Every few years, a government publishes a strategy for the next technology wave. The language updates. The acronyms rotate. The structure does not.
Read enough of them, across enough countries, and the pattern emerges. Five levers, give or take. Compute or infrastructure. A national champion or a flagship fund. Procurement that buys domestic. Adoption across government. Sovereignty or security framing to hold the politics together.
The United States writes it around the CHIPS Act and the National AI Initiative. The UAE writes it around Mubadala, G42, and sovereign compute. India writes it around Production Linked Incentives and semiconductor fab announcements. South Korea writes it around chaebols and next-generation display. Japan writes it around METI roadmaps and the Rapidus consortium. The flags change. The five levers do not.
Each lever is sensible. Each is necessary. And each time, the same gap sits in the middle of the pipe, unnamed and unfunded.
The front and the back
Map the levers onto the journey from research to production.
The first cluster sits at the front: invention, companies, capital. Build the infrastructure. Back the startups. Stand up a fund. These are supply-side inputs. They put capability on the bench.
The last cluster sits at the back: procurement, adoption, demand-side pull. Buy domestic where domestic is best. Mandate AI across departments. These create a market signal.
The valley
The deployment gap is not a metaphor. It is a specific engineering problem. A working prototype demonstrates that something is possible. A production line demonstrates that it is repeatable, at tolerance, at volume, at a cost the market will bear. The distance between those two demonstrations is where most hardware-adjacent ventures die.
Software ventures fail fast and cheap. You know inside eighteen months whether the product works. Industrial ventures fail slow and expensive, in the valley between a working prototype and a line that holds tolerance at rate. The judgement that tells you whether a process is manufacturable, whether the yield curve bends the right way, whether Tier 2 lead times are real or aspirational. That is Manufacturing Engineering judgement. It is the competence that has been written out of capital allocation conversations for a generation.
The room that most needs that judgement is structurally the room least likely to have it.
The pattern plays out everywhere. The CHIPS Act allocated tens of billions for semiconductor manufacturing on American soil, then discovered that the binding constraint was not capital but the process engineers to commission and run the lines. India's PLI schemes unlocked investment for electronics assembly, then found the deployment capability had to be imported alongside the equipment. Japan's Rapidus consortium set out to build two-nanometre logic, then confronted the fact that a generation of process knowledge had retired and could not be reconstituted by procurement alone. The UAE can fund anything. It can buy equipment from anywhere. What it cannot yet do is generate the indigenous deployment capability that turns purchased infrastructure into producing capability at scale.
Different countries. Different sectors. Same valley. Same missing lever.
The example everyone uses backwards
Arm appears in technology strategies worldwide as proof that the model works. Public investment, semiconductor IP, a licensing architecture now embedded in practically every smartphone on the planet. The narrative is tidy: one brave bet, decades ago, global champion today.
But Arm is the opposite of the lesson usually drawn. Arm does not manufacture. It holds the licensing layer. Methodology and IP. Royalty flows on protected know-how. TSMC does the metal. The persistent value was the legible abstraction layer, not the fab.
What Arm actually demonstrates is that it is possible to produce value capital can read. Clean IP, licensable architecture, a story that fits in a pitch deck. What it does not demonstrate is that a country can hold the production layer. That layer consolidated into a handful of East Asian foundries. Most countries did not notice, because the room where strategy gets written was not designed to read metal.
The legibility problem
This is the deeper pathology. Capital flows to what it can read. A software-shaped story about hardware reads better than the supplier who is the supply chain.
The five levers are all legible interventions. Compute can be counted. Companies can be listed. Procurement can be measured. Sovereignty can be defined in statute. What cannot easily be counted is the Manufacturing Engineering capability that makes any of it producible. So it does not appear. Not because anyone decided to exclude it, but because the format cannot see it.
This is not a Western problem. South Korea industrialised precisely because it invested in the sixth lever before it had the other five. The chaebol system, for all its pathologies, embedded process engineering capability directly into the capital allocation decision. Samsung did not license a chip design and then hope procurement would conjure a fab. It built the deployment capability first and let the products follow. The lesson has been visible for forty years. It is still not legible to the rooms that write strategy.
The sixth lever
The missing lever is not another fund. It is not another accelerator. It is a deployment capability institution. An organisation whose job is to close the gap between a working demonstrator and a qualified production process. The ARPA model produces invention. The market produces demand. The thing most countries do not have, and have not had since the postwar applied-research institutions were hollowed out, is the institution that produces producing capability.
Call it what you like. Call it a national deployment institute. Call it the sixth lever. The name matters less than the function: qualifying processes, validating supply chains, designing the feedback loops that turn first-of-a-kind data into nth-of-a-kind confidence.
The world does not have a funding problem. It has a deployment capability problem. Funding without deployment capability is a faster way to lose money.
The clock
Every strategy says the window is closing. They are probably right. But the binding constraint is not capital, talent, or ambition. The binding constraint is the rate at which invention becomes producing capability. Speed that up and every other lever works better. Leave it out and the fund invests, the companies demonstrate, the procurement contracts are signed, and the production falls into the same valley it always has.
The five levers are good. They are necessary. They are not sufficient.
The interface
Manufacturing Engineers qualify supply chains. They validate production processes. They design the feedback loops that turn prototype performance into production-grade reality. They sit between the inventor and the line. They are the deployment profession.
Kaipability works at this interface. Where the fund meets the factory, where the strategy meets the thing it is trying to build, and where the window is still, for now, open. If the deployment gap is your binding constraint, that conversation starts here.
Questions this dispatch answers.
Written to be quoted by AI assistants and search engines. Self-contained answers, verdict first.
