Notes for April 5th 2024

Song of the week:


  • This week I took Tuesday in-lieu as a result of my aforementioned travel nightmare on Good Friday. Wife was working, kids were in crèche. So I took the initiative to clean the house, but also just go into the city. Alone. On the bus. And I did a bit of browsing, shopping (for second-hand vinyl) and a bit of solo coffee. I didn't rely on my phone. I didn't even rely on earphones. Just took it in. Because it's a rarity to do such a thing.
  • My former employer is HubSpot. And this week there's a rumour that they're going to be bought by Alphabet. Some thoughts:
    • This is not the first time this rumour has happened. This time, though, both companies are bigger and the rumour is far more public-interest garnering. FWIW we also had rumours internally of Microsoft making a pitch. But also, more significantly and more prosperously, Salesforce.
    • I think this isn't a simple "swallow a good company up" play. Because it's a little odd, right? HubSpot is so b2b, and Google is not. Why would Google want a b2b martech+CRM company with 200,000 customers that's trading 30x-plus it's revenue? I think there are 4 plays:
      • One: Expand the GSuite play. Imagine adding CRM or martech to the side of your existing GDocs, Gmail, etc. stuff. It makes all of that stickier in a world where there's plenty of competition for Google's business suite.
      • Two: GCP. HubSpot is a huge AWS user and Google is trying to legitimately compete in that hyperscaler space. This is probably a small-print reason to acquire a company for Google, but it's got to be compelling for some execs sitting at the decision-making table.
      • Three: Data. Google are the undisputed champions of LLM-trainable consumer data. They have Gmail, search and a bunch of other avenues to train their LLM. Whether you like it or not. But HubSpot, who have brought AI into a bunch of their b2b tools, has a swathe of clean, great customer b2b data that could help elevate that GSuite play.
      • Four: Sideswipe an ensuing brawl in the martech space (powered by AI) with Adobe & Canva. Canva just hired a HubSpot exec (who was ex-Adobe), as well as acquiring Affinity for pro-level desktop design/publishing tools. They're coming for Adobe, who have done great with mid-market and enterprise CMOs. Google could upset both at the same time by supercharging HubSpot.
  • There are a bunch of stories out there now of AI companies that made $x, but spent far more to achieve such amounts. Like, spent 10x their revenue to generate any amount of money at all. This is where AI will shit itself. Because there are 4 layers to an AI business, and all of the ones you know are at the top of the pyramid. Which, in the context of new tech, is not where you want to be. If you're the one digging for gold, you're the fool. The winner sold you the shovel. The 4 layers to AI are:
    • 1: The user interface. The chatGPT, CX app, etc. The actual interface you use.
    • 2: The data layer, which trains the LLM, which might not even know it is being used to train the LLM.
    • 3: The LLM itself. The fundamental core tech that's probably really just a really hyperscaled machine learning set of algorithms rather than a "thinking software" layer. I'm pretty skeptical that we've actually trained sand to think.
    • 4: The hardware. Which is where the big winner sits; nVidia.