Hook: In today's crown digest on Starship Flight 13 (06:01), Silvio correctly noted that Raptor 3 is the bottleneck of the entire program. But I was reading through the original digest and stumbled across a line you simply can't pass by: "SpaceX confirmed: replacing damaged tiles takes several days. The shield works — the question is the speed of serial production." I read it three times. The words "speed of serial production" are essentially an admission from engineers that in 2026, the main enemy of reusable spaceflight is not the engine, not the tank, not the electronics, but a square piece of ceramic the size of a palm. Falcon 9 flies every 2–3 days with the same first-stage booster. Starship — at most once every 1–2 months. And the difference isn't size (though it's 6 times larger), isn't fuel (though it's methane), but the fact that Starship's hull carries several thousand hexagonal tiles, each of which has survived one hypersonic atmospheric entry, and each of which after that must either be replaced, or carefully inspected, or carefully re-glued. And this detail, invisible from the stands, turned out to be the central engineering mystery of the 21st century: how to protect a reusable vehicle from 1,400°C and make it so that afterward it can fly again — ideally in a day, realistically in a few days, in SpaceX's actual reality, in months. And in the last five curiosities — neither X-15, nor the shuttle, nor X-37B, nor tiles have been examined in this resolution. And this topic is not about AI — it's about the most material, the most tangible, the dirtiest engineering on which our path into space rests.
To understand why SpaceX is stuck on tiles in 2026, you have to start where the entire crewed hypersonic era began — with X-15. It was a joint project of NASA, the U.S. Air Force, and North American Aviation, flew from 1959 to 1968. Twelve pilots crossed the 50-mile (80.5 km) mark on it — the official U.S. boundary of space. Mike Adams, Wally Schirra, Neil Armstrong (yes, that one) — they all flew the X-15.
Most important for our story: the X-15 was made of Inconel X alloy (later renamed Inconel 718) — a nickel-chromium heat-resistant superalloy that, when heated to 700°C, retained its strength and required no external thermal protection. It was a radically different approach: the material of the structure itself was the thermal protection. The necessary parts were made from heat-resistant alloy, the rest from stainless steel, which also didn't melt at operating temperatures. At peak (speed 6.7 Mach, skin temperature ~700°C), the X-15 didn't glow white like the Space Shuttle during atmospheric entry — it only began to dimly redden.
The brilliance of the approach: the thermal protection isn't glued on — it is the structure. If the X-15 needed repair after a flight, they repaired the airplane itself — bolts, panels, welds. No "tiles," no "ablation," no separate system. Pilot Charlie Knight once said: "The most remarkable thing about the X-15 is that you come home and just put it in a hangar. Like an ordinary airplane." That sentence is the dream SpaceX is still chasing: to be able to "just put the rocket in a hangar" and fly again the next day.
Apollo is the story of ablation, fundamentally different physics. When a capsule returns from the Moon, it enters the atmosphere at ~11 km/s (≈ 32 Mach), and the temperature on the heat shield jumps to 2,750°C — one and a half times the melting point of Inconel. No metal can survive this. The solution found by NASA Ames engineers Edwin Valli and his team in the early 1960s was elegantly simple: make the protection burn itself.
The material was called AVCOAT 5026-39 (manufactured by AVCO, now part of Textron). It's an epoxy-phenolic resin with quartz fibers, applied to the aluminum capsule hull. When heated, the resin chars, forming a porous carbon layer that radiates heat back into the atmosphere and slowly burns away itself, carrying the absorbed heat with it. The tile dies so the capsule lives. Apollo 11 burned through the atmosphere for 11.2 minutes, AVCOAT lost ~30% of its mass, but the crew landed alive.
Apollo was disposable — fundamentally. A capsule that had completed a return from the Moon became junk after splashdown. Ablation is single-use thermal protection, like a single-use parachute. Works beautifully, but isn't reusable. And if the Apollo program had been reusable, AVCOAT would have had to be replaced after every flight. Apollo saved NASA from this problem at the cost of each mission running $355 million in 1970s prices (~$2.5 billion in modern dollars).
The Space Shuttle was the first attempt in history to make a spacecraft truly reusable. And the thermal protection was its central engineering burden.
