Raptor 4 Just Left Raptor 3 Behind... Starship's Historic Era Is Coming!

An adult elephant weighs around 5 tons, and a super-heavy booster has enough power to lift the weight of roughly 500 of them off the ground. But not with today's Raptor 3 configuration. I'm talking about Raptor 4. It won't just be more powerful. It'll inherit everything SpaceX is learning right now from the ignition and restart problems exposed during the first two flights of the new Starship configuration. And if Flight 14 proves those fixes work on Raptor 3, Raptor 4 could take that reliability even further. So how does Raptor 4 beat Raptor 3? Let's dig in. On the last day of 2024, someone on X asked Elon Musk how much thrust Starship actually had. His answer ran four lines. Once Raptor reaches 300 tons of thrust at liftoff, which Raptor 3 can probably do, and which Raptor 4 certainly will, the booster puts out 10,000 metric tons at the pad. 22 million pounds of force. Almost exactly three times Saturn 5. That was the first time the engine had a name and a number attached to it. And it came before Raptor 3 had flown a single mission. Then, in November 2025, he said something stranger. Raptor 3, he wrote, would probably be two to four times better than Merlin in dollars per ton of thrust. Raptor 4 should beat Merlin by more than 10 times in dollars per ton of thrust, with further improvement in thrust-to-weight ratio and specific impulse. Read that order again. Cost first, then thrust-to-weight, then efficiency. Raw thrust wasn't even the headline. And in February 2026, the last piece landed. Musk named Raptor 4 as the engine of choice for Starship Block 4. Not a concept. Not a fan nickname. A specific engine assigned to a specific vehicle. So, Raptor 4 is real. But notice what he measures it against. Not the F-1. Not the RS-25. Not the RD-180. He measures it against Merlin. The small, cheap, unglamorous engine SpaceX has cranked out by the thousand for Falcon 9. Nobody makes documentaries about Merlin. That's the point. You don't benchmark your next product against the thing you admire. You benchmark it against the thing you want to become. Remember that. It's where this whole story ends up. Now, the honest part. There's no public Raptor 4 hardware. No announced test fire. No date. Anyone quoting you an exact thrust figure is guessing. But there's a pattern. Raptor 1 came in around 185 tons. Raptor 2, around 230. Raptor 3, around 280. Every one of those numbers was said out loud before the engine existed. And everyone landed close. Musk's thrust targets tend to hold. His schedules almost never do. When he gives you a thrust number, take it seriously. When he gives you a date, add time. Which leaves the real question. Musk has said, Raptor is getting close to the limit of known physics on thrust. And that's not a figure of speech. Raptor already runs at the highest chamber pressure of any flight engine ever built. The thing standing in the way of more isn't thermodynamics. It's metal. There's a ceiling on what any known alloy will survive, while hot. Oxygen-rich gas is screaming through it at those pressures, and Raptor is already pressed up against it. So, if the power curve is flattening out, what is Raptor 4 actually for? The last two flights answer that. May 22nd, 2026. Flight 12. First flight of Starship V3. First flight of Raptor 3. All 33 engines lit. Then, a minute and 42 seconds in, one shut down. Not fatal. Super heavy is built to lose engines. The trouble came after separation. Booster 19 was supposed to flip, relight, and fly itself back toward the gulf. Telemetry showed an erratic engine startup. It couldn't relight the number it needed. The burn was cut short, and the booster went into the water hard. The FAA called it a mishap and grounded the program. Then, July 16th. The countdown for flight 13 gets scrubbed, and here's the reason. Four Raptors failed to ignite during startup. Not in flight. On the pad. Bolted to the tower. Plugged into every ground system on Earth, with engineers watching every sensor. And four of them just wouldn't light. July 24th. Flight 13 flies. And SpaceX had done homework. They rewrote the startup sequence specifically to be more tolerant of timing variability in engine startup. Their language, not mine. They were telling us where the problem lived. It worked. Super Heavy completed the high thrust portion of its boostback burn on all 33 engines. First time a V3 booster ever pulled that off. Real progress. Eight weeks after a failure. Then the burn ended early. And on the landing burn, only 10 of 13 engines relit. Five of those quit within seconds. By the time the booster hit the gulf, roughly five were still firing out of 13. An engine that won't light, an engine that lights and then quits, and a burn that terminates early may not share the same root cause. But operationally, they expose the same weakness. Raptor 3's startup and restart behavior still isn't as repeatable as SpaceX needs it to be. Before we go further, Raptor 3 deserves a fair hearing. Because the easy version of this story is wrong. Raptor 2 caught Super Heavy on the tower three separate times. Raptor 3 is two flights in and hasn't landed one. Looks like a step backward. It isn't. Raptor 3 never inherited Raptor 2's job. SpaceX changed the engine, the vehicle plumbing, the startup sequence, the boostback profile, the thrust level, and the thermal environment all at once. Raptor 2 flew a mature system. Raptor 3 is a new propulsion system being debugged in public. At full scale. In real time. Different tests entirely. And that debugging is exactly where Raptor 4 comes from. SpaceX is working startup timing, propellant conditioning, ignition sequencing, and engine-to-engine variability on Raptor 3 right now. You could watch it happen between flight 12 and flight 13, when they reworked the startup sequence and then got all 33 engines to relight for boostback. So the honest framing isn't that Raptor 4 arrives with some magic new igniter. It's that everything making Raptor 3 more reliable becomes Raptor 4's starting point. Which is why the stakes are so much higher than a lost booster. On Earth, a failed landing burn costs you a rocket and a bad afternoon. SpaceX planned to expend that one anyway. Run the same failure on Mars, and the engine that won't relight is the engine that was supposed to bring the crew home. There's no tower. There's no next flight. There's no help. So what does Raptor 4 actually have to deliver? Start with the thing Musk himself flagged. He used a term most people skipped right past. Induced mass. That's the mass an engine design forces the rocket to carry that isn't the engine itself. Pipes. Shielding. Interfaces. Brackets. Everything the engine drags along behind it. Weigh the engines alone, and Raptor 2 sits around 1630 kilograms against Raptor 3 at about 1525. Roughly 100 kilos. Fine. Not thrilling. Weigh the full installation. Engine plus everything the vehicle carries for it. And Raptor 2 runs about 2075 kilograms. Raptor 3 comes in near 1120. Over 100 kilograms saved per engine. Multiply by 33 and you've stripped about 38 metric tons off one booster. That's not engine dieting. That's SpaceX deleting the tax the engine charges the rocket. And Musk's own comment was that Raptor 3's induced mass, while far better than Raptor 2, still has a lot of room to improve. Read that as a direction of travel. So, restart better. Weigh the vehicle down less. Cost less to build. Why all three and why now? Because of what's waiting for it. Block 4 is a bigger stack running 42 engines total. And that's 42 across the whole vehicle. Not on the booster. 33 below. With the ship growing from six toward nine. A taller ship. More vacuum Raptors. More payload. And a mission list that eventually includes refueling in orbit and landing on another planet. You don't hand that vehicle an engine you have to babysit. Which brings us to the comparison everybody reaches for and gets backwards. The F-1 on Saturn F-5 produced about 680 metric tons of thrust. More than double a 300 ton Raptor. On raw power, one F-1 beats one Raptor. And it isn't close. Now ask a different question. How many times did an F-1 have to light? Once. One ignition. One burn. Then it dropped into the Atlantic and its working life was over. Single shot igniter. It was never designed to start a second time. Because nobody ever expected it to. A Raptor doesn't get that luxury. Booster. One flight. Lift off. Boost back. Landing burn. The ship on a Mars mission has to light for ascent. Again to reach its trajectory. Again to come down on Mars. Again to leave Mars months later. And again to land back on Earth. An F-1 only had to work once, fresh out of the shop. A Mars-bound Raptor has to restart again and again. After hours, days, and eventually months between burns. With nobody around to help. That's a different problem. And it explains a decision SpaceX made years ago that most people file under trivia. Why methane? The usual answer is soot. Kerosene cokes up in engines' insides. For something you throw away, who cares? For hardware you want to fly a hundred times, that's a cleaning job after every flight. Methane burns clean. The second answer is density. Hydrogen is more efficient, but it's so light and so cold that tanks balloon. Insulation gets absurd, and you can't store it for months. Methane is roughly six times denser and fits in a 9-meter tank. The third answer is the real one. The Martian atmosphere is about 95% carbon dioxide. And there's water ice under the surface. Run those together, and you can manufacture methane and oxygen on site. The chemistry is Victorian. The Sabatier reaction dates to the 1890s. What you need is CO2, ice, and power. You cannot refine kerosene on Mars. You cannot hold liquid hydrogen there through a launch window. SpaceX didn't pick methane for the trip out. They picked it for the trip back. That opens a question nobody puts on a thrust chart. Fuel made by a plant on Mars won't come off a certified production line. Purity drifts, temperature drifts, contaminants tag along. So either that plant has to deliver extraordinarily consistent propellant, or the engine has to tolerate more variation than today's Raptors do. One of those two has to give. And SpaceX isn't the only one working the Mars problem. This February, NASA fired a lithium-fed magnetoplasma-dynamic thruster at JPL at 120 kilowatts. Over 25 times the power of the thrusters flying on Psyche, under a program aimed at megawatt-class nuclear electric propulsion for crewed Mars missions. But look at what it does and doesn't do. Electric propulsion is a gentle, endless push. It's superb for crossing between planets, and completely useless for landing on one or leaving it. Even in a future where a reactor carries the crew across, something still has to fire hard on the surface of Mars to bring them home. That job stays chemical. Nuclear electric propulsion doesn't retire the restart problem. It concentrates it into the exact moments where failure is unsurvivable. Which is a good moment to ask why Raptor is such a strange, rare machine in the first place. Full-flow staged combustion isn't a new idea. The Soviets chased it with the RD-270 in the 60s, and it never flew. American engineers studied it and walked away. 60 years in textbooks, and exactly one company has ever put one in the air. The wall everybody hit was the oxygen-rich pre-burner. Hot, high-pressure oxygen stops behaving like anything you'd call air. It attacks metal. It will set fire to the engine containing it. Getting past that took custom superalloys and years of hardware failing on test stands. Which is why nobody copies Raptor from a blueprint. You can steal the drawing. You can't steal the test history. When nothing seems capable of overshadowing Starship launches anymore, SpaceX decided to outdo itself anyway. Announcing a new Starship variant that will launch from the biggest rocket launch facility on Earth. No exaggeration here. Elon Musk recently tweeted, Will enable over 2 million tons per year to orbit with Starship 4. A crucial step towards becoming a Kardashev 2 level civilization. This was a reply to an earlier tweet quoting his own post about Starbase Louisiana. Starbase Louisiana will ultimately have over a dozen launch towers. Enabling more than 30 Starship flights per day and making it the biggest launch site on Earth. What this signals is easy to predict. A massive new launch site is coming. And the next, bigger, more powerful generation of Starship will make its very first flight from there. Genuinely thrilling stuff. Well, let's start with the ground. Because everything else in this story is built on top of it. Literally. SpaceX has just unveiled Starbase Louisiana. A spaceport planned near Pecan Island in Vermilion Parish. Carrying a price tag north of 100 billion dollars. Musk described the finished site as having more than a dozen launch towers. Eventually enabling over 30 Starship flights per day. If that number holds, it would make this single facility the busiest launch site on the planet. Full stop. The scale is hard to overstate. We're talking about roughly 125,000 to 130,000 acres of coastal land. More than 200 square miles. Organized into five separate launch complexes. Each with two pads. For 10 pads total to start. SpaceX's own description of the site reads less like a rocket pad and more like a small industrial city. On-site propellant production. Dedicated power generation. Deep water shipping capability for moving hardware in and out. Vehicle processing facilities. And reportedly, even its own airport. Construction is targeted to begin in 2027. With the first launch penciled in for as early as 2029. Why Louisiana, of all places? The answer is geography. Elon just tweet, It's a great location for a wide range of azimuths. Louisiana's Gulf Coast lets Starship launch southward. Approximately 189 degrees azimuth. Across open water towards sun-synchronous orbits. Approximately 98 degrees inclination. Impossible cleanly from Florida, where southward paths would overfly Cuba. This unobstructed corridor supports high-cadence launches and reaches the sun-stable orbits needed for solar-powered orbital data centers and next-gen Starlink satellites. Now for the vehicle that's supposed to fill all those pads. Starship version 4, sometimes called Block 4, and in a recent tweet Elon Musk just wrote that Starship 4 is the next major evolution after the V3, currently flying test missions in 2026. And evolution undersells it a little. This is a genuine step up in almost every dimension that matters. Start with height. The current Block 3 stack stands at roughly 124.4 meters. Version 4 is expected to stretch to around 142 meters, with some configurations pushing close to 150. Picture three Statues of Liberty stacked directly on top of each other. That's roughly the scale we're discussing. None of Musk's tweets mention any increase in the rocket's diameter, so this growth is entirely about length, not girth. A longer upper stage designed to carry more propellant and greater payload volume. Then there's the engine count. Elon Musk tweeted last year, Starship V4 will have 42 engines when three more Raptors are added to a significantly longer ship. Total across both stages, 42 Raptor engines. The super-heavy booster keeps its familiar 33-engine cluster, but the upper stage, the ship itself, grows from six engines to nine. Three sea-level Raptors for landing, and six vacuum-optimized Raptors for efficient performance once it's coasting through space. Musk apparently couldn't resist noting that 42 is, as foretold in the prophecy, a nod to Douglas Adams that tells you exactly how confident this team feels about where the numbers landed. However, if SpaceX only increases the engine count on the ship, that means the super-heavy booster will retain its original 33-engine configuration. Of course, this requires SpaceX to develop an engine with enough power to make that configuration viable, the Raptor 4. If the Raptor 4 produces 300 metric tons of thrust at sea level, then, based on the thrust ratio observed between the Raptor 2 and Raptor 3, the vacuum-optimized Raptor 4 could produce around 320 metric tons. With the current configuration of three sea-level engines and three vacuum engines, the Starship upper stage would generate roughly 1,860 metric tons of thrust. If upgraded to six vacuum engines, the ship alone would produce around 2,820 metric tons of thrust. Those numbers exceed the total thrust of many launch vehicles currently in operation. Beyond raw power, Raptor 4 is also expected to bring further simplification compared with Raptor 3. This could include fewer potential failure points, improved durability, faster refurbishment, and an even stronger focus on full and rapid reusability. All of this could significantly reduce manufacturing costs, turnaround times, and operational complexity. Besides, combined liftoff thrust is targeted at roughly 10,000 metric tons, about 22 million pounds of force. For context, that's close to three times the thrust of the Saturn V, the rocket that took humans to the moon in the Apollo era. If version 4 flies as described, it will be, without much argument, the most powerful rocket ever built by our species. Fully reusable version 3 tops out around 100 tons to low-Earth orbit. Version 4 is designed to roughly double that, delivering 200-plus tons per flight in fully reusable mode. The tanker variant, the version that exists purely to refuel other Starships in orbit, is expected to deliver more than 200 tons of propellant per trip. And the crewed variant reportedly offers more than 1,000 cubic meters of pressurized volume, which is roughly 10% more livable space than the entire International Space Station packed into a single vehicle. And, as Elon Musk promised, the new Starship V4 could ultimately be capable of delivering 2 million metric tons of material into space every year. Let's do a little math. The really interesting part is that you don't have to take anyone's word for it. You can run the numbers yourself. Take Starbase Louisiana's target of 30 launches per day. Multiply that by 365 days a year and you get roughly 10,950 launches per year. Now multiply that by the current version 3 payload capacity of around 150 metric tons per flight at the maximum capacity SpaceX has publicly stated, and you get approximately 1.64 million metric tons per year. Think about that for a moment. Using the current