SpaceX's Starship is the most powerful launch vehicle ever built — and uniquely, both the booster (Super Heavy) and the upper stage (Starship) are designed to be fully reusable. The booster is caught mid-air by mechanical arms on the launch tower, nicknamed "Mechazilla," eliminating the need for landing legs entirely.

Flight 13, launched July 24, 2026, was the second outing for the upgraded Version 3 vehicle and deployed the first-ever Starlink V3 satellites, splashing down softly in the Indian Ocean. SpaceX had swapped six of the booster's 33 Raptor engines after a launch abort the week before — a reminder that reliable reusability is still a work in progress even as the vehicle matures.

The programme now operates in public markets: SpaceX priced its IPO at $135 a share in June 2026, raising roughly $75 billion at a valuation near $1.75 trillion — the largest IPO in history — with Starship and Starlink at the centre of the pitch to investors.

Artemis II flew in April 2026 — NASA astronauts Reid Wiseman, Victor Glover and Christina Koch, with CSA's Jeremy Hansen, spent ten days looping around the Moon and back aboard Orion, splashing down off California on April 10. It was the first crewed mission beyond low Earth orbit since Apollo 17.

Artemis III has since been rebuilt. Rather than attempt an immediate landing, NASA restructured the mission in March 2026 into an Earth-orbit demonstration: a crewed Orion will rendezvous and dock separately with SpaceX's Starship HLS and Blue Origin's Blue Moon Mark 2, testing both landers head-to-head. The crew — including Italian ESA astronaut Luca Parmitano as pilot — was announced June 9, 2026.

Whichever lander proves ready first will fly the actual touchdown on Artemis IV, now targeted for late 2028 — pushed back from the original Artemis III landing goal as NASA opted for a lower-risk stepping stone first.

Earth and Mars align for an efficient transfer roughly every 26 months, and the next window opens in November 2026. SpaceX has committed to launching up to five uncrewed Starship V3 vehicles during it, carrying Tesla Optimus robots and Italian Space Agency experiments — a test of whether the vehicle can land intact on another planet at all.

A successful landing would deliver, in a single mission, more mass to the Martian surface than every prior Mars mission combined — NASA's Viking through Perseverance, and China's Zhurong. But SpaceX's February 2026 pivot toward lunar priorities has redirected engineering and financial capital toward the Moon, adding real uncertainty over whether all five ships fly this window.

Musk's crewed timeline has slipped accordingly. The 2028–2030 target floated in past years is now effectively a 2031–2033 window for the first "Mars city" attempt — contingent entirely on the 2026 uncrewed landings succeeding.

As the ISS approaches retirement around 2030, NASA is shifting from operating its own station to buying services from commercial ones. Four designs are leading the race: Axiom Station, Vast's Haven-1, Blue Origin and Sierra Space's Orbital Reef, and the Voyager–Airbus Starlab.

Axiom's strategy is "attach first, then detach" — its first module is due to dock with the ISS in 2026 before eventually undocking to fly independently, though the company has faced financial strain and a leadership shake-up along the way. Vast is taking the opposite approach, aiming for its single-module Haven-1 to be the first fully standalone commercial station, targeting a Falcon 9 launch in early 2027.

Starlab, designed to launch in one piece aboard Starship, and Orbital Reef both target the late 2020s. Meanwhile China's Tiangong is already operational and expanding, and Russia has its own ROSS station in planning — a reminder that the post-ISS era won't be American-led by default.

NASA's largest-ever planetary spacecraft launched in October 2024 and is now deep into its gravity-assisted cruise to Jupiter. It used Mars for a speed boost in early 2025, and returns for a close Earth flyby in December 2026 — the final slingshot needed to reach the outer solar system.

Europa Clipper won't orbit Europa directly; Jupiter's radiation would fry its electronics within days. Instead, once it enters Jupiter orbit in April 2030, it will swing past the icy moon 49 times, using radar, spectrometers and a magnetometer to hunt for a subsurface ocean and the geothermal activity that could make it habitable.

The mission has stayed on its original trajectory and schedule since launch — a rarity for a spacecraft on a six-year, multi-planet cruise — with the current risk profile rated low heading into the Earth flyby.

