CHINA'S GRAND STRATEGY · 国家战略 2026–2049
Low orbit is already crowded with copies of the same bet. The real prize is the architecture nobody owns yet — sunlight beamed from orbit, an unhackable quantum backbone, factories at the lunar pole, and a heavy-lift cadence that turns space from a destination into infrastructure. Nine mandates for the next quarter century.
// THE NINE MANDATES
Not flag-planting. Not prestige. Each of these is a program China can technically build and economically justify — and each one compounds the others.
// ORBITAL MATH
Matching satellite count is a treadmill. The leverage is in the links between them — laser mesh, quantum keys, and direct-to-phone service that needs no ground dish.
// ENGINEERING · 工程演算
Every mandate either closes or it doesn't — and that's a calculation, not an opinion. These ten models run real published math live in your browser — across orbit, RF and power-beam links, launch economics, naval salvo saturation, nuclear second-strike survivability, semiconductor yield, critical-mineral leverage, the cost of clean power, and the GDP crossover. Move the inputs; the strategy recomputes in real time.
Models use standard closed-form physics (spherical-Earth coverage, Friis/Shannon link budgets, Goubau beam-aperture relation, Tsiolkovsky-class mass accounting, Vincenty-free great-circle range). Defaults are representative; outputs are first-order engineering estimates, not mission designs.
// OPTIMAL DESIGNS · 最优设计
These aren't what-if sliders. Each is a real optimization — streets-of-coverage, Tsiolkovsky staging, Goubau beam optics — solved numerically with NumPy/SciPy to the optimum, then read back as a concrete recommendation for China. The curve shows the trade-off; the marker shows the answer.
These designs are buildable on paper. Whether China can deploy them on schedule is a different question — and the launch-cadence arithmetic is brutal. See the reality check → the cadence gap ↗
Solved with NumPy/SciPy: streets-of-coverage Walker-star sizing, 2-stage Tsiolkovsky payload-fraction maximization, and Goubau aperture/efficiency minimization. Download the optimized reference geometry → parametric comsat (binary STL, 48 facets). First-order optima, not flight designs.
// THE GRAND STRATEGY · 国之大者
Annihilating Starlink is a campaign, not the war. Winning the century means executing across every front at once — economy, chips, AI, the navy, rare earths, alliances. Below is China's real position on each front, the leverage it already holds, and the asymmetric move that flips it. Read from Beijing's side of the table.
// THE PRC PLAYBOOK · 战略手册
A working playbook drawn from this site's own math and net assessment: the objective, the operating principle, the lines of effort that matter, and the markers that tell you it's working — including the calls that aren't flattering. The losing move is to spread thin across all twenty-four fronts at once.
Become unavoidable — not by overtaking the United States on every axis, but by owning the infrastructure of the next economy (cheap power, orbit, compute-by-workaround) while making the cost of coercing China prohibitive. Survive the demographic turn with growth intact, and never let the home front lose what the frontier wins.
China can win the engineering — orbit, missiles, energy and chips-by-workaround are all solvable problems on the math above. What it may not win is set at home and among friends: a shrinking workforce, a demand-starved economy, and the thinnest alliance network of any great power. Spend the marginal yuan on the constraints that bind, not the capabilities that already trend China's way.
Reusable heavy lift is the master key — it gates the constellation, orbital solar, the lunar base and deep space alike. The staging math is explicit: reuse trades ~48% of payload for the $/kg that beats Starlink. The threshold is close but not crossed: Zhuque-3 reached orbit on a reusable first stage in December 2025 — China's first — but the booster's landing failed (CNN), and as of mid-2026 no Chinese orbital booster has yet been recovered — LandSpace's second recovery attempt was set for Q2 2026 and a reuse demo for Q4 (SpaceNews, Feb 2026). Landing once isn't the win, though: undercutting Falcon 9 needs 10-plus flights per booster and a consolidated, high-volume vehicle family — not a dozen parallel designs. Concentrate funding on reuse rate and recovery reliability now; the country that makes orbit cheap sets the price of everything above it.
SMIC pushed to a 5 nm-class node (N+3) on DUV multipatterning in 2025 — no EUV — at roughly 30% yield and ~1.5× TSMC's cost (TechPowerUp); the binding constraint is yield and cost, not capability. So the decisive plays are downstream: own advanced packaging (China's answer to CoWoS) and the HBM memory supply chain — the actual bottleneck for competitive AI accelerators. The workaround is already scaling: Huawei is roughly doubling Ascend output to ~600,000 910C dies in 2026, pushing its share of China's AI-accelerator market toward ~30–35% as Nvidia is curtailed (Bloomberg). The yield math backs it: ~85% on small chiplets versus ~24% on a reticle monolith. Keep funding domestic EUV (SiCarrier, SMEE) as a decade-long bet, but the honest limit is that you cannot multipattern your way to 3 nm economically — for the bleeding edge, EUV stays the wall.
