Equity Monday: The AI race, crypto doldrums, and the future of fake fish

This time around we had a bit of a theme. See if you can spot it: Stocks are down around the world in the wake of Hamas' attack in Israel. Crypto is similarly in the red. China is working to bolster its national computing and data infrastructure. You know, the things that make AI function. In other words: the AI race between China and the United States is far from over. This Anthropic paper is a potentially big deal. The more that we understand LLMs, the better we can use them, right? Changes at YC! And from Startup Land: Wanda Fish Technologies just raised $7 million to make fake fish, and Lottie raised $21 million to keep tackling the UK care home market. That's our show for today! We have some really fun interviews coming up, so make sure to tune in on Wednesday! For episode transcripts and more, head to Equity’s Simplecast website. Equity drops at 7 a.m. PT every Monday, Wednesday and Friday, so subscribe to us on Apple Podcasts, Overcast, Spotify and all the casts. TechCrunch also has a great show on crypto, a show that interviews founders and more! Credits: Equity is produced by Theresa Loconsolo with editing by Kell. Bryce Durbin is our Illustrator. We'd also like to thank the audience development team and Henry Pickavet, who manages TechCrunch audio products. Learn more about your ad choices. Visit megaphone.fm/adchoices

• 0:10Speaker 00

  Hello and welcome back to Equity, the TechCrunch podcast where we unpack the numbers and the nuance behind the headlines. This is Alex. Good morning and welcome to our Monday show where I get caught up and I catch you up as well. It is October 9th, 2023 and on the show today, we have updates from the worlds of stocks and crypto, the rising global stakes of the AI race, a possible breakthrough on understanding LLMs, changes at a well-known startup accelerator, and two funding rounds that I found fascinating. Let's go.

• 0:45Speaker 00

  Let's kick off with a look at the world of money and that means stocks. Share movement was muted in Asia today. In Japan and South Korea, markets are closed for a holiday. Stocks are also mostly lower in Europe and they are set to fall here in the US. What's going on? Well, the attack on Israel by Hamas and concern of larger unrest in the Middle East is weighing on the global stock market and of course, we hope that everyone is as safe as they can be as quickly as they can be. Elsewhere, there are no earnings this week that we care about and there are no new IPO filings of note either, and that means we have ahead of us in financial terms, a pretty clear week. Much like the stock market, the crypto market is flashing red this morning. We're talking mostly

• 1:27Speaker 00

  red ink over the last week. Bitcoin is off about 2.5% in the last seven days. ETH is off a sharper nearly 8% over the same timeframe. And looking at charts, it does seem that many crypto tokens have had a pretty negative last 24 hours. What's driving the news? Well, recently, Yuga Labs announced layoffs. Now, we don't know how many people the company behind Bored Apes did cut, but it does show that even the biggest companies in crypto are stuck focusing more and trying to reduce their spend and burn. Recall as well that TechCrunch recently reported that Web3 funding from VCs in the last quarter was lackluster at best. And more recently, the

• 2:10Speaker 00

  block reported that Bitmain paid staff after delaying certain wages in September. Not a good look and perhaps implying that there is a very hard market out there today for Bitcoin miners.

• 2:23Speaker 00

  Next up, big news that matters, and I can't think of a bigger news item than how global this AI scrap is. Now rewind the tape. Before the LLM boom hit the global tech market like an earthquake, there was a lot of talk about China being ahead of the US in AI. Now, after the large language model moment, those arguments have died down, in part because I think the epicenter of the LLM moment has been firmly inside the US. However, the China-US rivalry is far from over. China's government recently disclosed that six of its departments are targeting 300 exaflops of national compute power by 2025, and that's up from about 220 exaflops today. There are other goals that were recently

• 3:08Speaker 00

  disclosed, including more storage capacity and quick data transfer targets inside the country. In short, it appears that the Chinese government is putting more and better national computing capacity in its near-term planning. And just guess what consumes a lot of data and compute power? Well, modern AI work, of course, so you can draw a link there as much as you like. The US is more diffuse and internally self-directed than China due to divergent government and economic setups. So, once again, we are seeing a rivalry of sorts between free market economies and more centrally planned efforts. Place your bets accordingly, but the US-China AI rivalry is far from over.

