- Installed capacity
- Recent additions
- Generation mix
- Forward target
Sources
The Missing Grid · Part I
“To talk about energy and the economy is a tautology: every economic activity is fundamentally nothing but a conversion of one kind of energy to another, and monies are just a convenient (and often rather unrepresentative) proxy for valuing the energy flows.”
— Vaclav Smil, Energy and Civilization (2017)
In June 2025, the State Grid Corporation of China commissioned the Hami-Chongqing transmission line: a 2,260 km long ultra-high-voltage line that carries thirty-six billion kilowatt-hours per year from the Xinjiang deserts to the Yangtze manufacturing belt. It is the 45th of its kind to enter service with lines 46 to 48 under construction. In the same 12 months, the Electric Reliability Council of Texas (ERCOT) curtailed more than 8 terawatt-hours (TWh) of wind and solar generation because the surplus could not be transported from West Texas to demand centers further east. Despite active buyer interest and the turbines spinning, the wires didn’t exist to transmit the energy.
Both events are visible expressions of a single global question: whoever can move bulk electricity at scale across long distances will set the terms of the next industrial cycle. China understood this two decades ago and committed seven hundred billion dollars to the problem (alongside an incredibly “favorable” political climate to simply build in). Now, every other major economy is in a position where catching up is simply the only available option.
We aim to dig deeper into the energy race and the supply chain powering it. China gets significant attention because the Chinese buildout is the framework everyone must follow or at least compete with. The US also receives sustained focus because the consequences for the AI buildout are most immediate and the political environment most nuanced. But the question is global. Whoever builds and controls the supply chain shapes the geography of industrial advantage for the next half-century.
While generation is by no means a solved problem and more generation capacity is essential to scale, we believe these problems are being addressed (that too very publicly and aggressively):
Hover or tap a country for its electricity generation profile.
Sources
The map’s most instructive cases aren’t the giants. Norway runs almost entirely on hydropower and exports the surplus to Europe. Iceland runs on hydropower and geothermal and routes eighty percent of its generation straight to heavy industry, a working template for how dispatchable power gets allocated to industrial load. Saudi Arabia, the UAE, and Qatar, all net hydrocarbon exporters, are building gas-and-solar capacity in transactional terms: every megawatt-hour of domestic burn displaced is a barrel freed for export. The buildout is happening across radically different fuel mixes, geographies, and political logics.
The fuel and technology behind the buildout are widening too. The USGS added 28.3 trillion cubic feet of recoverable Permian gas to American reserves in January 2026, on top of the roughly 250 trillion the Marcellus still holds. Small modular reactors signed their first hyperscaler power purchase agreements in 2024 and 2025.
Generation was the binding constraint, and the market has tackled it systematically. We believe that in the rush, the harder problem of moving the electrons got overlooked.
The international buildout is dramatically uneven. China has built two hundred and sixty gigawatts of large-scale interregional transmission since 2014, Europe forty-four, and North America seven. China has committed eighty-three billion dollars per year to grid expansion through 2030, with forty-five percent of that earmarked for HVDC. India has built nine thousand circuit-kilometers of its planned Green Energy Corridor and has another ten thousand approved. European transmission system operators are locking in framework agreements that pre-allocate the next decade of HVDC cable manufacturing capacity.
The United States is moving in the opposite direction. Lawrence Berkeley National Lab tracks an American interconnection queue of 1,400 gigawatts of generation and 890 gigawatts of storage; the historical completion rate is thirteen percent and the median wait is four years. The share of new American transmission projects with interregional reach has fallen from six percent to four. NERC’s January 2026 reliability assessment classifies thirteen of twenty-three regional grids as high or elevated risk over the next five years.
There are two rational responses to a transmission shortage.
The first is to stop trying to move electricity and bring the load to the fuel. Pennsylvania being a great example: the PPL reports more than sixty gigawatts of data center interest in its service territory alone with Talen, Constellation, and Blackstone signing multibillion-dollar deals to build combined-cycle gas plants on top of the Marcellus and hyperscale AI campuses behind the meter. Shippingport, the site of the first commercial nuclear reactor in 1957, will host a $3.2 billion gas plant feeding a single AI data center campus. Casey Handmer, the engineer who has argued for years that batteries and on-site generation will out-compete grid-scale transmission, predicted this exact arrangement in January 2024.