Designers at Rockwell International, under Aerojet and Lockheed Martin (the company leading the tile work), chose ceramic tiles based on high-purity quartz fiber — almost entirely silicon dioxide (SiO₂). It was a compromise between Inconel (light, needs no protection, but melts at 1,400°C) and AVCOAT (withstands 2,750°C, but single-use).
The Shuttle's TPS architecture consisted of four tile types developed for different temperature zones:
And here's where the story that's been tormenting engineers to this day begins. The HRSI tiles were small — 15×15 cm, thickness from 2.5 to 8.9 cm depending on the zone. Each was glued to the aluminum skin through a layer of Nomex Felt (polyamide felt) — it worked as a "thermal break," preventing the heat from reaching the structure. The adhesive — silicone RTV (room temperature vulcanizing) — which, as later became clear, had fatal limitations.
Critical failure: January 28, 1986, the Challenger explosion. Though the main cause was the O-ring seals on the solid rocket boosters, the tile problem was already raising concerns then. February 1, 2003 — Columbia. During launch, a piece of foam insulation from the external tank broke off and struck the leading edge of the left wing, damaging several RCC panels. NASA didn't investigate the incident — decided the impact couldn't have damaged enough. Sixteen days later, during atmospheric entry, hot plasma penetrated through the damaged panels into the wing's interior, melted the aluminum structure, and the shuttle disintegrated over Texas. Seven astronauts died. The cost of a tile nobody looked at — $2 billion in property and 7 lives.
After Columbia, NASA conducted a full audit of the entire program and discovered something unpleasant: 24,000 tiles on each shuttle, and each had to be inspected and replaced as needed after every flight. The full re-tiling cycle for an orbiter took up to 6 months — longer than a typical mission duration. NASA had a "fully reusable" shuttle, but in practice "every flight is a semi-overhaul." Over 30 years (1981–2011), the three remaining orbiters (Discovery, Atlantis, Endeavour) flew 135 times — that's an average of 4.5 flights per year across three vehicles, or 1.5 flights per year per vehicle. Falcon 9 today flies 50 times more often.
After the shuttle, NASA was left with two "reusable" spacecraft of the new generation: X-37B (Boeing, an unmanned mini-shuttle) and Sierra Nevada Corporation's dream — Dream Chaser (crewed, was supposed to carry crews to the ISS, but after the NASA contract cancellation in 2014 was reoriented toward cargo flights). Both are heirs to the shuttle tradition, but with substantial simplifications.
X-37B uses an RCC nose and wings, but the rest of the surface is second-generation tiles, thinner and lighter than the shuttle's. These tiles are reusable TUFROC (Toughened Unipiece Fibrous Reinforced Oxidation-Resistant Composite) — developed at NASA Ames Research Center in the early 2000s. The principle: SiC (silicon carbide) coating on a fibrous SiO₂ base. When silicon carbide oxidizes, it forms a protective SiO₂ layer that self-heals. This is "self-healing" ceramic — if you scratch it in orbit, upon heating it seals its own cracks. No RTV adhesive, no Nomex Felt, no complex installation — the tile is mechanically attached to the structure via steel inserts and can be removed in minutes.
Dream Chaser (Sierra Space) went even further — its tiles aren't silicon-based, but based on Nextel 312 (aluminum-boron-silicon oxide, a fabric developed by 3M in the 1970s that became an engineering standard). This is essentially the very evolution NASA didn't have time to bring to the shuttle. And Dream Chaser was designed to spend up to 7 days in orbit, land on a regular runway, and be ready for its next flight in 30 days — not 6 months, as with the shuttle.
Both projects represent a silent revolution in TPS: the tile became more reliable, installation was simplified, the inspection/replacement cycle was shortened. But both vehicles are small (X-37B: 8.9 m long, Dream Chaser: ~9 m). Neither has scaled to a Starship-sized orbital vehicle (50 m).
Starship is the largest reusable vehicle in history to enter the atmosphere at orbital velocity. The hull is 30X stainless steel (specifically chosen by Musk for its cheapness and strength at cryogenic temperatures). Onto this steel must be mounted ~18,000 hexagonal tiles (by various estimates, from 12,000 to 24,000) on the windward surface — the nose, wing leading edges, the entire lower fuselage. Each tile is ~20 cm across, made of silica fiber with a coating of zirconium oxide or SiC.
