Monumo, a Key Component of the AI Era Economy

All of the companies I have been working with have been those that are poised to capture benefits at the intersection of energy and information where I believe the most value is accruing at this point in the transformational changes brought about by the AI era. None better defines this than Monumo, which is building an AI foundation model for electromagnetics to be able to deliver design, control and monitoring for the world’s best motors, actuators and generators, These are at the core of both the energy production that AI era computation relies on and the agency that AI will deliver in the real world, through motion or actions.

Electromagnetics are at the core of power generation and delivery in the global economy.

The ‘old’, internet economic model.

The internet drove value creation downstream to replace traditional sales and marketing with newer forms of distribution that captured the benefits of networks that the internet delivered. Specifically these benefits consisted of virality (new users get more users) and network benefits (new users deliver value to existing ones). Broadcast era marketing of high production TV commercials with high profile stars were replaced by memes and influencers, while low cost, paper junk mail and cold calling were replaced by growth hacking and performance marketing. Meanwhile, traditional, in-person sales were replaced by ecommerce and bottom-up rollups of SaaS accounts or engineer-to-engineer selling, such as that practiced by Palantir. Virality meant things could grow abnormally fast in advance of monetization and network effects meant that markets were winner-takes-all. This meant growth was more important than revenue, until a market had been captured and that equity based, Venture Capital financing was needed to subsidise this growth and share some of the upside, to offset the risk.

The internet era business model was about distribution network effects.

The new, AI era economic model.

AI is not just another phase of the internet, it is a revolution of possibly greater importance, and while it will coexist with internet era economics, AI economics are totally different because of differences in where network benefits operate. AI software is not like codified ‘if this then that’, deterministic logic, running on top of a database. With AI, the model is both software and data store, it is probabilistic rather than deterministic, and the model itself literally is a network. As a result, the network benefits play out in the model itself dramatically lowering the cost of software production. Even if AI models themselves cost lots to train, much of the overall cost is in inference (running), the human development costs drop to zero and they are capable of generating new application code for very low cost. Meanwhile, unlike with databases where indexing means that the computational cost scales as the logarithm of the size of the data, AI models cannot be indexed in the same way, so the cost as the model size increases, gets exponentially larger. These two factors combine to invert the economics of software businesses, until now. In the past software was expensive to create and cheap to run, in the AI era software is cheap to build and expensive to run, driving value creation upstream to data, chips, data centers and energy provision. This is why we have seen such explosive growth in computing infrastructure spend, the valuation of hyperscalers and the ecosystem of providers of components of AI infrastructure hardware, such as Nvidia. It is why Amazon, Oracle and Microsoft, among others, have been literally looking to bring nuclear powered data centers online.

The AI era business model is about production network effects.

A correction doesn’t change the fundamentals.

Just as the dotcom era saw the bursting of a bubble created in internet stocks, but the internet continued and eventually grew to exceed the expectations of the bubble, the same could happen with AI if there is infrastructure overbuild. In addition, shifts in approach or architectural innovation, such as the shift away from training larger models to having existing ones think more about responses or the use of mix-of-experts to achieve something akin to indexing by switching off sections of models could trigger a correction where supply outstrips existing demand for AI computation. Nonetheless, the demand will continue to grow, so we can assume we are in an era where there will be a need for abundant computation and the energy required to deliver it.

The future economy will be based on the interplay of abundant energy and computation

Electricity 2.0

All of the energy required for AI, whether it be nuclear, solar, wind or hydro, with the obvious assumption that it will need to be zero emissions, will need to be electric and all of the electricity delivered, apart from solar, will be via generators. This will be at a time where this extra demand will be competing for resources with the electrification of much of human transportation, particularly cars. With generators, a significant portion of the fuel that will run the future world economy will flow through a device which has changed little since before the information economy existed and whose principal operational characteristics are determined by the shape of an electromagnetic field. Monumo therefore delivers a 21st century design of the device that powers the modern economy. But this is only half the story.

Generators are at the heart of all non-solar, clean energy

AI Agency

AI agents transform AI systems from ones which can answer questions to ones that can perform tasks. In the virtual environment, these are software tasks, such as booking a flight or reserving a restaurant. In the real world, AI tasks are things such as laying a brick wall or emptying the dishwasher. To date, robotics have been largely controlled by traditional software of fixed logic and perform specific tasks or ranges of tasks. AI allows robotics to become general purpose, spanning multiple tasks and with the ability to learn. General purpose technologies are rare and extremely valuable. As such, until this year, the range of estimates for the size of the robotics market was between $25 billion and $350 billion. AI robots are a fundamentally different technology, however, and when combined with the possibility that general purpose devices could achieve almost any task (capturing network effect in terms of ability) as of this year, Tesla suggested the market could be $7trillion, Ark invest put the figure at $24trillion and Nvidia suggested it being $50trillion. A recent Morgan Stanley report in humanoid robotics put the total addressable market as $60trillion, or the majority of global GDP, based on the assumption that AI robotics can deliver a workforce of any capability at any scale. Where Monumo fits into the AI agency economy is that general purpose AI robotics will be electric (Boston Dynamics is retiring its last hydraulic system) and will depend on the design of the worlds best electric motors for actuators with sufficient efficiency, reliability, precision control.

