00:20:58 · Jordi Hays, John Coogan
[00:20:58] S0: He's baiting for sure. He's like how dare you accuse. Yeah. He's like you can go back and look at how many times I've used [00:21:04] S1: Dell for millions of I created Dell. I coined that word. [00:21:08] S0: There's a YC company called Dell [00:21:12] S1: Oh, yeah. [00:21:13] S0: Yeah. As well So that's doing [00:21:16] S1: he says, but it's interesting that the goo that the earnings call was Ruth Porat's last one as CFO. Porat earned a lot of plaudits for getting Google spending under control in the late twenty tens, but what seems clear in retrospect that is that 52,500,000,000.0 that Google spent on CapEx in 2024 was too little. And so he's kind of like, you know, noodling on this idea that maybe the CFO was too cautious going into an AI boom. [00:21:48] S0: You could lose your job [00:21:49] S1: over that. Called a maybe Ruth Porat was calling top signals like us. It looks terrible. Fortunately, we don't lose our job and we There's [00:21:57] S0: nothing wrong with calling top signals. [00:22:00] S1: There there there's something wrong [00:22:01] S0: with underinvesting. Water if you're calling specifically calling the top. Yes. Trying to identify top signals [00:22:07] S1: Yes. [00:22:08] S0: Is more of a meta game.
00:22:08 · Jordi Hays, John Coogan
[00:22:08] S0: Is more of a meta game. [00:22:09] S1: Yeah. Yeah. Yeah. So, basically, Google, it it it does seem like Google under invested in CapEx in 2024 based on demand, wound up with that massive backlog, missed on top line, couldn't couldn't make enough money, couldn't generate enough revenue, had higher margins, that's great, but but didn't deliver on the cloud side on on the actual scale. So Ashkenazi's Ashkenazi's calls, meanwhile, have repeatedly reiterated that Google Cloud is particularly supply constrained. They don't have enough servers. And the company has now surprised investors twice in six months with the scale of its CapEx plans. Let's go. Says, this, in my opinion, is incredibly bullish for Google. Go back to this, the disruption lens, which we were kind of noodling on earlier. The exact avenue from which you would expect management resistance to a disruptive innovation to flow is from the is the CFO office. Like the CFO should be resistant to disruptive innovation. And when you look at the history of companies that got disrupted, the CFO is saying, we have a good business. Let's just keep printing cash. Like this new thing, let's not worry about it. Let's not Nokia [00:23:17] S0: CFO. Exactly. Are you guys seeing the numbers?
00:23:17 · Jordi Hays, John Coogan
[00:23:17] S0: CFO. Exactly. Are you guys seeing the numbers? [00:23:20] S1: Yeah. But, like, for a lot of those companies that got disrupted, it's like they were printing cash, high dividends for decades even after the disruption happened. The next, the iPhone comes out. And in order to actually do something to compete, they have to completely go into start up mode. They have to burn a ton of cash, cut their dividend, stop stockpiling cash, issue debt, do a ton, raise more equity maybe. Like, they have to become a new product development company, and it would be very, very difficult, usually. [00:23:48] S0: BlackBerry's annual revenue in 02/2005, 1,900,000,000.0. BlackBerry's They completely BlackBerry's annual revenue in 2024, 580,000,000.
00:23:48 · Jordi Hays, John Coogan
[00:23:48] S0: BlackBerry's annual revenue in 02/2005, 1,900,000,000.0. BlackBerry's They completely BlackBerry's annual revenue in 2024, 580,000,000. [00:23:59] S1: It's a cybersecurity company now. They bought a couple cyber security assets while they were valuable, while they were like, you know, while the stock was up. And then eventually, they wound down the phone business and just kind of became a cyber security company. Because obviously they had a ton of enterprise contracts because BlackBerrys were sold into like large enterprises as you know, work phones. Very few people had BlackBerrys as like everyday phones. But so they had all those, they had like solid sales team, solid connections, bought a new asset and were able to continue the business even though obviously it's not the company that it once was. So Ben Thompson goes on to say there's a there does seem to be a major shift in mindset in terms of the company's willingness to lean into AI, particularly for a segment Google Cloud that even in the best case scenarios is significantly lower margin than the company's core business. What's interesting about that margin point, however, is that it too is another reason to be bullish on Google's prospects. Google is the only company we talked about the TPU thing. They're the only company with an at scale ASIC alternative to NVIDIA's GPUs which should give them a meaningful cost advantage to that end to the extent that cloud compute [00:25:13] S0: At scale [00:25:13] S1: to a
00:25:13 · John Coogan, Jordi Hays
[00:25:13] S1: to a [00:25:13] S0: point where they've they're gonna be able to support SSI. SSI. Which is crazy. Not just their own needs. [00:25:18] S1: And then and then there's also another question [00:25:20] S0: in the sorry. The dynamic that's interesting is, like, OpenAI needs to massively scale infrastructure and and as we covered earlier this week commit to spending tens of billions of dollars with with Oracle. Yep. And their revenue today doesn't support that. Obviously, their their growth trajectory is absolutely insane. But OpenAI being in a position where they are competing head on with Google for this very very critical consumer use case, search knowledge retrieval [00:25:50] S1: Yep. [00:25:51] S0: And ultimately usefulness. It is a extremely tough position for OpenAI to be in.
00:25:57 · John Coogan
[00:25:57] S1: Yeah. So the other hyperscalers are working on ASICs that are alternative to NVIDIA GPUs, but they all seem to be following the similar path, trying to get line time at TSMC, not fully at scale. Then there's also the question, you know, I think a lot of people's mind is like, is this the end of history? Is there something that's coming down the pipe that could disrupt the NVIDIA GPU monopoly, the CUDA ecosystem, or even the Google TPU? And semi analysis has a funny meme here way down in the stack about the next generation of chips that were that that have been pitched from startups like Etched and a few others. So SemiAnalysis says, although baking transformers into silicon may sound cool, it's most mostly just a marketing slogan. 90% of transformers flops are just GEMMs and two fifty six by two fifty six or one twenty eight by one twenty eight systolic arrays in TPU and in TPU tranium are already optimized for these. Even modern GP GPUs with tensor cores are optimized well for GEMMs. Even if you bake transformers into silicon, aka just create a giant systolic array, most of your die area will still be taken up with SRAM cells. That's the memory. And you will still face the memory wall since your HBM memory bandwidth will be the same as GPGPUs, TPUs, and Trainium. And so the meme is transformers are just 90% matmoles. George Hotz had a similar analysis or take just saying that if you look at the actual energy use of GPUs right now, there isn't that much opportunity to squeeze more value out of them. They're pretty efficient and most of the most of the energy that goes in goes into these very, very specific math calculations that are already pretty optimized. We did hear from someone something about the this is from Martin Scarelli that Jane Street or was it was it Jane Street? One of the high frequency trading firms figured out how to run one of those very, very basic map mole calculations on a GPU more efficiently, and I guess kind of open sourced it or something or got out into the into the research world. So it it doesn't it doesn't feel like at this moment, I still think that there's a interesting bull case for the etched crew in some ways, in some niches or something, some specific models.