An illuminated path connecting industrial history with a technological frontier

Our thesis · 2026–2045

Two curves.One economic re-founding.

The marginal cost of intelligence and the marginal cost of energy are both collapsing toward zero, and almost everything interesting in the next two decades falls out of their collision.

01Intelligence↓

02Energy↓

The collision

Either curve alone would define an investing generation.Both at once is not a market cycle.It is a re-founding of the economy.

Two luminous streams converging into a single field — 01Energy

01 · Energy

Electrons become the new oil.

Abundance becomes the strategy.

Solar plus storage is now the cheapest electricity in human history in most geographies, and the cost curve has not flattened.

The 20-year story is not simply that renewables replace fossil fuels. The more consequential story is what happens when energy becomes so inexpensive that we stop economizing it. Desalination at scale, direct air capture, synthetic fuels, vertical farming, and industrial heat. Entire industries once considered thermodynamically wasteful can become viable.

Our thesis is not to bet primarily on generation as it commoditizes, but on everything downstream of cheap electrons: grid-orchestration software, diverse storage chemistries including sodium-ion, iron-air, and thermal storage, interconnection and permitting infrastructure, and the industrial processes reinvented when energy becomes five times cheaper.

The wildcard is nuclear: small modular reactors and potentially fusion in the 2035+ window, driven less by climate policy than by AI datacenter demand. Compute is becoming the first customer in decades willing to pay a premium for firm, dense, 24/7 power. In effect, compute demand may underwrite the energy transition.

A crowd follows an illuminated route while a builder constructs another path — 02Mobility

02 · Mobility

Autonomy eats the car.

Then it reshapes the city.

Autonomous driving is beginning to work in the real world. The 20-year implication is larger than robotaxis: it is the unbundling of car ownership and the repricing of urban land.

If transport-as-a-service becomes three to four times cheaper than ownership, parking is released, commute radii expand, and real-estate value redistributes. The second-order bets are more interesting than the first-order ones.

Logistics networks redesigned around autonomous freight, with middle-mile operations as the wedge.

Electrified aviation opening regional routes to a new cost and operating model.

Software coordinating fleets, curb space, charging, and the physical movement of goods as an API.

Physical infrastructure layered with networks, coordination systems, and machine intelligence — 03Technology

03 · Technology

Agents and the trust problem beneath them.

AI agents move from chatbots to economic actors: negotiating, transacting, hiring one another, and running workflows end to end.

The less obvious call, and the one we anchor the fund on, is that the bottleneck shifts from capability to coordination and trust. When millions of agents transact on behalf of people and companies, the economy needs identity, provenance, reputation, verification, dispute resolution, and payment rails built for machine-speed and machine-volume transactions.

Alongside that, robotics becomes viable because the AI brain problem is solved before the hardware problem. General-purpose manipulation in warehouses, kitchens, and eventually homes arrives in the 2030s. Biology becomes an engineering discipline through AI-designed proteins, programmable therapies, and longevity treated as a condition rather than a fate.

Compute itself becomes geopolitics. Chips, datacenters, and energy form a single strategic stack that nations contest in the way they once contested oil.

The meta-thesis

Intelligence industrializescoordination.

The 20th century was about making energy and labor cheap enough to industrialize atoms. The next 20 years are about making intelligence cheap enough to industrialize coordination; the constraint moves from “can we do it?” to “can we trust it, permit it, price it, and connect it?”

The biggest returns will not necessarily come from frontier capabilities themselves; those commoditize quickly. They will come from the connective tissue: trust layers, orchestration layers, market mechanisms, and the physical infrastructure, including grids, fabs, ports, and water, that the software world overlooked for 30 years and now urgently needs upgraded.

Physical-world competence becomes scarce and valuable again.

Teams fluent in thermodynamics, supply chains, and regulatory reality, not just software, will win the largest markets, because that is where software is heading next.

Pattern recognition

Every generation or two, a fundamental input cost collapses.

Everyone who understood it early looks like a genius for the rest of their lives.

1880s

Steel & oil

Carnegie and Rockefeller noticed before others that the price of a critical input was about to fall off a cliff and positioned at the chokepoints.

1980s

The transistor

The funds backing the microprocessor era did not need to choose perfectly. They needed to be directionally present as compute became a million times cheaper.

2000s

Bandwidth & distribution

A handful of funds understood the internet as an economic phase change rather than a technology trend and captured extraordinary value.

Now

Intelligence & energy

For the first time, two input-cost collapses are happening at once. Either alone would define an investing generation. Together they redraw the economy.

A vast physical structure under construction at an illuminated horizon — 04Our hunting ground

Where value accumulates

The chokepoints.

Most capital will chase the obvious layer: the models, the chips, the things on magazine covers. That layer will produce spectacular winners and spectacular carnage because frontier capability commoditizes faster than anyone expects.

We are going where value always accumulates in a phase change. When intelligence is abundant, what remains scarce? Trust, coordination, energy, and physical-world competence. When a billion AI agents transact on behalf of humans, someone owns the rails they trust. When electricity approaches abundance, someone orchestrates the grid that delivers it. When software collides with atoms across logistics, water, manufacturing, and mobility, the winners are the rare teams fluent in both. That is our hunting ground.

The window

The people who invested in 1999 were late to the internet’s first act and early to its crash. Those who invested in 2009, into the wreckage when it felt reckless, caught mobile, cloud, and one of the greatest venture vintages in history.

From 2045, this window will look as obvious as the internet looks from today.

Your grandchildren will not ask whether you saw it. Everyone saw it.They will ask what you did about it.

Explore our focus