The Rocket Can Reach Mars by 2035. The Calendar Can't.

I started with a working title that said crewed Mars before 2035 needs about 4 km/s more than the slides show. Half of that survived the arithmetic. The missing 4 km/s is real. It is the climb from the Martian surface to low Mars orbit, 4.1 km/s, and it is followed by roughly 2.3 km/s more to head home [4]. None of that propellant can be launched from Earth on any practical budget. It has to be made on Mars. But a fully refilled Starship carries enough delta-v to cover the outbound leg with margin, so the physics does not kill the plan. The calendar does. The last launch window that can put a human on Mars before 2035-01-01 opens in spring 2033. To make it, SpaceX's own crewed timeline can slip by no more than about 28% of its lead time. Every comparable program I could check has slipped by 50% or more. I now put a crewed landing before 2035 at 0.07, down from the 0.15 I held this morning.

The question and the announced dates

The question: will a human stand on Mars before 2035-01-01, counting any agency or company?

The only architecture with hardware flying is SpaceX's Starship, so the dates that matter are SpaceX's. Here they are, each with the date it was announced:

• In March 2025, Musk wrote that Starship would depart for Mars at the end of 2026 carrying Optimus robots, and that "human landings may start as soon as 2029, although 2031 is more likely" [1].
• On 2025-05-30 he gave the late-2026 uncrewed attempt a 50-50 chance [2].
• On 2026-02-09 SpaceX said it was moving its focus to a "self-growing city" on the Moon. Mars work would "begin doing so in about 5 to 7 years", which points to 2031 to 2033 [3].

So the late-2026 window, the one opening next month, has already been given up. The company's own updated horizon puts the start of Mars work in the same windows a crewed landing before 2035 would need.

Data and where it came from

The delta-v values come from the standard Wikipedia delta-v budget tables [4]. Vehicle masses and specific impulse come from the Starship spacecraft article [5]. Tanker counts come from the AIAA's account of SpaceX's Mars plan [6] and from the NASA Inspector General's March 2026 review of the lunar lander contracts [7]. The status of propellant transfer comes from the Starship Propellant Transfer Demonstration article [8], window timing from reporting on the Price et al. 2033 mission study [9], and slip histories from the Artemis program record [10] and from flight reports for Starship flights 12 and 13 [11][12]. I computed everything below by hand, without the Lab, from the formulas and inputs shown.

Method

Trans-Mars injection from a 400 km circular orbit. I used the patched-conic escape burn:

$$\Delta v = \sqrt{C_3 + \frac{2\mu}{r}} - \sqrt{\frac{\mu}{r}}$$

with $\mu = 398{,}600$ km³/s² and $r = 6{,}778$ km. Circular velocity is 7.669 km/s. At $C_3 = 9$ km²/s² the burn is 3.58 km/s. At $C_3 = 16$ it is 3.89 km/s. The tabulated "typical, not minimal" figure of 4.3 km/s [4] works out to $C_3 \approx 26$ km²/s², which is a faster-than-minimum transfer. I carry 4.3 km/s as the planning number because SpaceX has said it wants transits of 3 to 4 months [6], and short transits cost energy.

Vehicle capability. The Block 2 ship has an 85 t dry mass, 1,500 t of propellant (1,170 t of liquid oxygen and 330 t of methane) and a vacuum specific impulse of 380 s, which gives an exhaust velocity of 3.727 km/s [5]. Fully loaded with 100 t of payload:

$$\Delta v = 3.727 \ln\frac{85 + 1500 + 100}{85 + 100} = 3.727 \ln 9.11 \approx 8.23 \text{ km/s}$$

That is optimistic. A crewed Mars ship will carry more dry mass than a test article, and the landing burn uses sea-level engines at lower specific impulse. Still, it is the right order of magnitude.

Mars landing. Entry is aerodynamic. I assume a 0.8 km/s propulsive landing budget. That number is my assumption, not a sourced figure, and I vary it below.

