Alternative Data Weekly #281

Theme: Insight per Unit of Attention

Mar 13, 2026

Special thanks to our sponsor Carbon Arc.

Carbon Arc’s AI platform gives access to proprietary economic and transaction data. Ask research questions in natural language for as little as $20 a month, and get structured results across many use cases in seconds. From equity research to prediction market edges — and everything in between.

Get started with a month free trial, using the coupon code ADW30.

QUOTES

“A year ago, it was an open question whether AI tools were making programmers more productive. Now I think the evidence is clear — and it’s only going to get more lopsided as AI models continue to advance.” – Timothy B Lee

News
Pods
Charts
Final Thoughts (4K Subscribers + 35% open rate – Thank you!)

#1 – Asymmetrix published Not everything is a data business - Anthropic announces a flurry of partnerships with data providers. March 2026.

My Take: There is a big, unanswered question about where the value flows in the world of data. Data ownership or data distribution/access. This article makes the case that there remains a lot of opportunity in the data & analytics industry (I agree) as AI interfaces make data more accessible in some sort of Jevons Paradox-type of growth. Human insight on top of great tools and unique data will create a ton of value.

#2 – Kyle Poyar published A new vision for AI pricing. March 2026.

My Take: this is interesting to me. I’ve historically not liked pricing with “credits”. I never know how much I am spending and it feels like it is intentionally opaque. That said vendors want to (and should) get paid for value provided. There are services I am paying $20/month for that I love & use daily. There are services that I pay much more for and rarely use, but “everyone has access”, so you need it to (right?). Better pricing models that are more aligned with the value added is a positive. The author uses the analogy of a car (lease/car purchase = cost; gas = value). The amount your pay should be aligned with the value you are getting from the tool. MAybe driverless taxis (see podcast below) line up cost/value more closely as you are only paying for rides (going from A—>B).

#3 – Andi Mann published Why Your Observability Platform Has Become A Bottleneck. February 2026.

My Take: modern computing systems now change faster than humans can observe them.
SREs (System Reliability Engineers) are caught in the middle. They are responsible for keeping real-time systems healthy, but the tools available to them were designed for an era in which data could be collected, stored, and analyzed later. Today’s anomalies don’t matter tomorrow. They matter right now.

“The ultimate measure of success is reducing the number of times a human needs to open a dashboard at all.“

BONUS: Yordan Ivanov published The Smart Engineer’s Framework to Staying Current Without Noise. March 2026. “The real objective is different. You want to maximize insight per unit of attention.”

What else I am reading:

Tavis Lochhead published The 2026 Investment Data Stack. March 2026.

Jody Hesch published The Proof is in the Pipeline (Part 1). March 2026.

Daniel Beach published Data Engineering Blogs to Follow. March 2026.

Dhaval Vithalani published One Observability Goal, Four Agents, Many Trade-Offs: Why No Single Agent Fits Every Use Case. February 2026.

Timothy B Lee published The Pentagon’s bombshell deal with OpenAI, explained. March 2026.

James Miller & Nick Zervoudis published The Business Case Isn’t the Decision. March 2026.

Source: The Data Exchange with Ben Lorica published an interview with Evangelos Simoudis Is Waymo Actually Profitable? The Real Cost of the Robotaxi Revolution. February 2026.

My Take: Autonomous vehicles (AVs) are an interest of mine. At it’s core, this is a massive data play.
I learned a lot of interesting new things in this podcast. How to measure success? Cost per mile? Cost per vehicle? Sensors (Waymo) vs no sensors (Tesla)
Lemonade (the insurance company, not the drink) is taking Telsa data into consideration for insurance rates.
Waymo is now multicity and expanding quickly, Tesla is in Austin & several pilots, Zoox is in Vegas and is testing SF.
Dark Spaces will be the future, meaning no people are needed … dark roads, dark warehouses, dark industrial kitchens, dark data centers (in space?).

HIGHLIGHTS (34-Minute Run Time)

Minute 01:00 – discussion kicks off; where do we stand with autonomous vehicles economically.
Minute 06:00 – involvement of human operators
Minute 06:45 – insurance issues
Minute 08:45 – Waymo economics; losing money now
Minute 10:20 – AV 2.0
Minute 13:00 – Does Lidar work well in all-weather situations?
Minute 15:30 – nothing is 100% accident proof
Minute 16:00 – economics are improving
Minute 18:20 – “Data center rebellion is here” (not just US)
Minute 26:45 – will we have an overcapacity problem?
Minute 29:00 – half life of the AI chips