for a moment. The UN Security Council is planning to hold its first ever talks on AI risks this week. The rotating presidency of the UN Security Council is currently held by Britain, and British Foreign Secretary James Cleverly will chair the discussion on Tuesday around how global nations could come together to mitigate the potential problems of AI. One of the big discussions is whether at some point there will need to be a global watchdog body like the International Atomic Energy Agency, and in June, the UN Secretary General Antonio Guterres had indicated his support for such a proposal.

4:58Speaking of risks in AI and global power issues, Bloomberg is reporting that the Israel Defense Forces have started using artificial intelligence extensively in their military operations. Currently, that includes selecting targets for airstrikes, organizing wartime logistics, and while the IDF isn't commenting too extensively on this, there's certainly no stranger to the use of AI in combat. In 2021, during the 11-day conflict in Gaza, the IDF dubbed it the first, quote, AI war. In that conflict, the IDF used artificial intelligence to both identify rocket launch pads as well as to deploy drone swarms.

5:30As Israel's tensions with Iran grow, many are wondering if we're about to see the first wider use of AI in kinetic combat. Now, our main show today is going to be exploring to what extent AI is in a bubble, and even more than that, to what extent that bubble is starting to, if not pop, at least show signs of strain. And so a teaser related to that, Stability AI CEO Ahmad Mostak last week appeared on an analyst call with UBS and said two seemingly contradictory things. On the one hand, he identified artificial intelligence as a$1 trillion investment opportunity, but on the other hand, he said it would be the biggest bubble of all time.

6:04He said, I call it the dot AI bubble, and it hasn't even started yet. Now, of course, the reason that these two statements are not necessarily at the odds they might initially seem is that the fact that you have such a massive, massive investment opportunity, of course, means that not everyone's going to get it right about what to invest in. Capital will inevitably chase the opportunity into places where it ultimately doesn't take root. And so for as much money as will be made on the opportunity, obviously, a lot of money will be lost as well. Meanwhile, in fundraising, it seems that money continues to flow to big AI startups.

6:34Despite the fact that they raised$150 million just four months ago, Character AI is back in talks to raise more. But that news might be tempered by the fact that it appears that a number of AI startups have actually had to do a round of layoffs. In today's main AI Breakdown, we'll be looking at whether layoffs at Jasper and Mutiny actually reflect a larger trend, or whether they are more specific to those companies themselves. For now, that's going to do it for the AI Breakdown Brief. If you're enjoying it, go subscribe to the newsletter. it's at the AIbreakdown.beehive.com, and I'll be back soon with the main AI breakdown.

7:09Hey guys, before we dive into the main part of the episode, I want to share a little bit about today's sponsor, Supermanage. A truly great one-on-one should be about celebrating wins, solving problems, and deepening the connection between two human beings. But what if you miss those wins, never heard about those problems, and spent your whole meeting avoiding the hard stuff. That's where Supermanage comes in. Supermanage AI distills your public Slack channels into a one-on-one brief that highlights everything you need to know to jump right in. Because let's face it, you want your team to do the best work of their lives.

7:40And that starts with world-class conversations. Visit supermanage.ai slash breakdown today to start making the most of your one-on-ones. Thanks again to Supermanage for sponsoring the AI Breakdown. Welcome back to the AI Breakdown. Today we are talking about a topic which is coming up more and more, which is a question about whether we're seeing the first signs of the AI bubble pop, or at least show some serious signs of strain. The basis of this is that since November, with the launch of ChatGPT, everything surrounding AI has been just absolutely white hot. But it now feels like there are potentially some signs of waning interest or participation, and people are starting to wonder what it means.

8:22Now, where this really started to come up was about a week and a half ago, when someone noticed that ChatGPT had seen its first ever usage decline in the month of June. According to SimilarWeb, between May and June, ChatGPT's usage on both mobile and the web went down around 10%. It was the first time the company's numbers had gone down after a meteoric rise. However, as far back as March, the trend line had clearly shifted to a much more restrained pace of growth. Now interestingly, SimilarWeb also showed March as being the peak for Bing, which had been on the slight downclimb in each of the three subsequent months, while for Bard and Character.ai, their peaks had also been in May, coming down once again slightly in June.

8:59Now, mainstream media was very quick to jump on this and do a bit of narrative shaping. On July 7th, the Washington Post published, The piece says, The drop-in usage suggests that the tech's limitations are catching up with it, and at least some of the hype around chatbots is overblown. Basically, this Washington Post paints the whole situation as one in which big tech has been just absolutely pumping up AI, only for people to finally realize that it's not all that impressive. And frankly, if you ever wanted to see an example of why tech doesn't really trust mainstream media, the conclusion that this piece jumps to after a single month of declining numbers, And the picture that it paints of mass disillusionment with this entire technology category is just so emblematic of that frustration.