While it’s clear that agency in the real world is the other half of the AI agent economy that starts to become real in 2025, physical AI may take longer and what really matters is the path to physical AI, or its go to market.

Electric motors and actuators are at the heart of all physical AI to deliver agency in the real world

Automotive as the GTM for Physical AI

The car industry is being completely disrupted by the transition from Internal Combustion Engine (ICE) vehicles to Electric ones (EVs), where 2024 marked the year when it became clear that China, the world’s largest car market, would be an electric only market for cars.

Electric cars may look the same as traditional ICE vehicles, but with 80% less moving parts and differentiation based on batteries, electric motors and software, their business model, expertise requirements and supply chains are entirely different. Monumo obviously acts as a critical, component of enabling delivery of the best electric motors, optimised for the driving profiles of particular vehicles and works with many OEMs. Monumo’s go to market with AI enabled design optimisation is the first step on a path to developing a Large Engineering Model (LEM) whose core component is the optimum shape of electromagnetic fields and the motor or generator design required to produce them. An LEM has additional benefits beyond motor design, extending into effective operation and fault monitoring. Continuous monitoring of deviations in the magnetic field, corresponding to particular environments or use, allows for fault detection and precise motion and torque for optimised use and reliability. Monumo therefore offers the potential to not only deliver higher efficiency, but increase reliability and lower costs and material dependencies such as rare earth metals or sensors. By placing the electromagnetics AI model within a motor controller, Monumo delivers a truly intelligent electromagnetics, which gathers sensory information which creates a network effect of ongoing improvement through use. A further benefit is that since an AI model such as Monumo’s is based on physics, it is anchored in ground truths which we now know can vastly improve the accuracy and performance of AI models in post-training inference.

Intelligent Electromagnetics

EVs are merely the first step in a journey that EVs pave the way for in terms of intelligent motors. The reason for this is that EVs and their AI led software innovation have paved the way for autonomous vehicles (AVs), where autonomous vehicles can be thought of as AI robots that operate in a 2D decision space (turn left or right) for motion. Extending to 3 dimensions and picking things up and putting them down and you have a general purpose robot. AVs therefore provide a highly controlled environment as a go to market for general purpose robotics and the largest EV manufacturers, BYD and Tesla are both actively developing general purpose robots. Some of the most advanced uses of industrial robotics to date have been in automotive factories, and as the automotive industry is in the best position to manufacture general purpose robots it is also in the best place to be an early adopter of them within the controlled environment of its own factories. This creates a flywheel of using robotics to scale the production of robotics.

Automotive as the go to market for generalised robotics

As a result, the disruption to the automotive industry which threatens extinction for many existing manufacturers, opens up opportunities in a robotics market for the survivors which are vastly greater and which Nvidia CEO, Jensen Huang, suggests will be the largest technology sector of all.

The AI model for the Robotic Nervous System

Just as the human brain and nervous system are interconnected but different, AI robotics requires different AI models to process information to make decisions about actions and to process the actions themselves. A robot’s “brain” makes decisions on what tasks to perform, while its “nervous system” handles how to execute those tasks at the hardware and actuator level. In typical robotics systems this consists of three layers, where monumo operates at the low level:

• The high-level AI (brain) decides what actions to take and plans broader paths or goals.
• The mid-level motion planner translates these goals into detailed trajectories over time, ensuring the motion does not exceed torque or speed limits.
• The low-level controller directly handles the actuators, using sensor feedback to monitor and correct for any deviations in real time.

By combining world-centric planning with actuator-level control, modern robotics can achieve complex tasks in dynamic environments. The addition of advanced modeling (including electromagnetics) and AI for fault detection and optimization further refines the precision and reliability of these systems.

The robotics nervous system

At the center of AI and Energy
Monumo ultimately becomes the intelligence within the nervous system for AI enabled motion and action in vehicles, drones and robotics and in the generators that produce the electricity that powers them, to position itself as a key component at the hub of half of the world’s economy in the era of AI agency and clean energy.