rocket, not some future version, simply operating Louisiana at its eventual full capacity already gets you remarkably close to 2 million metric tons per year. Add the nearly doubled payload capacity of version 4, even with a lower flight rate while the site is being upgraded, and the 2 million ton figure starts to look less like hype and more like straightforward mathematics. The math actually checks out, at least in theory. Still, while the numbers may sound reasonable, inspiring, and impressive, there are plenty of people who argue that Elon Musk's plan has significant weaknesses. Based on an animation video simulating the proposed launch tower layout at Louisiana, one ex-user argued that concentrating 12 launch complexes in a single area cannot simply be scaled up by multiplying the output of a single Falcon 9 launch pad by 12. Why? Because those 12 launch pads would share a much smaller number of fuel storage facilities, around 6 tanks, according to the current plan, along with shared transportation routes and vehicle loading and handling operations. Concentrating that much infrastructure in one location could actually create bottlenecks that distributed launch sites would avoid. 12 launch pads sharing 6 fuel storage facilities and a single loading pipeline present a completely different operational challenge from 12 launch pads spread across independent sites, each with its own supporting infrastructure. There's another question people are raising. If Starship's rapid reusability works as promised, why does the factory keep expanding? SpaceX's current production capabilities at Starbase Texas and Cape Canaveral have been described as almost absurdly large, reportedly capable of handling more than two dozen vehicles simultaneously. If the core value of Starship is rapid reusability, then, in theory, you shouldn't need to keep building new factories just to keep up with the production of new rockets. Yet, the released images appear to show a facility potentially two or even three times the combined capacity already available at other locations. So, what do you think about these conflicting opinions? Let me know in the comments. Your perspective might help challenge or even enlighten the way I and many others are thinking about this. Of course, with any massive project, opposition is inevitable. However, Elon Musk's mission and vision are much bigger than SpaceX itself. It all comes back to the Kardashev Scale, a framework developed in the 1960s for measuring how advanced a civilization is based on the amount of energy it can harness. A Type 1 civilization uses all the energy available on its home planet. A Type 2 civilization harnesses the full energy output of its star, something comparable to a Dyson swarm or massive space-based solar infrastructure. A Type 3 civilization would control the energy of an entire galaxy. By most estimates, humanity today is only around 0.7 on the Kardashev Scale. We haven't even mastered the energy resources of our own planet, let alone those of our star. Musk's argument is simple. You can't build star-scale energy infrastructure without first solving the problem of moving massive amounts of material into orbit cheaply and repeatedly. That's the entire chain of logic. Getting massive amounts of material into orbit enables the construction of megastructures. Megastructures allow us to capture enormous amounts of energy, and harnessing that energy is the practical meaning of becoming a Kardashev Type 2 civilization. According to Musk, any self-respecting civilization should be aiming for at least Type 2. But not everyone believes the Kardashev Type 2 concept is realistic, and some of the criticism comes from people who actually study the subject. Jason Wright, an astronomer at Penn State University, who has spent years searching for signs of Dyson spheres around other stars, acknowledges that launching a million satellites would technically be a step in the right direction. But he is also quick to point out that such a step would still fall far short. Fully harnessing the energy of the Sun, as a true Type 2 civilization would require, would demand more raw material than the entire asteroid belt contains. A million satellites might seem astonishing to us today, but compared with the scale of a true stellar energy civilization, that number would barely scratch the surface. Whether this is a realistic near-term goal, or simply an idea far ahead of its time, remains an ongoing debate within the scientific community. But either way, that doesn't make us any less excited about the next-generation Starship V4 and SpaceX's $100 billion spaceport.