Still in pre-training... eu/acc Pee-seed/seed in Lovable, Tandem Health, Bottlecap AI, Recursive Superintelligence, AMI Labs, Callosum etc.
Stockholm
Joined June 2018
- Tweets6K
- Following565
- Followers2.3K
- Likes1.5K
Learned about Dunba'rs number last week
The cognitive limit of approximately 150 stable social relationships that a human can comfortably maintain at any given time
Choose wisely
Oliver Molander retweeted
We’re going to see a bunch of neolab to neocloud pivots next year
Oliver Molander retweeted
This Apollo report, claiming AI has returned "zero financial payoff" outside of the Mag 7, is a complete non sequitur.
They show that profit margins and revenue per employee is flat outside of the Mag 7, therefore the rest of corporate America is not seeing any returns to AI spend.
Who sees the problem with this analysis?
Obviously: this is not how you measure the contribution of a variable. You need to measure returns between companies that use MORE AI vs that use LESS AI. A flat aggregate like profit margins across all companies is totally incapable of answering this question.
Breaking out Mag 7 vs everyone else is also not the right unit of analysis. Mag 7 are mostly AI sellers, everyone else is mostly an AI buyer.
Look at the backdrop. Interest rates are rising, oil is at $116, inflation is rampant, and consumer sentiment is a record low. If the AI buyers are holding onto their profit margins despite that, this is consistent with AI is actually buoying the economy.
TL;DR: the conclusion of this article is completely ass-backwards. Very strange writeup.
Source: apollo.com/wealth/insights-n…
Oliver Molander retweeted
Databricks Chief AI Scientist Jonathan Frankle (@jefrankle) says the bottleneck to enterprise AI adoption is evals, not models.
"The models are good enough at a lot of really complex enterprise tasks and knowledge tasks"
"the real challenge is describing what you want AI to do and knowing whether AI did it"
Oliver Molander retweeted
It is very important for the entire ai startup ecosystem for key inference vendors to be 1) capable of getting capacity that they can freely allocate, 2) secure, 3) reliable, 4) always driving cost and perf
let’s not forget how much of an enabler cpu-era cloud was for innovation! no capex! outsource much of your infra ops (and more every year)!
every small-mid customer “owning, building and running their own landed power, cooling qnd chips” completely ignores the lessons of economic history. we had colo 20y ago…public cloud won
in an age where closed frontier labs weigh every (limited) GPU served by api vs. internal training and 1st party product use, inference abundance beyond them is a prerequisite for a rich startup ecosystem
Oliver Molander retweeted
Brutal truth from @Lethain
“If you're in [middle] management (…) you sort of have to decide, am I retiring now or am I re-skilling?”
He did just this: step down from leading a large team (250 ppl) to be hands-on with a small team (25 ppl), ship code & re-learn the new “rules”
Why a 3x CTO went back to a 25-person team. From Will Larson @lethain, Imprint CTO and 4x book author.
"It's a hard time in middle management, right? There's a lot more competition for these roles. What I found personally very rewarding is getting back into the details of writing software.
When I left Carta, my team was around 250. When I joined Imprint, the team I was leading was 25.
This a unique moment to actually re-skill. If you're in management, I think right now either you have to get back into the details, or the world is changing so radically that the profession you know will not be valued in three years. You sort of have to decide, am I retiring now or am I re-skilling?
I think there has to be some openness to taking a backward step or a lateral step. What I knew I wanted was getting back into the details and relearning how to do software in the current age. That's what I believe is going to be the unlock to the longevity of my career."
Yesterday we got Fable 5.1 level intelligence as a local model
(Kimi K3.1 Pro, open weights coming Nov 4th)
I remember the GLM founder saying he expects Fable 5 capabilities from China by the end of the year
It seems to have happened
Chart: Artificial Analysis
TikTok is targeting 4 GW compute capacity by the early 2030s in the US
This might be the most underrated AI story of today.
According to CNBC, TikTok's US business is looking to secure multiple gigawatts of computing capacity, with one source suggesting the company is targeting 4 GW by the early 2030s. I don't think people understand how significant this is. We're talking about a social media platform potentially looking to secure enough power and computing infrastructure to rival some of the largest AI data center developments in the country. For perspective, Meta's massive Hyperion data center campus in Louisiana is targeting around 5 GW of capacity.
TikTok already relies primarily on Oracle Cloud Infrastructure, and the company has confirmed that $ORCL will remain its primary cloud provider in the United States. On top of that, TikTok is retraining its recommendation algorithm using American user data and is developing AI products focused on large model inference and creative automation. In other words, this isn't necessarily just about supporting people scrolling through videos anymore. TikTok could be preparing for significantly more AI workloads which would require enormous amounts of computing power.
Now connect this to everything else we've been discussing. Microsoft, Amazon, Google, Meta, OpenAI, and xAI are already competing to secure power, GPUs, land, and data center capacity. Now we're seeing another major technology platform potentially entering that same competition at the gigawatt scale. And remember, these companies cannot simply decide they want another gigawatt of computing capacity and have it available tomorrow. They need electricity, substations, cooling infrastructure, networking equipment, GPUs, memory, and facilities capable of supporting all of it.
This is exactly why I remain so bullish on the companies supplying the infrastructure behind AI. Whether we're talking about $NVDA, $MU, $ORCL, $BE, or the broader networking and power ecosystem, the number of potential customers competing for limited resources continues to grow. What's particularly interesting is that this demand is increasingly coming from companies outside the handful of frontier AI labs everyone focuses on. I think we're still underestimating just how much infrastructure the AI economy is going to need over the next decade.
I'm still amazed how Sweden can punch above it's weight in both music (Abba, Avicii, Swedish House Maffia, Max Martin, Ludwig Göransson, etc) and technology (Spotify, King, Klarna, Skype, Mojang, Minecraft, Legora, Lovable, etc)
Spotify was in so many ways the perfect manifestation of this
And Europe's first $100B tech startup outcome
Oliver Molander retweeted
Yesterday brought two data points about the health of inference businesses : SpaceX's third-quarter report & DeepSeek's inference post.
Overall, they demonstrate that inference businesses can be quite healthy even at current prices & efficiencies.
Oliver Molander retweeted
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine.
• Best-in-class fast local coding
• Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark
• Built from scratch in under 4 months
Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.