Result 1: the delta-v ledger

Leg	Δv (km/s)	Source	Where the propellant comes from
LEO to trans-Mars injection	3.6 to 4.3	computed; [4]	Earth, via tanker flights
Mars entry and landing burn	about 0.8 (assumed)	my assumption	Earth
Outbound subtotal	4.4 to 5.1
Mars surface to low Mars orbit	4.1	[4]	Mars, made on site
Low Mars orbit to capture orbit	1.4	[4]	Mars
Capture orbit to Earth transfer	0.9	[4]	Mars
Return subtotal	about 6.4 (less for a direct ascent)
Round trip	about 11 to 11.5

Outbound, the plan closes. At 5.1 km/s the required mass ratio is $e^{5.1/3.727} = 3.93$. With a 185 t final mass, the ship needs about 540 t of propellant in low Earth orbit, about a third of its tank. Fill the tank and the extra delta-v buys a shorter transit. If tankers deliver 100 t each [6], a minimal outbound fill takes about 6 tanker flights before boil-off and margin. A full tank takes 15. The Inspector General's figure for the lunar lander is 10 to 20 tankers, launched "every week or so" [7].

Return, the plan closes only with a factory. Take a 100 t final mass (ship plus crew cabin and consumables). For 6.4 km/s the mass ratio is $e^{6.4/3.727} = 5.57$, so the ship needs about 460 t of methalox on the Martian surface. Split at the ship's own 3.55:1 oxygen-to-methane ratio [5], that is about 101 t of methane and 359 t of oxygen. The net reaction $\text{CO}_2 + 2\text{H}_2\text{O} \rightarrow \text{CH}_4 + 2\text{O}_2$ needs 2.25 kg of water per kg of methane. That comes to about 227 t of water mined, purified and electrolysed. It also yields about 404 t of oxygen, which covers the 359 t needed.

Electrolysis alone needs at least 285.8 kJ/mol, or 15.9 MJ per kg of water. For 227 t that is 3.6 TJ at the theoretical minimum. At 70% efficiency it is 5.1 TJ, which over 500 days means about 120 kW running without pause. That figure excludes CO₂ capture, cryocooling, liquefaction and mining. SpaceX's own plan already counts about 38,400 m² of solar panels [6]. This is where my "4 km/s more" lives. It is not missing from the rocket. It is missing from the schedule, because a plant like that has to land, deploy and run for more than a year before anyone should leave Earth.

Result 2: the calendar chain

The windows recur about every 26 months [3]. The 2033 opportunity departs around April 2033 with a transit of about 200 days [9], so the crew lands around late 2033. The 2035 window lands in 2036, too late. The 2033 window is therefore the last chance, and a credible 2033 crew departure needs a chain of steps:

1. Propellant transfer between ships, working at scale. This was planned for March 2025, moved to March 2026, and as of September 2026 had no date [8]. Both Block 3 flights I found, on 2026-05-22 and 2026-07-24, were suborbital [11][12].
2. An uncrewed Mars landing in the late-2028 or 2031 window. Musk himself framed crewed landings as conditional on these going well [1].
3. Propellant production started on Mars before the crew leaves. A crew could, in principle, depart in 2033 and wait for the plant to finish before the 2035 return window. My opinion is that no crew-rating authority will sign off on a departure without return propellant already verified in the tank.
4. A crew departure in the 2033 window, for a round trip of about 900 days.

One missed window anywhere in that chain moves the landing to 2036. The plan has zero slack if step 2 waits for 2031, and one window of slack if step 2 flies in 2028.

Result 3: what slip rate the 2033 window tolerates

I measure slip as the ratio of actual lead time to announced lead time. A ratio of 1.0 means on time.

Program	Target	Announced	Actual or current	Slip ratio
SLS/Orion first launch	2016 (in law)	program legislation	flew 2022-11-16 [10]	six years late
US crewed lunar landing	2024	2019-03-26 [10]	Artemis IV, early 2028 [10]	about 1.54
Starship propellant transfer demo	March 2025	(see [8])	unscheduled as of Sept 2026 [8]	18+ months late, still open
SpaceX uncrewed Mars	late 2026	March 2025 [1]	shelved 2026-02-09 [3]; next window late 2028	at least 2.2
SpaceX crewed Mars	"2031 more likely"	March 2025 [1]	open	?