Similarly, Saudi Arabia is using cheap gas to colocate AI capacity in NEOM. The Gulf states are courting hyperscalers with promises of behind-the-meter power. China runs Inner Mongolia coal plants connected directly to local server farms.
However, we believe that this approach works across a small set of isolated projects but would not scale to a country. The Marcellus, the Permian, and the Haynesville together hold the majority of recoverable American natural gas but cover a small fraction of the country by area. The argument for collocation is geographic but so is the rebuttal since generation sources are extremely distributed. In the US, solar is best recoverable in the Southwest, wind in the high plains, hydroelectric flows in the Pacific Northwest and Quebec but the load grows in Atlanta, Phoenix and northern Virginia. The same logic applies internationally with India where renewable resources are concentrated in Rajasthan and Gujarat but the load concentrates in the eastern and southern coastal cities. China’s generation resources sit west of the Hu Line (the demographic boundary drawn in 1935 along which ninety-four percent of the population lives on forty-three percent of the land) and hence transmits most of their energy across. Europe has similar dynamics, the crucial argument being: the fuel and the load have always been in different places, the function of transmission is to make their geographies converge.
The second response is to build transmission infrastructure: we believe the best path for that to be high-voltage direct current. Alternating current, the dominant mode for the last century, loses energy over long distances, cannot connect grids running at different frequencies, and cannot run undersea past roughly fifty kilometers because the cable behaves as a continuous capacitor. HVDC has none of these limitations. The break-even distance against AC is about three hundred and fifty miles overhead and fifty kilometers undersea. Above those thresholds HVDC is cheaper per megawatt-mile, requires a fraction of the right-of-way at equivalent power transfer, uses less metal per unit of capacity, and can be controlled directly by the grid operator rather than flowing along the path of least impedance.
Right-of-way for 6 GW of long-distance transmission
China operates roughly three hundred gigawatts of HVDC across forty-five commissioned ultra-high-voltage lines covering 52,300 kilometers. State Grid figured out in 2004 that the only way to industrialize the country at scale was to build wires across the Hu Line. Within a year the National Development and Reform Commission had approved a national plan combining UHV DC for bulk transfer with UHV AC for distribution. State Grid placed two thousand engineers on the project and funded research at three hundred Chinese universities. The 2026 commissioning of the Gansu-Zhejiang line will introduce the world’s first 800 kV voltage-source-converter HVDC system, eight gigawatts of capacity moving thirty-six billion kilowatt-hours per year from the Tibetan Plateau toward the Pacific coast.
India is planning nine new corridors totaling thirty-three gigawatts with its longest commissioned link, the 1,838-kilometer Raigarh-Pugalur line, moving power from Chhattisgarh (landlocked Central India) to Tamil Nadu (South India). Europe has wired the North Sea with subsea HVDC interconnectors that delivered twelve percent of UK electricity demand in 2024. Brazil’s Belo Monte 800 kV line carries four gigawatts of Amazon hydropower 2,543 kilometers to São Paulo. Australia’s Sun Cable, twice bankrupted and twice resurrected, will if completed move solar power 4,300 kilometers under the seabed from the Northern Territory to Singapore. Morocco’s Xlinks project will move desert solar 4,000 kilometers under the seabed to the United Kingdom.
The United States operates roughly eleven gigawatts of HVDC, most of it built between 1970 and 1986, and the country has commissioned only one major project since: SunZia, which took seventeen years from approval to energization. The 1990s unbundling of generation from transmission removed the regulated utility’s incentive to roll long-haul wires into its rate base, and siting authority is now distributed across fifty state utility commissions, FERC, and the regional transmission organizations in a way that any large project survives only by outlasting its opponents. Grain Belt Express needed ten years just for a Missouri certificate, Clean Line Energy collapsed after a decade of state-by-state battles, and SOO Green ended up buried along a freight rail right-of-way because the rail operator refused overhead access.
Hover or tap any project for details.