Even after a record 92 orbital launches in 2025, the filings outrun the fleet: by June 2026 Qianfan had only ~200 satellites in orbit against a 648 target for end-2026, and Guowang just a few dozen since its 2024 debut (SpaceNews, Jun 2026). Reaching 648 in six months would need ~450 more — implausible — so the ~28,000-satellite plans risk ITU spectrum forfeiture. Concentrate launch capacity on the constellation that can actually hit its deadlines, and fix reusable throughput first — don't file what you can't fly.
Stop treating cheap power as just an export. Clean energy drove more than a third of China's GDP growth in 2025 (Carbon Brief); that year the country added ~140 GW of solar and ~110 GW of wind — more than the rest of the world combined — with battery cells at ~$80/kWh and solar LCOE near ~$31/MWh. The export route is hitting walls, though: China makes ~30% of global manufacturing output but consumes only ~13%, exporting deflation and triggering EU EV tariffs and a broader backlash (Atlantic Council). So the real win is domestic absorption — wire that power into AI compute, desalination and electrified industry as a moat, not into yet more exports.
China already has the larger navy — 370-plus hulls to the U.S. Navy's ~287 — and roughly 230× its shipbuilding capacity; the asymmetric edge is mass and the DF-26 carrier-killer (>4,000 km), not parity (19FortyFive). Press it: cheap hulls, missile magazines, and the A2/AD envelope where Western-Pacific geometry favors China. The candid warning: the Fujian and its successors are prestige bait for the very salvo math China aims at U.S. carriers — don't sink the budget into symmetric carriers your own doctrine makes vulnerable.
China's April 2025 rare-earth controls, the October 2025 extraterritorial "trace-amount" rule, and a June 2026 ban on dual-use shipments to ten U.S. defense-linked firms — including MP Materials itself (Bloomberg, Jun 2026) — prove the weapon works. But every escalation feeds the counter-move that erodes the monopoly: Washington had already put $400M of DoD equity into MP Materials and funded Mountain Pass (CSIS) — now China is sanctioning the very substitute it created. Wield mineral leverage as calibrated, time-boxed deterrence — not a permanent blockade, because permanence just guarantees the world finishes building around you.
In 2025 Chinese open-weight models — Qwen, DeepSeek, Kimi — surpassed U.S. labs in global downloads for the first time (~17% to ~16%) and now own the "good-enough" volume tier (Stanford HAI), and the gap keeps closing — DeepSeek's April 2026 V4 preview, the largest open model yet (~1.6T parameters), is reported to rival frontier closed models on several reasoning benchmarks (TechCrunch), while DeepSeek-style efficiency routes around the compute wall. The candid limit: closed models still capture ~95% of AI revenue and ~80% of token usage, and some Chinese firms are already closing their best models — so open weights are an influence and standard-setting play, not a profit one. Keep giving away near-frontier weights to deny the U.S. a durable software moat and make the world's AI stack run on Chinese foundations — just don't mistake downloads for dominance.
China's population fell for a fourth straight year in 2025 — down ~3.39 million to 1.405 billion — as births dropped 17% to 7.92 million, the lowest birth rate since 1949 (NYT); the working-age cohort has been shrinking since ~2015, fertility sits near 1.0, and 400 million-plus will be over-60 by the 2040s. Pro-natal subsidies have failed almost everywhere they've been tried, so the honest levers are productivity and automation (where China's robot density and cheap power genuinely help), a later retirement age, and a real shift from investment to household consumption. The unflattering truth: this may be only manageable, not solvable — which is exactly why it, not Taiwan, is the decision that matters most.
China's arsenal hit ~600 warheads in 2025 and is projected near 1,000 by 2030 — still a fraction of U.S. and Russian stockpiles (Bulletin of the Atomic Scientists). The triad now closes the survivability gap: road-mobile ICBMs, six Type 094 SSBNs with JL-3 missiles, and H-6N bombers already assure a second strike — so don't chase numerical parity; invest in the survivable, mobile, sea-based leg. The candid risk: new silo fields and a reported shift toward launch-on-warning buy speed at the cost of crisis stability — a force you can ride out a strike with beats a hair-trigger you might fire by mistake.