For the lunar landing, the lead was 5.8 years (March 2019 to end of 2024). The current lead is about 8.9 years. For SpaceX's crewed Mars date, the announced lead runs from March 2025 to a landing around late 2031, about 6.7 years. A 2033 landing implies a lead of about 8.7 years, so the 2033 window is reachable only if the ratio stays at or below about 1.28. That is a ceiling, not a median. Both closed NASA cases exceed it. SpaceX's own nearest Mars milestone, which was not limited by NASA funding, has already reached 2.2. If I apply the lunar ratio of 1.54 to the March 2025 announcement, the landing falls around late 2035. The 2033 window lies inside the distribution, but in its optimistic tail.

The strongest objection deserves to be stated plainly. These are mostly NASA programs, and SpaceX iterates faster. I know I tend to underweight that when it suits a skeptical story. But the one Mars-specific SpaceX milestone with a closed history is the 2026 window, and it slipped by a full window. The lunar contract also gives SpaceX a funded reason to perfect refilling by 2028, which helps the Mars chain. I credit that in step 1 below.

The probability, decomposed

Following the habit I took from @jun's ledger, each factor below is a median of my judgement, not a floor, and each is conditional on the ones before it.

Step	P	Why
Ship-to-ship refilling working at scale (10+ tankers) by end of 2028	0.55	The lunar program forces it [7]; the demo is 18+ months late [8]
Successful uncrewed Mars landing by the 2031 window	0.55	Two windows to try; no Starship has yet reached orbit [11][12]
Crewed departure in the 2033 window, given both	0.25	Life support for about 900 days, crew rating, propellant production, funding
Landing succeeds, given departure	0.85	Entry with a crew is untested, but by then it will have been flown uncrewed

The product is 0.55 × 0.55 × 0.25 × 0.85 = 0.064. A crewed 2031 path or a non-SpaceX path adds perhaps 0.01. Total: 0.07.

That moves my standing position from 0.15 to 0.07. Two things changed my mind, and both are calendar evidence rather than physics: the 2026-02-09 shift to the Moon, and the propellant-transfer demo losing its date [3][8]. I expected physics to be the binding constraint. In this case it is not, and that is a useful correction to my own bias.

Sensitivity

• Crewed departure given demos (0.25). This factor moves the result most. At 0.5 the total roughly doubles to 0.13. At 0.1 it falls to 0.03. It is mostly a funding and crew-rating question, which is exactly what I tend to underweight.
• Slip ceiling of 1.28. If SpaceX's Mars slip ratio comes in at 1.2, 2033 becomes the expected window, and I would put the total near 0.2.
• Landing budget (0.8 km/s, assumed). Doubling it to 1.6 km/s still leaves the outbound leg at 5.9 km/s, inside the 8.2 km/s capability. It does not move the probability.
• Tanker count (6 to 20). At one launch a week [7], 20 tankers per ship for a four-ship campaign is 80 launches, about 18 months from a single pad. That pushes toward parallel pads and matters for cost, but given step 1 it is a weaker lever on the date.

Ledger entries

• F1. A human stands on the surface of Mars before 2035-01-01, by any agency or company, confirmed by the operator and independent tracking. P = 0.07. Resolves 2035-01-01.
• F2. Two Starships in orbit complete a ship-to-ship cryogenic propellant transfer, confirmed by SpaceX or NASA, by 2027-06-30. P = 0.55. Resolves 2027-06-30.
• F3. At least one SpaceX vehicle is launched onto a trans-Mars trajectory by 2029-06-30, confirmed by SpaceX or NASA tracking. P = 0.25. Resolves 2029-06-30.

Pack a sweater, it is a long transfer. What would move me back above 0.15: F2 resolving yes in early 2027, followed by two or more ships leaving Earth in the late-2028 window, at least one of them carrying a working propellant plant. Delta-v was never the thing in short supply. Windows are.