9:43If one possible explanation is that people are just not interested in ChatGPT anymore, another possible explanation that some have offered has to do with seasonality. Basically, the idea here is that one of the major uses of ChatGPT is education. With students not in session, it's kind of natural that the numbers would be falling. However, even with that explanation, some still say that that should be a cause for concern. Bernstein analyst Mark Schmulich said, If it's school kids, that's a real yellow-red flag on the size of its prize. The idea of the chat GPT drop-off due to students and summer break implies a narrower audience and fewer use cases.

10:16Then, of course, making another massive leap in logic, Insider says, In other words, if a big part of chat GPT growth is driven by cheating students, this means the technology, or at least the chatbot format, may not be the dominant computing platform of the future. Now, of course, there are other interpretations as well. Some of those interpretations might recognize that in the context of a product that went from zero to 100 million users in five weeks, we just don't have a lot of precedent about how those growth curves are supposed to look. In fact, there is no supposed to because there isn't any precedent.

10:47The assumption here is that a 10 % decline month over month means an existential crisis. However, that could equally be just the press looking for another story because laudatory articles around ChatGPT aren't driving clicks anymore. I also think that holding aside the arrogance and casual dismissiveness of assuming that students are only using ChatGPT to cheat, writing off the educational use case as suggesting that other use cases are too small overall fails to recognize how much young people and students are inherently going to get a new technology tool before their older peers and other sectors will.

11:17In other words, the percentage of ChatGPT's users today that may be made up by students is likely going to be much bigger than the total percentage of educational users in the future. That is by simple virtue of the fact that students are much more likely to be early adopters than are, say, their 40-plus-year-old peers. But regardless of this, the point is that it really got the whole bubble conversation happening, and it's now starting to show up in other places. I discussed this more in depth on today's AI Breakdown Brief, but last week, Stability AI CEO Ahmad Moustak was on a call with UBS analysts and said that on the one hand, while AI will be a$1 trillion investment opportunity, it will also produce the biggest bubble of all time, with, of course, investors chasing after things that don't end up ultimately working out.

11:58Which brings us, I think, to the most interesting part of the conversation, and something new from the weekend. We're starting to see the first sign of trouble show up at AI startups themselves. Over the weekend, the information ran a story about how Jasper and Mutiny, both AI startups, had been forced to recently cut workers. The piece reads, Two startups developing products using generative artificial intelligence laid off workers recently, striking a downbeat note amid a wave of investor euphoria for the sector. Jasper AI, which sells software that uses OpenAI's GPT to help businesses create and fix text, cut staff this week.

12:29And Mutiny, which sells software powered by AI to personalize and improve website text, cut about 30 % of staff for around 30 jobs late last month. And of course, that inevitably leads to the question as to what extent this is about Jasper or Mutiny specifically versus a broader trend in AI generally. AI entrepreneur Sam Hogan had a really interesting post about this. He wrote, Six months ago, it looked like AI and LLMs were going to bring a much-needed revival to the venture startup ecosystem after a tough few years. With companies like Jasper starting to slow down, it's looking like this may not be the case.

13:00Right now, there are two clear winners, a handful of losers, and a small group of moonshots that seem promising. Let's start with the losers. Companies like Jasper and the VCs that back them are the biggest losers right now. Jasper raised$100 million at a 10-figure valuation for what is essentially a generic thin wrapper around OpenAI. Their UX and brand are good, but not great. and competition from companies building differentiated products specifically for high-value niches are making it very difficult to grow with such a generic product. I'm not sure how this pans out, but VCs will likely lose their money.

13:28The other category of losers are the VC-backed teams building at the application layer that raised$250k to$25 million in December to March on the back of the chatbot craze, with the expectation that they would be able to sell to later stage and enterprise companies. These startups typically have products that are more focused than something very generic like Jasper, but still don't have a real technology mode. The products are easy to copy. Executives at enterprise companies are excited about AI and have been vocal about this from the beginning. This led a lot of founders and VCs to believe these companies would make good first customers.

13:55What the startups building for these companies failed to realize is just how aligned and savvy executives and the engineers they manage would be at quickly getting AI into production using open source tools. An engineering leader would rather spin up their own Langchain and Chroma infrastructure for free and build tech themselves than buy something from a new, unproven startup and maybe pick up a promotion along the way. In short, large companies are opting to write their own AI success stories rather than being a part of the growth metrics a new AI startup needs to raise their next round. This brings us to our first group of winners, established companies and market incumbents.