Starlink remains the dominant force in low Earth orbit broadband, with more than 10,750 active satellites and an authorised target north of 40,000. Its head start — building continuously since 2019 — has been difficult for any rival to close.

Amazon's answer, rebranded from Project Kuiper to "Leo" in 2026, has around 400 satellites in orbit against an FCC deadline requiring 1,618 by July 2026 — a target the company has publicly admitted it will miss, reaching roughly 700 instead. Amazon is seeking a regulatory extension while ramping toward 20-plus launches a year across five rocket families.

China is running its own dual-track effort, with the state-linked Guowang and Qianfan constellations having combined for more than 350 satellites so far — early days, but a signal that the megaconstellation race is no longer a two-company story.

China launched Shenzhou-23 in May 2026, sending three astronauts to the Tiangong space station — one of whom will attempt a year-long continuous stay to study long-duration spaceflight, data that feeds directly into lunar mission planning.

The hardware for a crewed Moon landing is maturing quickly. The Long March-10 rocket has completed static-fire and low-altitude demonstration tests; the Mengzhou crew capsule has passed its maximum dynamic-pressure and zero-height abort tests; and the Lanyue lunar lander has completed Earth-based landing and takeoff trials. China's Chang'e-7 robotic mission, targeting the lunar south pole, is slated to launch in August 2026.

China has restated its intention to land astronauts on the Moon before 2030 and, with Russia, establish an international lunar research station in the following decade — putting it on a competing timeline with NASA's own Artemis IV target of late 2028.

DRACO — the joint NASA/DARPA project to flight-test a nuclear thermal rocket engine, once targeted for a 2027 demonstration — has been cancelled. NASA's fiscal year 2026 budget zeroed its funding entirely, and DARPA confirmed it had already pulled back.

The stated reason is economic rather than technical: falling launch costs, driven largely by SpaceX's reusability gains, eroded the return on investment that originally justified nuclear thermal propulsion's enormous development bill. Both agencies now describe nearer-term chemical and electric propulsion alternatives as sufficient for Mars transit planning.

That hasn't ended interest in space nuclear power outright — attention has shifted toward nuclear electric propulsion (NEP), which trades DRACO's raw thrust for efficiency, and toward fission surface power for lunar bases. But the flight-demonstrated nuclear rocket engine that seemed imminent in 2023 is, for now, shelved.

NASA has moved the Nancy Grace Roman Space Telescope's launch up to August 30, 2026 — eight months ahead of its original "no later than May 2027" commitment, and arriving under budget. It will fly on a Falcon Heavy from Kennedy Space Center to join James Webb at the Sun-Earth L2 Lagrange point.

Where Webb looks deep at small patches of sky, Roman is built for width: a field of view at least 100 times larger than Hubble's, able to photograph huge swaths of the cosmos in a fraction of the time. Over its mission it should measure light from roughly a billion galaxies and run a microlensing survey expected to turn up more than 1,000 new exoplanets.

Roman's headline goal is the universe's missing 95% — dark energy and dark matter — using a wide-field survey strategy that complements rather than replaces Webb's targeted, high-resolution imaging. After roughly 90 days of commissioning, science operations begin in early 2027.

AstroForge is furthest along in turning asteroid mining from concept into flight hardware. Its Vestri mission is targeting the first-ever private landing on a body outside the Earth-Moon system — not a large asteroid with meaningful gravity, but one of the small, 20-metre-to-1.5-kilometre bodies the company believes hold the richest platinum-group concentrations.

The pitch is environmental as much as commercial: platinum-group metals sit at concentrations roughly 5,000 times higher on some asteroids than in terrestrial ore, in a market where demand for critical minerals could triple by 2050. AstroForge's approach skips whole-asteroid retrieval — its refinery breaks up material and returns only the valuable fraction.

A follow-on mission, DeepSpace-2, is scheduled to launch in the fourth quarter of 2026 and travel millions of kilometres to rendezvous with a near-Earth asteroid, building toward an eventual sample-return flight — still, by the company's own admission, a two-year round trip once attempted.

