We believe frontier technology is entering a new phase. The first phase of the technology cycle was dominated by software, cloud computing, and digital platforms. The emerging phase is increasingly defined by the physical infrastructure required to make intelligence economically useful at scale.
Artificial intelligence is becoming the central demand driver across this ecosystem. More AI requires more compute. More compute requires data centers, semiconductors, networking, and cooling. Those systems require enormous quantities of reliable electricity. The result is a capital-intensive investment cycle spanning digital infrastructure, power generation, transmission, storage, and advanced energy systems.
At the same time, intelligence is moving from the digital environment into the physical world. Robotics, autonomous systems, and machine intelligence are beginning to converge, creating the foundations for autonomous factories, logistics, transportation, defense systems, construction, and eventually infrastructure itself. The investment opportunity may extend well beyond the robot manufacturers to the enabling ecosystem of sensors, actuators, specialized semiconductors, batteries, simulation, software, and fleet management.
Five foundational layers
We see five foundational layers emerging in the frontier economy.
AI, agents and machine intelligence
The demand engine. Every layer below exists to feed it or to put it to work.
Semiconductors, data centers and networking
Where intelligence is manufactured. Supply is set by fabs, packaging, and power, not by demand alone.
Nuclear, gas, renewables, storage and grid
The binding constraint. Firm, deliverable power is increasingly the scarce input.
Robotics, drones and autonomous systems
Where intelligence meets labor. Value spreads to the parts and software every fleet needs.
Space and other extreme environments
The next operating domain, from orbit to the deep subsurface.
A second generation
Synthetic biology, advanced materials, quantum computing, advanced manufacturing and next-generation propulsion.
Pick the bottleneck, not the winner
We seek to identify the infrastructure, enabling technologies and scarce resources required for the adoption of frontier technologies at scale.
The investment implication is significant. The opportunity may increasingly lie not in predicting which individual technology company becomes the winner, but in identifying the scarce infrastructure and enabling technologies that every winner will require.
In our experience, scarcity in this cycle is physical and slow to relieve: grid interconnection and firm power, advanced packaging and specialized chips, cooling and heat rejection, permitting, skilled construction labor, and supply chains that meet domestic sourcing and foreign-entity rules. Assets that sit on these constraints can earn durable economics even when the leaders above them change.
How we read the market
Periods of pullback in AI and frontier names are part of any capital-intensive cycle, not proof that the cycle is over. We separate price from progress. The questions we ask are practical: is capacity being used, is power contracted, is demand coming from customers who pay, and is the build funded by cash flow rather than leverage alone.
Physical constraints cut both ways. They slow deployment and test patience, and they also protect the value of assets that already have power, permits, and supply. When sentiment resets, disciplined investors get better entry points into the same long-term demand.
Three horizons
Commercial now
AI infrastructure, data centers, semiconductors, power infrastructure, nuclear energy, robotics and autonomous systems.
Emerging platforms
Space infrastructure, synthetic biology and advanced materials as growing platforms for economic growth.
High upside, higher uncertainty
Quantum computing, fusion, brain-computer interfaces and space manufacturing.
One industrial ecosystem
This creates a broader investment framework. Frontier technology should increasingly be viewed as an interconnected industrial ecosystem rather than a collection of individual technology sectors.
The shift we are watching is from software to infrastructure, from digital intelligence to physical intelligence, and ultimately from today’s infrastructure to autonomous infrastructure.
What would change our view
Efficiency gains that sharply cut power and compute per unit of intelligence. Long delays in nuclear and grid build-out. Capital costs that outrun returns. Policy shifts that break supply chains. We track each of these, and high conviction is not low risk.
For investors, the key question becomes: where are the bottlenecks, scarce assets and enabling platforms that will capture value as intelligence moves into the physical economy?
This note is education for accredited investors, not a solicitation or a prediction about any single issuer.
← All Insights