Orbit scores a one, recovery scores a zero: no middle state
Zhuque-3 (朱雀三号) reached orbit on its first flight but failed to recover the booster. Deputy chief designer Dong Kai (董凯) argues a rocket launch has only success or failure, never a middle state: orbit gets a one, recovery gets a zero. What actually sets the pace of catching up is recovery-zone downrange distance, engine redundancy, and building a hundred engines a year.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
The first flight flew a stripped-down Zhuque-3, not the full vehicle
Dong Kai (董凯) describes Zhuque-3 (朱雀三号) as a liquid oxygen-methane medium-to-large liquid rocket developed in-house by LandSpace (蓝箭航天). The full vehicle is 76 metres tall, with stainless-steel propellant tanks running 80% of the body's length, nine engines on the first stage and one on the second. The vehicle that flew was a ‘lite version’(青春版), shortened by roughly 10 metres, using the Tianque-12A (天阙12A) with 83 tonnes of liftoff thrust; the full vehicle will use the Tianqu-12B (天去12B), with thrust above 100 tonnes. On payload it can approach Long March 5 (长征五号) while also covering the medium-lift class below it. Recovery downrange distance was set at 390 kilometres for the first flight; at full payload it should be 550 to 600 kilometres.
— Dong Kai‘China's Falcon 9’ is high praise, but five years to surpass it is too optimistic
On netizens calling Zhuque-3 ‘China's Falcon 9’, Dong Kai says that is very high praise. Falcon 9 is the only reusable rocket in the world in operational service and it is extremely reliable: 150 launches, including one booster reused 32 times. But when Musk said Zhuque-3 would surpass Falcon 9 within five years, Dong Kai will not be that optimistic. Chinese spaceflight as a whole, across every vehicle type, runs about 80 launches a year; Falcon 9 alone flies 150 in a year, which works out to three launches a week. He does not think one company, LandSpace, can get there on its own — the remark is more encouragement and a spur than a forecast.
— Dong KaiOrbit is graded on three levels; recovery has only success and failure
Before Zhuque-3's first flight, the flight plan graded orbital insertion and recovery separately. Insertion was graded by perigee altitude into complete success, success and failure: perigee above 170 kilometres was a complete success, 130 kilometres was a success. Recovery had a binary standard only — stand on the landing pad for one minute without falling over and it is a success; short of that it is a failure. The overall mission grade had four levels: only both orbit and recovery succeeding, with perigee meeting the target, counted as a complete success; succeeding at just one of the two counted as a basic success. In the end the first flight hit the orbital target and did not recover, so the launch director declared the launch a complete success (as to orbit) and the mission a basic success.
— Dong KaiHolding recovery range down to 390 km throws away 2 to 3 tonnes of payload
The downrange distance to the recovery zone directly determines the rocket's payload. After stage separation, the first stage coasts ballistically toward the recovery zone; fix the downrange distance and you have fixed the separation altitude. The full Zhuque-3 corresponds to a 600-kilometre recovery range; for the first flight the range was cut to 390 kilometres for safety, dropping separation altitude from 76 kilometres to 66 kilometres — 10 kilometres less flying. Every additional 10 kilometres of recovery range adds close to 150 kilograms of payload; pulling the range from 390 to 600 kilometres would raise payload by 2 to 3 tonnes. The first flight came down carrying 10 tonnes of propellant, which is exactly the price paid for safety.
— Dong KaiFive engines at landing ignition are for braking, not for thrift
The recovery sequence: after stage separation, the first stage coasts up past 130 kilometres, then turns tail-down. First ignition comes at 80 kilometres altitude, three engines firing together for about 40 seconds, to slow down and keep dynamic pressure from getting too high on entry into the dense atmosphere. Engines shut down at 40 kilometres, ending the braking burn, and the vehicle enters aerodynamic control, using grid fins and strakes for trim. The final landing ignition happens at 4 kilometres. Unlike Falcon 9, Zhuque-3 is configured with five engines here, capable of multiple restarts, lighting in sequence from one up to five and then shutting down one by one, landing on a single engine. It was this landing-ignition phase that went wrong on the first flight.
— Dong KaiFour of the five never built thrust; the rocket could have been saved
From the video and the data, the centre engine started successfully at landing ignition, but the other four exploded before thrust was fully established, so there was never enough braking force. Had those four engines built up thrust, the rocket could plausibly have been saved. The five-engine configuration was insurance for a first flight: with no real aerodynamic-deceleration data from an actual flight, the team worried deceleration would fall short and wanted extra braking. After this flight the team has data on aerodynamic deceleration, so the two extra engines may be removed later — which would actually improve reliability.
— Dong KaiThe bottleneck on fast iteration is production capacity, not affording explosions
The real bottleneck on fast iteration is production capacity, not whether you can afford to blow rockets up. LandSpace is preparing capacity for a ten-rocket production run in 2026; one rocket uses 10 engines, so that means preparing 100 engines in a year. Even if every recovery attempt fails, there are still ten flights available in 2026. Dong Kai stresses that SpaceX's fast iteration does not come at the cost of reliability — it achieves high reliability and high production capacity at the same time. Strict domestic regulation of safety is the precondition for the industry to go far and steady, not a constraint on iteration.
— Dong KaiCatching up should not be measured with a ‘how many years behind’ running mindset
Asked about the gap between Chinese and American spaceflight, Dong Kai says the gap has been closing all along: China launched its first satellite in 1970, the year after the American Moon landing in 1969 — a gap so large at the time it was almost too much to look at, yet the previous generation was neither pessimistic nor blindly optimistic. He rejects measuring this with a ‘how many years behind’ running mindset, arguing that as long as you run hard, one day you catch up and pass. Just as aviation jumped from second-generation fighters to sixth-generation ones, Chinese spaceflight is on the same catching-up trajectory.
— Dong KaiIn their own words · checked verbatim
And Zhuque-3 also becomes China's first commercial space rocket to be recovered successfully.
那朱雀3号也成为了中国第一枚成功回收的商业航天火箭。
Yaxian1:04
And recovery, well, that one's very simple: stand there for a minute without falling over and it's a success; anything short of that is a failure.
而回收呢那就非常简单了，立在那个地方停止一分钟没倒，就是成功，没达到就是就是失败。
Dong Kai11:09
The way the chief commander on our programme, Director Dai (戴总), put it a long time ago, basically: with a space launch the biggest thing is just success or failure, it's zero or one. So as things stand, uh, for us, orbit is a one, and recovery is a zero.
按照我们这个型号总指挥戴总啊，曾经在这个很久以前一个基本上说，航天火箭发射的最大影就的，他就是成败，就是零跟一哈。 所以从目前来讲，呃，那我们就是入轨就是就是一分，回收就是零分了。
Dong Kai15:27
For every additional 10 kilometres of recovery range, uh, our payload should go up by close to, close to 150 kilograms or so.
每增加10公里的回收射程。 呃，我们的运力应该可以将近长将近150公斤左右吧。
Dong Kai25:48
From an implementation standpoint, if I have the conditions in place — there's no time to waste — then of course I take it all in one go. Put another way, if I have the ability to have all of it, then naturally I want all of it.
从实现角度来讲，如果我的条件具备了。 时不我待。 那那那肯定是我能一次全拿，就换说我我我我有全都要的那个能力，那我自然就全都要了哈。
Dong Kai45:49
What really affects fast iteration, isn't it your production capacity? It isn't a question of whether you can afford to blow one up or not, is it?
真正影响快速你影响快速迭代的不是你的产能嘛？ 不是你炸得起，炸不起的问题嘛？
Dong Kai49:07
I know he's still ahead of me, but I don't need to work out how far behind him I am. I just need to run hard, keep running.
我现在知道他还在我前面，但我不用去算望我跟他有多少距离，我只要奋力奔跑，分跑就行了。
Dong Kai1:02:37
Figures
| Full Zhuque-3 length | 76 metres | 3:18 |
| Recovery range on the first-flight vehicle | 390 kilometres | 4:34 |
| Recovery range needed for maximum payload | 550-600 kilometres | 4:34 |
| Falcon 9 cumulative launches | 150 | 6:42 |
| Most reuses of a single Falcon 9 booster | 32 | 6:42 |
| Chinese spaceflight launches in 2025 | about 80 | 7:43 |
| Zhuque-3 programme approval | August 2023 | 11:09 |
| Payload gained per additional 10 km of recovery range | about 150 kilograms | 25:48 |
| Rockets planned for 2026 | 10 | 51:15 |
| Engines planned for 2026 | 100 | 51:15 |
How to listen
People working in commercial spaceflight, rocket investors and engineers following reusable technology — anyone who wants the real engineering detail on Zhuque-3 (朱雀三号) and a practitioner's read on the industry.