• 3:48Speaker 00

  Sticking to the AI theme, the world is abuzz from a recent paper from AI startup Anthropic called, and I quote, "Towards monosyllabicity: Decomposing Language Models with Dictionary Learning." That's a lot of syllables. What does it do? Well, it claims to have sorted out how, in small, large language models, to, quote, "find a path to breaking down complex neural networks into parts that we can understand, building on previous efforts to interpret high-dimensional systems in neuroscience, machine learning, and stats." Now, what does that mean? It means that the black box of LLMs can be somewhat figured out in terms of what is doing what and why. Anthropic said in the paper that the basic math of LLMs is

• 4:32Speaker 00

  Understood, but quote, "We don't understand why those mathematical operations result in the behaviors that we see." Now, hearing this news over the weekend about the paper, I was honestly pretty stoked. Seems very, very cool. More or less, learning why LLMs do what they do is a big task, but it could make the technology, I don't know, understandable, tunable, and maybe even better for use in corporate and personal settings. So this could be a breakthrough. There's a lot to think about here. One fun question that I have had bouncing around in my head is this. As we get closer to AGI, which I think is going to be more gradual than all at once,

• 5:12Speaker 00

  will our ability to understand the models in question become fuzzier or less fuzzy? Now, the Anthropic news points towards the latter, but I was always leaning towards the former. So now I'm very curious, as AI models become more complex, will we better understand them or understand them less well? Well, we'll find out soon, but Anthropic is pushing that tuner in one direction. As you can imagine, AI is going to cause regulatory headaches all around the world, but don't worry, startup land is gearing up for that work. Now, recently, TechCrunch reported that Y Combinator, a well-known startup accelerator, announced three recruits that are going to help out in its visiting partner group.

• 5:52Speaker 00

  That is Tyler Bosamini, Nate Smith, and Pete Cooman, former YC founders that are now kind of returning to the nest. Also, the company has hired Luther Lowe, who previously worked at Yelp in a public policy role. He'll be reprising that role at YC. Now, you don't hire a DC veteran in the policy space as a tech company or accelerator if you don't want to have a say. So startups are getting a little bit more politically involved, just as regulators are waking up to the fact that, hey, technology companies have lots of power. Call it a meeting of the minds.

• 6:28Speaker 00

  Putting aside the government and AI and LLMs, let's talk about some startups. Yeah. All right. First up, Fake Fish. Now, Wanda Fish Technologies is an Israeli food tech startup that just raised $7 million in seed funding, or perhaps, given that it's fish, egg funding? I don't know. To accelerate the initial production of cultivated bluefin tuna. Now, the company aims to create whole-cut cultivated fish fillets, starting with bluefin, and they're going to use muscle and fat cells to actually replicate the texture, flavor, and nutritional value of wild-caught fish. That's according to Daphne Heffetz, the company's co-founder and CEO, talking to TechCrunch. This $7 million round was led by Aqua Spark, which was joined by existing investors The Kitchen Hub,

• 7:13Speaker 00

  Peregrine Ventures, Pico Venture Partners, More VC, and CPT Capital. That's a lot of people. Anyways, to date, Wanda Fish has raised $10 million in venture-backed funding. Love the tech, love the work, woo. But why don't we take the tuna and instead make it taste like beef, which is delicious, because fish are gross?

• 7:35Speaker 00

  Let's move on from Fake Fish and instead talk about caring for our elderly relatives. Yes, let's talk about home care. Lottie, a UK-based startup, has built an online marketplace that helps those seeking care find the best care homes and services for their needs, while also catering to additional categories such as home care itself. Now, the startup was founded out of London in 2021 by a couple of brothers, Will and Chris Donnelly, and the company now claims some 500,000 monthly users, which it says represents 300% growth in the last 12 months. Now, the company raised $7 million last year and today announced a $21 million Series A

• 8:17Speaker 00

  that was led by Excel and participated in by General Catalyst. Home care is a big deal. Populations around the world are graying and need help, and governments are often unable or unwilling to step up enough, so hence the private market. And there have been some issues at care-focused startups before, including Papa here in the US. We did a whole show about that. But there are other examples of care-focused startups that are doing well, including Winnie, even if that company is more focused on childcare instead of elderly care. More or less, the care market is huge, and so seeing startups there, not that big of a surprise.

• 8:56Speaker 00

  Oh, that is the end of our lovely Monday show, but don't worry, Equity is back on Wednesday and Friday. And if I may be so bold, we have a couple of really cool interviews coming your way. So pay attention to our Wednesday episodes because those are going to rock. But if you need even more of us in the interim, of course, we are EquityPod on Twitter and Threads. And if you want to hear my boring musings as a dad, well, I'm Alex over there on Twitter. Equity has two sister podcasts, Chain Reaction and Found, that are well worth your time. Give them a look in every podcast app out there and we'll talk to you soon. Goodbye.

• 9:30Speaker 00

  Equity is hosted by myself, editor in chief of TechCrunch Plus, Alex Wilhelm, and TechCrunch senior reporter, Mary Ann Azvedo. We are produced by Teresa Locan Solo with editing by Kel. Bryce Durbin is our illustrator and a big thank you to the audience development team and Henry Pickavet, who manages TechCrunch Audio Products. Thank you so much for listening and we'll talk to you next time.