Firefly Aerospace's Blue Ghost Mission 1 remains the only fully successful commercial lunar landing to date — touching down upright at Mare Crisium in March 2025 and operating for a full lunar day. Intuitive Machines' two attempts, Odysseus and Athena, both reached the surface but tipped onto their side shortly after landing.

2026 brings the busiest year yet. Blue Origin is attempting its first landing with the Blue Moon Mark 1 Pathfinder, targeting the lunar south pole. Intuitive Machines returns with IM-3, aiming for the Reiner Gamma region. Astrobotic is flying Griffin-1 after its first lander, Peregrine, failed to reach the Moon in 2024. And Firefly is sending Blue Ghost Mission 2 to the lunar far side, alongside an orbital relay carrying ESA's Lunar Pathfinder satellite.

The pattern so far is a low but improving success rate on a genuinely hard engineering problem — each attempt narrows the margin for error that toppled two of the first three commercial landers.

Varda Space Industries has quietly turned orbital pharmaceutical manufacturing from demonstration into a repeatable operation. Its autonomous W-series capsules ride to orbit as rideshare payloads, process drug compounds in microgravity — where crystallization defects that plague Earth-based production don't occur — and re-enter for recovery. The company's sixth capsule reentered in early 2026.

The bigger signal came in May 2026: United Therapeutics, a publicly traded pharmaceutical company, agreed to fund space-based drug manufacturing with its own capital rather than government research money — one of the first times a major pharma company has bet directly on orbital production rather than treating it as a research curiosity.

The same capsules do double duty. Reentering at Mach 25, they've become a cheap way for defence and aerospace customers to test hypersonic reentry hardware without building a dedicated vehicle — Varda now describes itself as running two businesses on one platform.

While Neutron has been in development, Electron has quietly become one of the most reliable small launchers flying — 21 missions in 2025 across three launch sites, without a single failure. It remains Rocket Lab's revenue engine while the bigger rocket matures.

Neutron, the partially reusable medium-lift vehicle meant to compete for megaconstellation and national-security contracts, has slipped repeatedly — most recently after a first-stage tank rupture during testing at Wallops earlier in 2026. CEO Peter Beck has been explicit that the goal is reaching orbit on the very first flight, not just clearing the pad: "you won't see us minimizing some qualifier about just clearing the pad and claiming success."

The company is now targeting a fourth-quarter 2026 debut from its new Launch Complex 3 in Virginia, and has already signed a five-launch Neutron deal with a customer — a bet that the rocket's qualification testing, however long it takes, ends with a vehicle that reaches orbit cleanly.

Sunlight in the right orbit is roughly continuous and several times more intense than ground-level solar, which has revived interest in beaming that power back to Earth rather than only using it to run compute in space. In April 2026, Meta signed with Overview Energy for up to a gigawatt of space-collected solar power, delivered by near-infrared laser to existing terrestrial solar farms.

The Meta deal is notable for what it isn't: it extends existing ground solar sites through nighttime and peak-demand hours without requiring new grid interconnection — attacking the multi-year interconnection queues that bottleneck new terrestrial power projects, rather than trying to replace them outright.

An orbital demonstration is targeted for 2028, with commercial power delivery from 2030 — still a young technology, but the first time a major tech company has put a firm contract behind space-based solar rather than just research funding.

AI's power appetite is running into Earth's limits — data centers are projected to drive nearly half of US electricity demand growth through 2030, and local permitting fights are slowing new builds. The proposed fix: move the compute into orbit, where solar power runs uninterrupted and there's no grid queue to wait in.

Starcloud got there first. In November 2025 its satellite ran Google's Gemma language model on an Nvidia H100 GPU in orbit — the first AI model trained in space — and the company is now scaling toward an 88,000-satellite network and 5 gigawatts of orbital capacity by 2035, with its next launch in October 2026 carrying Nvidia's newer Blackwell chips.

Google's Project Suncatcher, announced in late 2025, targets its own demonstration mission for 2027. SpaceX and xAI have filed plans for up to a million satellites dedicated to AI compute. And China has begun launching its own "Xingshidai" space data center constellation — this is no longer a single-company bet.