The constraints are political, not technical, and the politics are changing. Capital is sitting in infrastructure funds, sovereign pools, and hyperscaler balance sheets with nowhere to deploy at scale, while the firms that design and supply Western HVDC equipment are either American or operating American factories. What changed is the demand: AI training clusters that need gigawatts in places the existing grid cannot reach, manufacturing reshoring pulling industrial load into states without the local generation to support it, and chip fabrication and defense industrial capacity that the federal government has explicit reasons to power. Build incentives that did not exist a decade ago now sit on both sides of the aisle, and the cheapest path to a national AI buildout, a reindustrialized supply chain, and energy leadership runs through HVDC corridors the country has not yet decided to build. State Grid takes a project from approval to commissioning in two or three years, POWERGRID does the same in India, and European TSOs do it under EU cross-border rules. The engineering is not the bottleneck. The decision is.
China’s twenty UHVDC lines delivered 705 terawatt-hours in 2024, of which fifty-seven percent came from renewables, almost all of it hydropower. Wind and solar averaged only about a fifth of UHV-delivered power, and the corridors designed for one hundred percent variable renewables (Qinghai-Henan, in particular) historically delivered less than half their rated annual volume until distributed synchronous condensers were installed at the sending end. China designed the program in 2014 to move emissions from Beijing’s lungs to Xinjiang’s, not to decarbonize, and the N-1 reliability criterion still forces eastern provinces to build local backup regardless of imported power. China’s grid is not yet a clean grid, but it exists, and that distinction matters: a country can retrofit a network it has built, but no one decarbonizes a network they have not. Casey Handmer’s argument that batteries shift power through time more cheaply than transmission shifts it through space is correct on its own terms but answers a different question. Only wires move gigawatts across a continent.
The next four parts of the series make the case in detail and trace the path forward.
Part II takes up the technology and its history. It traces why direct current lost to alternating current in the 1890s, what brought it back, and why offshore wind and cross-continental subsea cables cannot use AC at all.
Parts III and IV map the global supply chain end to end. At the top sit three converter station builders (Hitachi Energy, Siemens Energy, GE Vernova) that supply essentially every Western project. Below them sit the seven largest HVDC cable manufacturers, concentrated in Italy, Denmark, Japan, Korea, France, and increasingly China. Further down still are the transformer steel mills and copper foundries that feed every cable factory, and the small fleet of specialized cable-laying vessels that installs every subsea project on earth. Power transformer lead times have doubled since 2021 and now exceed four years; subsea cable lead times exceed five. Hitachi Energy alone holds a thirty-billion-dollar order backlog, tripled since 2020, and GE Vernova reports that Q1 2026 electrification orders from data centers alone exceeded its full-year 2025 totals. On the demand side, the European market is settling into framework agreements that pre-allocate the next decade of capacity, with TenneT reserving seven thousand kilometers of cable and RTE locking five suppliers through 2028. The series traces the chain from raw materials through manufacturing to the TSOs, merchant developers, and hyperscaler customers placing orders against it, and identifies where capacity can plausibly grow on each layer.
Part V works through policy, the China question, and capital. China is the most instructive case in the world, vertically integrated and fast-permitting, and now extending Chinese hardware, Chinese technical standards, and Chinese financing into the developing world through what State Grid has rebranded the Global Energy Interconnection. Consequently, each of the major markets is attempting to shorten the gap: Europe is procuring through multi-year framework agreements; India has thirty-three gigawatts of new HVDC corridors planned with POWERGRID committed to most of them; the United States has FERC Order 1920 as a foundation, the BIG WIRES and SITE Acts still alive in Congress, and the DOE’s Transmission Facilitation Program disbursing under the Bipartisan Infrastructure Law. Permitting reform is the highest-leverage fix in each of these markets, and the political climate for it is the most favorable it has been in years. The series draws on field reporting with the operators, developers, regulators, equipment makers, and capital allocators building this market across the United States, Europe, India, and China. It names the specific policy moves that would shift the needle in each market and traces where the capital is going, from sovereign and hyperscaler balance sheets through merchant developers and European framework holders to the small but growing class of dedicated transmission infrastructure funds.
We believe this is the most consequential infrastructure buildout of the next decade. This overview reflects conversations with operators, engineers, regulators, and capital allocators thinking about it, and we would welcome more. If this space excites you, reach us here.
Headline statistics: