Perseverance has been collecting rock cores in Jezero Crater since landing in February 2021, sealing them in titanium tubes about the size of a cigar. By September 2025 it had collected 27 rock cores, plus regolith and atmospheric samples, all cached for a future mission to retrieve. The question many people ask is reasonable: if the rover is already on Mars with sophisticated instruments, why go to the enormous trouble of bringing the rocks home?

The answer is precision. Rover instruments are remarkable, but they cannot match Earth laboratories, which can date rocks to within millions of years, detect organic molecules at parts-per-billion levels, and use instruments too large, heavy and power-hungry to fly. Some questions — did these rocks once hold microbial life, and exactly when did Mars's climate change — can only be answered by putting the samples under Earth's best microscopes, mass spectrometers and clean rooms.

The original Mars Sample Return architecture was extraordinarily complex: a lander would fetch the tubes, a small rocket would launch them from Mars — the first launch from another planet — an orbiter would catch them, and a return capsule would bring them to Earth. It was also extraordinarily expensive. Early estimates reached about 11 billion dollars, with a return date stretching to 2040, and an independent review judged the plan unsustainable in that form.

In 2024 NASA opened the design to new ideas, and in January 2025 it announced it would study two landing architectures in parallel: one using a sky-crane system similar to the one that landed Curiosity and Perseverance, and one using commercial capabilities. The revised cost range fell to roughly 6 to 7 billion dollars, with a return targeted for the 2030s rather than 2040. A final design decision was expected in the second half of 2026.

Step one is already done: Perseverance collects and seals the samples, and some tubes are deposited in a depot on the Martian surface as a backup. Step two is a sample retrieval lander that lands near Jezero Crater, picks up the tubes — either directly from the rover or from the depot — and loads them into a small ascent vehicle. Step three is the Mars Ascent Vehicle launching into orbit, carrying an orbiting sample container designed to hold about 30 tubes.

Step four is rendezvous: a European Space Agency Earth Return Orbiter captures the container in Mars orbit, seals it in a protective system, and flies it back to Earth. Step five is entry, descent and landing of the sample capsule in a desert recovery zone, followed by transfer to a specially built receiving facility. Every step has been done in some form before except the Mars launch and the orbital capture, which is why those two are the highest-risk parts of the whole campaign.

Several engineering changes cut cost and risk at once. The lander platform would use a radioisotope power system instead of large solar panels, making it resilient during Martian dust storms that can blanket the planet for weeks. Sample cleaning and handling steps were moved to the Martian surface rather than inside the spacecraft, simplifying the orbiter. And the parallel study of a commercial lander option opened the door to using existing industry hardware rather than building everything bespoke.

China's Tianwen-3 mission, planned for around 2028 with a two-launch architecture, adds competitive pressure and scientific interest: if it returns samples first, the comparison between two independent sample sets would be scientifically valuable in itself. For NASA, the revised plan's virtue is not just lower cost but earlier science — a 2030s return means researchers working today could still analyse the samples during their careers, rather than handing the project to the next generation.

Risk remains at every handoff. The retrieval lander must land precisely near the depot, the ascent vehicle must work first time with no possibility of repair, and the orbital rendezvous must succeed millions of kilometres from Earth with light delays of up to 20 minutes each way. Planetary protection adds another layer: the samples must be sealed so thoroughly that no Martian material can escape, and the receiving facility must handle them under strict containment until they are proven safe.

Budget politics is a separate risk. Sample return has already been reworked once, and a multi-billion-dollar programme spanning more than a decade must survive several funding cycles. Perseverance itself helps here: engineers have certified its wheels and arm actuators for an extended mission through at least 2031, including a single-sol driving record of 411.7 metres set in June 2025, so the samples' guardian is healthy. The rocks are waiting; the question now is whether the retrieval fleet will be funded and flown in time to collect them.