Table of Contents
The gap between a promising lab result and a product people can buy is usually measured in years, not months. A material that works on a microscope slide may fail at 10,000 units a week. A drug that shrinks tumours in mice still needs three phases of human trials. Even software, which moves faster, has to clear security reviews, procurement and the simple question of whether anyone will pay. How new technologies move from research to market is less a single leap than a chain of small, unglamorous handoffs.
That chain has a name in policy circles: the valley of death. It sits between public research funding and private investment, and it kills projects that are scientifically sound but commercially unproven. The technologies that dominate headlines, from quantum computing to solid-state batteries, are exactly the ones that spend longest in this gap.
The valley of death is a proof problem before it is a money problem
Researchers are trained to reduce uncertainty about how nature works. Investors need uncertainty reduced about whether a product can be made, sold and supported. Those are different questions. A university lab can produce a gram of a new cathode material; a factory needs to produce tonnes with consistent particle size, low defect rates and a supply chain for every input. The first question is scientific. The second is operational, and it is usually harder.
Public programmes try to bridge the gap. In the US, Small Business Innovation Research grants move from roughly $300,000 in Phase I to around $2 million in Phase II. Europe’s innovation council and national proof-of-concept funds play similar roles. The money helps, but it does not answer the commercial question. Founders still have to find a buyer, a manufacturing route and a regulatory path.
Stage one: turn a discovery into a repeatable process
A result that works once in a carefully controlled experiment is not yet a technology. It becomes one when someone can repeat it on a bad day, with a different operator, using materials from a different supplier. That is why scale-up is such a filter. Processes that rely on a postdoc’s steady hands often collapse when handed to a contract manufacturer.
Scale-up is where elegant science meets messy reality
CRISPR gene editing was first demonstrated in 2012. The first approved CRISPR therapy, Casgevy, reached patients in 2023. Eleven years sounds slow until you list what had to happen: delivery methods, off-target effects, manufacturing, clinical trials in sickle-cell disease, and a reimbursement model for a one-time treatment with a multi-million-dollar price tag. Each step exposed a new constraint.
Battery startups face a similar grind. A coin cell that survives 500 cycles is not a car battery. Pouch cells must pass abuse tests, then modules, then packs. Every step changes the chemistry, the safety case and the economics. The lab record may survive; the original design often does not.
Stage two: find a customer before the product is finished
Many technical teams treat sales as something that happens after the product is ready. In practice, the first customer is part of the development process. They define the acceptable failure rate, the integration requirements and the price that makes the switch from the incumbent worthwhile. A pilot with a friendly corporate partner can be a useful reference, but it can also become a permanent science project. The signal that matters is a repeat order or a paid renewal, not a press release.
Financial markets offer a blunt test. An algorithm can look brilliant in a backtest and lose money once latency, slippage and market impact are included. AI trading systems only become products when they survive live capital, compliance reviews and the operational risk of a bad day. The market does not care how elegant the model is.
Pilots are not sales
A hospital may trial a diagnostic device for six months. A factory may run a predictive-maintenance sensor on one production line. Those pilots generate data, but they do not prove demand. The transition from pilot to purchase usually requires a budget line, a procurement process and someone willing to put their name on the decision. That internal champion matters as much as the technology.
Stage three: treat regulation as a design input
Regulation is often described as a hurdle at the end of development. It is more accurate to treat it as a design input from the start. Medical devices, food additives, chemicals and autonomous vehicles all have pre-market gates. The FDA’s 510(k) pathway can clear a moderate-risk device in months; a novel high-risk device may need a premarket approval application with clinical data and years of review. The difference shapes everything from trial design to the company’s cash runway.
AI policy adds another layer because the rules are still forming. South Africa’s AI policy approach is watched as a test case for how emerging economies balance innovation with data protection and public trust. Companies that engage early can help shape standards. Those that wait inherit compliance costs and may find their product category defined by someone else.
Stage four: manufacturing and unit economics decide who survives
A pilot line is not a factory. Pilot lines are allowed to be slow, manual and expensive. Factories have to run at yield, on schedule, with labour and energy costs that show up in every unit. This is where many well-funded hardware companies discover that their bill of materials was never designed for mass production.
Consumer hardware often succeeds by combining proven subsystems rather than inventing new physics. Sun Home’s Eclipse saunas, for example, package red-light and infrared emitters into a home format. The innovation is integration, safety and cost control, not a new law of thermodynamics. That is a common pattern in commercialisation: the winning product is often the first one that is reliable and affordable, not the first one demonstrated.
Unit economics are brutal at small volumes. A contract manufacturer may quote $200 per unit at 1,000 units and $45 at 100,000. The gap changes pricing, margins and the addressable market. Investors call this the cost curve. Founders call it the reason they need a second financing round.
The people who move technology across the gap
Commercialisation is a team sport with unfamiliar roles. The scientist who invented the technology is rarely the person who can run a factory or negotiate a licensing deal. The handoffs between those people are where projects stall.
- Technology transfer officers decide which inventions get patented, which get published and which get abandoned. They also negotiate licences and spin-out terms.
- Incubators and accelerators provide lab space, legal help and a first network of customers, usually in exchange for equity or fees.
- Corporate venture scouts bring pilot customers, manufacturing partners and a route into established supply chains.
- Regulatory consultants turn standards into test plans before the design is frozen, which is cheaper than redesigning later.
None of these roles guarantees success. They reduce the number of avoidable failures. A good tech transfer office can save a startup a year of patent mistakes; a bad licensing deal can hand away the core IP.
Adoption is uneven, and that is normal
Diffusion of technology rarely follows a smooth curve. The first buyers are not average consumers. They are hospitals, factories, developers, utilities and government agencies with a specific problem and a budget. They tolerate rough edges because the alternative is worse. Mainstream customers wait for lower prices, better support and proof that the product will still exist in three years.
The timeline varies widely by sector. Some technology trends change everyday life within two years; others take a decade to move from niche to normal. Electric vehicles followed a classic path: early adopters, tax incentives, charging infrastructure, then mass-market models. Generative AI moved from research demos to public products in months, but the enterprise adoption curve is still climbing because data governance and accuracy take time.
The market is a test, not a finish line
A technology has moved from research to market when a customer pays for it more than once, when the manufacturing line can repeat the result, and when the regulatory file is boring enough that no one talks about it. Those milestones are less glamorous than a breakthrough paper, but they are the ones that keep a product alive.
The next wave of technologies will follow the same route. Some will stall at the prototype. Some will find a first customer and never find a second. The ones that make it will have solved the unglamorous problems: repeatability, reimbursement, yield and support. That is the real work behind every product that looks inevitable in hindsight.