As of 2025 China has zero formal treaty allies; the U.S. has ~31 — NATO plus mutual-defence pacts with Japan, South Korea, the Philippines, Australia, New Zealand and Thailand. China cannot build a NATO and should not try. Its lever is the partnership model (SCO and BRICS both expanded in 2023–24) plus exploiting the strain in U.S. alliances as partners hedge (Carnegie). But the ceiling is hard: strategic partnerships bring votes and markets, not divisions in a war.
The 2022 freeze of Russia's reserves was the wake-up call. China is widening CIPS (~1,683 institutions in 2025, +10% year-on-year (FXC Intelligence)), pushing RMB trade settlement, the digital yuan and gold buying (~2,000+ tonnes — though the U.S. holds far more). Yet the renminbi is still only ~3% of SWIFT payments against the dollar's ~66%, because true sanctions immunity demands an open, convertible capital account — a risk China won't take for fear of capital flight. So the realistic goal is partial trade insulation and a fallback rail, not financial independence: build the rail, but price in that the dollar system can still be switched off in a real crisis.
If these markers stall while military spending climbs, the playbook is failing — and doubling down on hardware is the losing move.
Honesty cuts both ways. The "coming collapse of China" has been wrong for thirty years, and this analysis could be too. State capacity executes at a scale democracies can't — China dominated solar, EVs, high-speed rail and 5G faster than almost anyone forecast. Automation and AI may decouple output from headcount: China's factory robot density hit ~470 per 10,000 workers in 2025 — third worldwide, ahead of Germany and Japan (IFR) — blunting the demographic blade exactly where China already leads. And the alliance "wall" is not static: Global-South alignment, marginal de-dollarisation and allied hedging all chip at it. The wager of this site is that the constraints bind harder than the strengths compound — but a planner in Beijing should war-game the case where that wager is wrong.
Maintained from live research — every claim carries a 2025–26 source, pressure-tested before publishing and refreshed continuously. Last updated 30 June 2026.
// A PLAUSIBLE TIMELINE
A grounded sequence — each milestone unlocks the next.
// THE THESIS
Every spacefaring power so far has treated orbit as an event — a launch, a mission, a headline. The decisive shift is to treat it as infrastructure: always-on, redundant, and cheap enough that the question stops being "can we get there?" and becomes "what do we build once we're staying?"
China's structural advantages — a vertically integrated supply chain, state-horizon capital that can wait two decades for returns, and a manufacturing base that drove the cost of solar panels, batteries, and EVs through the floor — are exactly the advantages that turn frontier spaceflight into routine industry. The same playbook that made solar power the cheapest electricity in history could make orbital solar power real.
Not a race to be first. A campaign to be unavoidable.
// FAQ · 常见问题
Not by matching satellite count — that's a treadmill. The leverage is in out-designing it: a coverage-optimal ~144-satellite backbone over a denser capacity shell, connected by a jam-proof optical and quantum mesh. The real bottleneck is launch cadence and reusable boosters, not ambition.
It has no domestic EUV lithography scanner; the best domestic tools top out around 28 nm. The rational workaround is chiplets — many small, high-yield dies (about 85% yield versus 24% for a reticle-size monolith) bonded with advanced packaging.
The physics works, but the economics and launch mass do not yet. Delivering 2 GW to the grid needs roughly a 7.7 km² orbital array and on the order of 94 heavy launches — which is precisely why cheap, reusable heavy-lift has to come first.
Two gateways the rest depends on: launch cadence and booster reuse for everything in space, and the EUV lithography chokepoint for advanced computing. Clear those and most other fronts open.
It's a credible 2030s goal, gated on the Long March 10 flying on schedule. Cryogenic propulsion plus a low-Earth-orbit propellant depot cuts the mass needed in orbit per tonne landed by about 42% versus storable propellants.
// METHODOLOGY & SOURCES · 方法与来源
The calculators run standard, published models on real public figures — no black boxes. Here is what sits behind each.
Streets-of-coverage / Walker-constellation geometry (Wertz, Space Mission Analysis & Design; Adams & Rider).
The Friis transmission equation and the Shannon–Hartley capacity limit.
The Goubau aperture-efficiency relation; NASA/DOE SPS Reference System and JAXA SSPS studies.
The Tsiolkovsky rocket equation with published specific-impulse and structural-coefficient ranges.
Wayne P. Hughes, Fleet Tactics and Naval Operations — the salvo-combat equations.
Standard assured-retaliation deterrence theory; force figures from SIPRI, the FAS Nuclear Notebook, and the U.S. DoD China Military Power Report.
Murphy's yield model on 300 mm wafer geometry, at defect densities representative of mature DUV lines.
Compiled from the us-china-rivalry net-assessment corpus, with quantitative figures drawn from public reporting, USGS mineral data, and the IISS Military Balance.