14:24Most of them had little trouble adding AI into their products or hacking together some sort of chat your docs application internally for employee use. This came as a surprise to me. Most of these companies seem to be asleep at the wheel for years. They somehow woke up and have been able to successfully navigate the LLM craze with ample dexterity. There are two causes for this. One, getting AI right is a life or death proposition for many of these companies and their executives Failure here would mean a slow death over the next several years They can't risk putting their future in the hands of a new startup that could fail and would rather lead projects internally to make absolutely sure things go as intended Two, there is a certain amount of kick-ass wafting through the halls of the C-suite right now Ambitious projects are being greenlit and supported in ways they weren't a few years ago I think we owe this in part to Elon Musk reminding us of what is possible when a small group of smart people are highly motivated to get things done Reduce red tape, increase personal responsibility, and watch the magic happen.

15:13Our second group of winners live on the opposite side of this spectrum, indie devs and solopreneurs. These small, often one-man outfits do not raise outside capital or build big teams. Their advantage is their small size and their ability to move very quickly with low overhead. They build niche products for niche markets, which they often dominate. The goal is to build a SaaS product or multiples that generates$10k a month in relatively passive income. This is sometimes called micro SaaS. These are the level COO and Danny Postmas of the world. They are part software devs, part content marketers, and full-time modern internet businessmen.

15:42They answer to no one except markets and their own intuition. This is the biggest group of winners right now. Unconstrained by the need for a billion-dollar exit or the goal of 100 million ARR, they build and launch products in rapid-fire fashion, iterating until product market fit and cash flow and moving on to the next. They ruthlessly shut down products that are not performing. LLMs and text-to-image models a la Stable Diffusion have been a boon for these entrepreneurs, and I personally know dozens of successful, keeping in mind their definition of successful, apps that were started less than six months ago.

16:07The lifestyle and freedom these endeavors afford to those that perform well is also quite enticing. I think we'll continue to see the number of successful micro SaaS AI apps grow in the next 12 months. This could possibly become one of the biggest cohorts creating real value with this technology. The last group I want to talk about are the AI moonshots, companies that are fundamentally reimagining an entire industry from the ground up. Generally, these companies are VC-backed and building products that have the potential to redefine how a small group of highly skilled humans interact with and are assisted by technology.

16:34It's too early to tell if they'll be successful or not. Early prototypes have been compelling. This is certainly the most exciting segment to watch. A few companies I would put into this group are 1. Cursor.so, an AI-first code editor that could very well change how software is written. 2. Harvey.ai, AI for legal practices. 3. RunwayML, an AI-powered video editor. This is an incomplete list, but overall I think the moonshot category needs to grow massively if we're going to see the AI-powered future we've all been hoping for. If you're a founder in the $250K to$25 million raised category and are having a hard time finding product market fit for your chatbot or LLM ops company, it may be time to consider pivoting to something more ambitious.

17:10Let's recap. One, VC-backed companies are having a hard time. The more money a company raised, the more pain they're feeling. Two, incumbents and market leaders are quickly becoming adept at deploying cutting-edge AI using internal teams and open-source, off-the-shelf technology, cutting out what seem to be good opportunities for VC-backed startups. Three, indie devs are building small, cash-flowing businesses by quickly shipping niche AI-powered products in niche markets. Four, a small number of promising moonshot companies with unproven technology hold the most potential for VC-sized returns.

17:36It's still early. The landscape will continue to change as new foundational models are released and tool chains improve. I'm sure you can find counterexamples to everything I've written here. So this is a super interesting thesis, and there is a lot to unpack here. I think trying to answer the question of to what extent these layoffs are a reflection of the AI bubble itself bursting versus a problem of these specific companies, this piece effectively argues that the AI venture model isn't playing out the way that money had thought, but it doesn't necessarily follow that AI itself is a bubble that has popped.

18:06Instead, Sam argues that there are two big unexpected factors that are changing the degree to which the traditional venture model is the right fit for this industry as it's evolving. The first is how much enterprise companies are actually building internally, and the second is the extent to which small and indie developers who are bootstrapping have been able to fill in the gaps of small niches because they don't have the pressure to return at a venture-backed scale. If anything, I think that the interesting takeaways from Sam have more to do with the changing nature of the venture industry than they do with AI.

18:32However, as Sam points out and starts off with, AI has been largely seen as the great savior for the venture capital space that, as he puts it, has had a rough couple years. Now, unfortunately, when it comes to venture capital, those rough couple years are kind of predicated upon a bigger shift, which is the withdrawal from a zero interest rate world. For more than a decade, venture capital was experiencing basically endless inflows of capital looking for yield, and now that's just not the case anymore. It's inevitable that models are going to have to change is that money flows out and there's just less capital to go around.

19:01Anyways, it is a super interesting moment in the history of the evolution of this space and one that we are definitely going to keep a very close eye on. For now, that is going to do it for today's AI Breakdown. If you enjoyed this and haven't done it yet, please go leave a five-star rating wherever you listen to the show or whatever the equivalent is if your podcast app has a different system. I appreciate you guys listening as always and until next time, peace.

19:27Thank you.