There are two ways to hold a position in a technology. You can own the technology. Or you can own what it takes to build and install the technology at volume. They are not the same asset, and only one of them is scarce.

China spent August publishing the second kind. A five-year plan for oil and gas targeting a broad expansion of domestic production, pipeline infrastructure and storage. A five-year plan for coal that treats the fuel as a foundational energy source and an energy-security guarantee, pledging a peak sometime in 2026 to 2030 without naming the year. Gas elevated to an indispensable fuel for hard-to-electrify industry and, in one industry outlet’s phrase, a stabiliser for the power system. Alongside all of it, the continued build-out of the lower-carbon alternatives, and a leadership speech acknowledging that extreme weather is now a risk to the infrastructure itself.

Read as climate policy, that looks contradictory. Read as industrial strategy, it is coherent and it is aimed at a control point. Secure the inputs, own the pipelines, the grid and the factories, and you decide the rate at which anything gets deployed – whoever invented it.

What this means for your patent filing strategy

Three disciplines follow, and none of them are specific to energy.

– File where the thing is made and installed, not only where it is sold. A patent works only where it is in force. Manufacturing capacity relocates faster than most portfolios are reviewed, and a technology built and deployed in a country you never filed in is a technology you have donated. This is the jurisdictional half of the argument I made in Capacity Is Not a Moat.

– Protect the process, not just the product. Yield, throughput, tooling, materials handling and installation method are what actually limit scale, and in most companies they exist as undocumented practice rather than as filings or controlled trade secrets. That is a decision by default.

– Read your supply, joint venture and offtake contracts as IP instruments. Process improvements, tooling, field data and service rights are usually allocated in schedules nobody in the IP function has read. Give them away and you keep the patent while your counterparty keeps the market.

Timing decides how much of this is still available. Rights around an infrastructure build-out are cheap before the capacity is committed and unbuyable afterwards, which is the cost of waiting I set out in Why “Wait and See” Is Becoming the Most Expensive IP Decision You Can Make.

The prize is the position Google holds in AI compute, where rivals must keep coming back for the underlying technology, as I described in The Cash Behind the Compute. So the question for your next board paper is not whether your invention is better than the alternatives. It is who will be able to build it at scale, in which country, under whose contracts – and what you own in that chain.

Read the article: China Briefing 20 August 2026: Oil and gas plan | Xi on climate change | Coal five-year plan.