There is a pillar of iron in Delhi that has stood in the open air for sixteen centuries and has barely troubled to rust. The smiths who forged it are nameless now. They left no memorandum and raised no round, and yet they had already solved a problem of corrosion that metallurgy would spend the following centuries catching up to. This is the oldest and most stubborn fact about inventions and discoveries - the need for willing, patient machinery to carry it from the moment of insight into ordinary life.

That pillar is a standing question. A thing can be got right decisively, and centuries before anyone can explain why it works and still end as a curiosity rather than an industry, admired by visitors and built upon by no one. The distance between those two fates is worthy of interest and of pursuit. The gap between knowing a thing and making it at scale is an engineering problem before it is anything else - a matter of institutions, of temperament, and above all, of capital willing to wait. And like every engineering problem, it yields to those who refuse to leave it alone.

The world's easy money has spent two decades learning the reflexes of software: the marginal cost of nothing, the patch shipped overnight, the instinct to move first and repair later. In the physical world, these reflexes are not merely useless. They are dangerous. Matter accepts no update over the air. A reactor core, a wafer, a new alloy, a vaccine - each must be right before it is sent into the world, not corrected once it is there. Deep technology inverts nearly every instinct venture capital has spent a generation perfecting. It rewards not velocity but patience; not the wrapper but the constraint; not the demonstration but the plant.

Picture the founder who walks out of a laboratory holding something genuinely new - a component for a quantum machine, a strand of engineered biology, a metal that behaves as no metal has before. She steps to the edge of a chasm. On the near bank stand the grant and the university, which can carry a technology from first principles to rough proof - from what the engineers call Technology Readiness Level one to perhaps four. On the far bank stands the market, which will touch nothing it cannot buy at scale and trust absolutely. Between them lies the most expensive emptiness in modern enterprise: the years and the millions required to build the first machine that makes the thing - the first line, the first factory.

Traditional venture capital, bound to the seven-year fund and the quarterly logic of the mark-up, looks across that emptiness and declines to cross. The great industrial houses, schooled in reliability, wait for a scale that cannot exist until someone first dares to fund its creation. And so, the discovery sits on the near bank, ageing, while the world is told generously that it was merely ahead of its time.

This emptiness has a name in our work. We call it the translation gap: the void between the research paper and the plant. Crossing it is the entire purpose of this fund.

How we try to cross it comes down to the three convictions earned slowly and paid for. They are less a strategy than a set of instincts we have stopped arguing with, and we return to them before every company we choose to back.

The first is that constraint, not convenience, is where durable value is made. We are drawn to founders who build against the hard ceilings of the physical world - the density of energy in a cell, the shedding of heat in a crowded chip, the strength of a material at a temperature that would ruin an ordinary one. Most software can now be copied over a weekend. A company that has spent five years learning to defeat a law of physics cannot be. Constraint breeds architecture, and architecture is the deepest moat there is.

The second is that the factory is not an afterthought to the science, it is the science's final and hardest chapter. The passage from a working prototype on a bench to a First-of-a-Kind plant is not a question of scale but of a different discipline altogether - a second invention, quieter and less celebrated than the first, and far likelier to break the company that attempts it. We look for founders who understand in their bones that manufacturing design is product design, and who have made their peace with the truth that the first production run will be costly, humbling, and full of failure. That is not the risk to be avoided. That is precisely where the value lies, waiting.

The third is that no fund, and no laboratory, builds the future alone. What a young deep-tech company most often lacks is not capital but a first customer - someone prepared to buy the earliest and most expensive units before the price makes sense, and by buying them, prove the thing can be sold. That customer can be the state, or it can be industry. Where the work touches a nation's sovereignty - compute, energy, defence - the state is the natural first buyer, as it was for our atomic and space programmes, which moved because the government purchased the first reactor and the first launch when no market would. Everywhere else it falls to industry, our incumbents and the world's strategics, to step out of the spectator's seat and become the first customer and manufacturing partner while doing so is still uncomfortable.

We have sat across the table from a few hundred deep-tech companies over these last years and put our capital behind more than thirty of them. Two or three have since wound down, and it is that last, smallest figure we return to most, because the companies that failed did not fail in the way the world outside assumes they must.

The assumption is that deep technology dies of its science: that the physics will not hold, the molecule will not bind, the device will never quite work. Now and then, that is exactly what happens. But by the time a company reaches us the science has usually already done its part; someone has proven it on a bench, in a paper, in the patient artificial calm of a laboratory. What undoes a company is almost always what comes afterwards - the long and unforgiving passage from a thing that has worked once to a thing that can be made ten thousand times, and sold as many.

Deep technology is fashionable now, and we have learned to keep fashion at arm's length. A great deal of money has arrived these past two years chasing the vocabulary of the field - climate, defence, semiconductors, and above all the promise of artificial intelligence, with little patience for what those words cost to build in the real world, or for the long interval between an announcement and the first rupee of revenue, and the longer one still before revenue hardens into income.

What we underwrite instead are the fundamentals, and they are unglamorous questions. Whether the thing can truly be made. What it will cost to make. Who, precisely, is making it, and against which limit of the physical world they have chosen to press. And, in the end, the question on which all the others rest: who will buy it, and why, and for how long. At the genesis of a company, when there is little in the room but a founder and a conviction, these are the hardest answers in the world to come by. Getting to them early, and honestly, is the greater part of what we do.

All of this - the questions of cost and scale and the first buyer, finally comes down to a kind of person. The ability that carries a company across the translation gap is the old scientific temper - the readiness to test rather than believe, to give up a wrong answer without ego, to stay with a stubborn problem long after the excitement has drained out of it. That, and not any single technology, is what we have quietly taken as our north star, and what we are truly backing. We invest in a cast of mind - in founders who care more for an atomic truth than a quarterly metric, and who would sooner be slow and right than fast and forgotten.

The companies such founders will build are, for the most part, not yet companies at all. They are experiments running tonight in unremarkable rooms, tended by people who have chosen the long road knowing precisely how long it is. Our task is to find them early, and to stand with them across the translation gap when the rest of the market has looked away.