- Forests Over Trees
- Posts
- Can math save OpenAI?
Can math save OpenAI?
Plus: Crypto is back and NASA is outsourcing moon missions
Welcome back to Forests Over Trees, your weekly tech strategy newsletter. It’s time to zoom-out, connect dots, and (try to) predict the future.
Here’s the plan:
Tech News Takes — super-short analysis and commentary
Strategy Tools — strategy nuggets (for business and life)
F/T Shoutouts — sharing launches, tech events, and other reads
Can math save OpenAI?
Plus: Crypto is back and NASA is outsourcing moon missions
Photo by Chris Liverani
⚡ Tech News Takes ⚡
What’s up: OpenAI put out a new model (“o1” aka “Strawberry”), its first model trained to handle math and other complex, multi-step queries. It’s 3x more expensive than GPT-4o, their most recent general-purpose model.
So what: I’ve loved testing this new model, and it’s a huge unlock for more learning use cases! But this doesn’t change my mind about OpenAI’s longer-term prospects. Branding, model size, and memory are not moats. Math won’t be either.
What’s up: On Tuesday, hours before a Snap event to launch their new “Spectacles” smart glasses, Meta announced this deal. Savage. Meta and Ray Ban have partnered on smart glasses since 2019, but now they’re committing to each other through 2030.
So what: Aside from being a brilliant thunder-stealing move, this partnership is smart. The 2023 model they released together beat expectations. Plus, as the AI, battery, and screen tech inside gets more efficient, these glasses will become super powerful. They’re on a compelling, ‘disruptive innovation’ path (low-cost, simple starting point to dislodge more complex attempts from Hololens, Snap, etc.).
What’s up: The Houston-based company will support NASA with near space (Earth to moon) communication and navigation services, including an array of satellites. Their specialty is lunar missions, and this win builds on their February 2024 lunar lander – the first privately run mission of its kind.
So what: I was surprised to see a non-SpaceX name on a satellite-related mission, although diversification is probably good for NASA (SpaceX runs 78% of US launches). But I’m less surprised at the eye-popping contract value, because 1/ we are in an all-out race to the moon against China and 2/ new-worlding is big business.
What’s up: First, former President Trump launched a de-fi lending project (World Liberty Financial). Then, Michael Saylor bought another $1.1B Bitcoin. Finally, Circle, the ‘USDC’ stablecoin issuer, is moving to One World Trade before their IPO.
So what: I know what you’re thinking – who cares about crypto right now!? For one thing, AI-hype sucked the oxygen out of the room. Plus, now you have headlines like the above, which sound borderline irresponsible. But as with other innovative tech, some of the crazies will turn out to be right. So don’t fall for the herd mentality, positive or negative. There’s real value here, and someday we’ll find a middle ground between the hype of a few years ago and the hate of right now. I look forward to that.
When I was learning to drive, I was terrified I’d fail my driver’s test… But over time, I realized there are tons of bad drivers who passed and are doing just fine. So here’s the thing — starting a business… is EXACTLY like that. Nobody has it all figured out. If you have an idea, try it. This week’s Sponsor — Shopify — can help. Learn more here. |
🛠️ Strategy Tools 🛠️
5 Forces for OpenAI
Today’s strategy tip is about being a jedi – in touch with the force(s)...
Earlier, I mentioned I don’t think OpenAI is in a defensible position.
But it’s been a while since I explained why, and it’s the perfect chance to revisit a classic strategy framework: Porter’s Five Forces.
So here’s a reminder on the five forces of competition in an industry: bargaining power of customers, bargaining power of suppliers, threat of new entrants, threat of substitutes, and rivalry among existing firms.
When any of those forces increases – you are weaker.
So let’s pretend you’re Sam Altman (congrats!). Here’s the situation:
Customers – Mix of enterprises + cloud co’s. There are fewer AI model companies than enterprise buyers, so you have negotiating leverage there. But cloud providers are a different story – they’re consolidated, strong, and building their own models.
Suppliers – Cloud co’s, again. This time for computing power to train and run the models. And indirectly Nvidia – one of the most powerful companies in the world – for its cutting-edge chips.
New Entrants – It’s expensive to compete directly. But your friendly neighborhood Meta is footing the bill and handing out open-sourced models like Halloween candy.
Rivalry – It’s incredibly high. Every other LinkedIn post is a graph (aka scoreboard) charting your performance and how dead-even you are with your competitors.
Substitutes – Google search? Ok, this is a bright spot at least.
So Sam… are you feeling confident? Think you can succeed by flirting with Nvidia while you stiff-arm the hyper-scalers and gnash your teeth at the other model companies?
I’m exaggerating, but you get the idea. This is a tough position.
🌲 F/T Shoutouts 🌲
Art for the Tech Future – Deep in a Twitter rabbit-hole a few years ago, I ended up following a bunch of AR/VR builders, and I’m glad I did. Really nice Vision Pro app demo from Ian Panchevre and Kaleido. Better tech doesn’t have to mean worse art.
Keeping Fans Hungry – Thank you to the reader/streaming-exec who shared this incredible report from Variety. This was my favorite pair of graphs – the watch time of binge vs weekly release shows. See the spike? It’s a mid-season gap to keep ‘em hungry.
Source: Variety
Stronger PM/ENG Relationships – For the many of you that work with engineers and/or PMs, I can’t recommend this latest Lenny interview enough (with Camille Fournier, ex-CTO at Rent the Runway). Worth a listen.
Want to suggest a shoutout? Send it here.
The forest is growing.
Feel free to share this post.
To connect or collaborate, please reply to this email or send me a note.
How was it today? |