A grab bag of short observations
I spent most of the day driving to Portland and back. It wasn't a great day to write longer form content, but here is a grab bag of things that have been getting my attention.
Ted Gioia
I get a lot of inspiration from Ted Gioia. He writes with clarity and passion about many topics. Every year, I enjoy his 100 best recordings; I am introduced to so much new music from this list. Without Ted’s list, I would never have listened to Timo Lassy and Jukka Eskola and their Nordic jazz. I wish Ted would publish a Spotify or Apple Music playlist with all his recommendations, but he is allergic to the large streaming platforms. But it’s worth digging through his list.
He also has great advice on writing — “I became a writer by writing.” If you want to be a writer, you just have to start writing and learn as you go.
AWS and Complexity
Amazon AWS has a rich and complex architecture for its AI Cluster buildout. It will take me days to begin to understand all of this. AI Clusters are some of the largest machines that mankind has ever built; it is not a surprise that there is a ton of complexity.
That said, AWS has a penchant for complexity. Amazon AWS CEO Matt Garman at re:Invent 2024: “In fact, at this point, we’ve grown to where EC2 has 850 different instance types across 126 different families. What that means is you can always find the exact right instance type for the workload that you need.” — or you will never pick the right instance type. And Simon Willison on Amazon’s new models — “Gaining API access to anything on AWS remains horrifyingly difficult - seriously, 17 steps! Some day a PM at AWS is going to figure out that fixing this will have a material impact on Amazon's bottom line.”
Annual Roundups
Benedict Evans does a nice annual presentation on tech trends. This year, the theme is obviously “AI eats the world.” I think Benedict understates the impact.
A16Z’s big ideas in tech for 2025 and Y Combinator Requests for Startups 2025. Take these with a grain of salt; both organizations have vested interests, but they are still interesting.
Rabbit Holes
Non-admitted insurance — I had never heard of this type of insurance. It makes sense that in really high-risk areas, we may want to allow the development of non-standard insurance products. The risk is, of course, that these insurers are not actually able to meet their obligations.
More than you ever wanted to know about heat pumps.
AI Tools
One of the best users of AI tools gives his counsel on when to use them, when not to use them, and how to use them.
Scaling
Any use of photonics is pretty exciting when it comes to extending Moore’s Law — Nvidia, AMD and Intel Invest in Startup Bringing Light to Chips.
SemiAnalysis also has a great optimistic article on AI scaling. SemiAnalysis is a great resource.
Automotive
GM shuts down Cruise, and Ben Thompson assesses the move. Ben’s closing words are, “…GM is probably screwed anyways.” This sentiment is broadly relevant for legacy automakers — automakers spending lots of money on software for their dwindling fleet of cars are probably making an expensive mistake.
Comments ()