AI Infrastructure · Investment & Research

Building and investing in the infrastructure layer of the intelligent economy.

We research, invest in, and build long-term ownership in the companies powering intelligence — across digital intelligence and the emerging frontier of physical intelligence.

How Compute Notes works

Research is the engine.

Every position starts as research. Research sharpens into an investment thesis. The thesis becomes long-term ownership in the companies building the infrastructure of the intelligent economy.

01 · Intellectual engine

Research

Public writing on the economics of compute, models, and machines.

02 · Framework

Investment Thesis

Where the infrastructure layer of the intelligent economy will form.

03 · Ownership

Ventures

Long-term ownership in the companies building it — starting with Watt.

The Platform

One thesis, two frontiers.

Intelligence is scaling in two layers — digital and physical. Each needs an infrastructure and economic layer underneath it. Compute Notes researches, invests in, and builds across both.

Digital Intelligence

Watt

AI Compute Infrastructure

The optimization layer for enterprise AI adoption — managing AI workloads, cost, and performance across complex model ecosystems.

Active venture · onboarding design partners
Physical Intelligence

Robotics / Automation Pilot

Physical AI Infrastructure

Autonomous machines, robotics, and industrial automation — AI infrastructure extending into the physical world.

Pilot development stage
Investment Thesis
Every major technology platform eventually develops an infrastructure layer. AI is developing two.
Stage 1 · Digital Intelligence

Models, compute, software infrastructure

Intelligence delivered as software. The bottleneck becomes cost and control — the consumption layer of the agent economy. Watt is our first venture here.

Stage 2 · Physical Intelligence

Machines, robotics, autonomous systems

Intelligence delivered through machines. As robots and autonomous systems scale, they need their own perception, control, and unit economics — the next infrastructure layer.

We research, invest in, and build long-term ownership in the companies shaping the intelligent economy — approaching AI as a unit-economics problem, informed by a background in M&A and financial due diligence. Not a service provider, and not a fund with a fixed clock: an operator-investor building conviction positions over a long horizon.

Research Areas

Where we research, invest, and build.

01

AI Compute Infrastructure

The financial and consumption layer of the agent economy.

  • Watt Venture
  • AI Consumption Layer Core
  • AI FinOps Core
02

Physical AI Infrastructure

Intelligence meeting the physical world — perception and control.

  • Robotics Pilot
  • Autonomous Systems Pilot
  • Industrial Automation Pilot
03

Spatial Intelligence

The next interface between humans, machines, and space.

  • Drones Watch
  • Spatial Computing Watch
  • Space Infrastructure Watch
Ventures & Ownership

A portfolio built around the thesis.

Watt is not the whole story — it is the first proof point. Each venture tests the thesis in a real market before the next frontier is funded.

Digital Intelligence · Active venture
Watt
Building the optimization layer for enterprise AI adoption — helping organizations manage AI workloads, cost, and performance across increasingly complex model ecosystems.
In active development · onboarding design partners · first proof point of the thesis
Visit Watt →
Applied proof point

AuditMate

An AI co-pilot for audit professionals — the optimization thesis applied to a real, paying workflow.

Applied proof point

DealPilot

An agent for M&A and transaction-services work, built from years inside diligence rooms.

Physical Intelligence · Active pilot

The second frontier, already in development — a real pilot in a real market, not a slide.

Physical AI Infrastructure
Robotics / Automation Pilot
Exploring how AI infrastructure extends from digital systems into the physical world — through autonomous machines, robotics, and industrial automation.

Our first pilot applies Physical AI to agriculture, using intelligent systems to transform existing industrial workflows. Agriculture is a compelling entry point: real-world environments, proprietary data generation, and clear operational economics.

The underlying technology has broader applications across industrial automation, infrastructure inspection, energy, forestry, and other complex environments.

Stage: Pilot development
Research

Writing in public on the economics of compute.

The intellectual engine behind every thesis and venture — published openly.

Read all essays →
Newsletter

Get Compute Notes.

The economics of compute — and what we're building underneath it.

Or subscribe at newsletter.computenotes.co