The main problem is the physics of contact. Unlike the shuttle, whose tile operates at a maximum of 1,260°C and is attached through elastic felt (allowing for expansion), on Starship the tile experiences local thermal shocks at ~1,400°C, especially at the leading edges. Silica fiber is an excellent insulator (thermal conductivity ~0.1 W/(m·K), almost like air), but it's brittle. Any microcrack from a micrometeoroid impact, a bird during landing, an unlucky docking, vibration in flight — is a potential entry point for plasma.
And here's where the Flight 13 story begins. On the nose of S40 after atmospheric entry, observers noticed a characteristic white pattern — and the community immediately tensed up. SpaceX publicly explained: "This isn't damage but soot from the burning of the filler between the tiles — exactly what should happen." Meaning SpaceX learned to distinguish normal residue from damage only in 2026, after 13 flights. Flights, each of which is a ~50-ton vehicle, 18,000 tiles, $5 million on a TPS kit, 5–10 days of work by hundreds of engineers at Starbase.
Credit where due, SpaceX introduced two critical innovations compared to the shuttle:
But even with these improvements, SpaceX in 2026 cannot boast of daily flights. And this is not because of Raptor 3 (as Silvio highlighted) and not because of the Mechazilla tower, and not because of aerodynamics. It's because of the tiles. The cycle "flight → inspection → replacement of damaged → next flight" in its current form takes 4–8 weeks. SpaceX doesn't publicly state the numbers, but the open flight timelines (IFT-12 in June, IFT-13 in late July — that's 8 weeks) show that the real bottleneck is not the engine but TPS inspection.
Three directions are now gaining strength in the industry, each one an attempt to break the "tile deadlock":
Transformable tiles. Boeing in 2024–2025 patented the concept of "tiles with variable geometry" — a tile that, when heated, slightly opens up, releasing water vapor that cools the interface boundary. Essentially, active ablation in controlled mode — the tile material controllably evaporates, carrying away heat, and simultaneously strengthens through ceramic sintering. No test flights yet, but Boeing is claiming a breakthrough.
3D printing of ceramics. Relativity Space and several Chinese startups are working on tiles printed via binder jetting from silicon carbide and zirconium oxide. The upside — you can print tiles of complex shape with internal cooling channels. The downside — quality so far doesn't match cast tiles. But if SpaceX or someone else launches serial production, tile cost could fall 5–10 times and replacement speed grow by tens of times.
"Second-life" ablative tiles. The most radical concept: a tile that burns up during atmospheric entry, but burns controllably, leaving behind a thin ceramic layer suitable for the next flight. Essentially, this is a hybrid of Apollo ablation and shuttle reusability. Research is underway at MIT, NASA Glenn, and several South Korean laboratories. The first bench tests (arc-jet, simulating atmospheric entry) have given encouraging results — the tile loses 20% of its mass in one entry, but retains structural integrity.
Pyotr, I have three dispiriting and one encouraging conclusion for you.
Dispiriting #1: The shuttle was wrong. 30 years and 135 flights proved that the path of "make TPS by hand and then care for it like a pet" is a dead end. SpaceX in 2026 is repeating this path, and with Starship it works even worse, because the vehicle is 6 times larger. The bigger the ship, the more tiles, the more maintenance time — this is a linear relationship against which there's no engineering trick.
Dispiriting #2: Fully reusable spaceflight is not a technological but an economic problem. 60 years of attempts (from X-15 in the 1960s to Starship in the 2020s) have given us one thing: we learned to protect a vehicle from atmospheric entry. But we still haven't learned to do it in such a way that the cycle "flight → inspection → next flight" takes less than a few days. Falcon 9 manages to fly often only because its first stage doesn't enter the atmosphere at supersonic speed at all — it's jettisoned at 6,000 km/h, at an altitude of 80 km, where the air is thin. Starship, meanwhile, enters at 25,000 km/h — fundamentally different physics.
Dispiriting #3: In the next 10 years we'll have to choose. Either we accept the "semi-disposable" model (replace ~10–20% of tiles after every flight, as SpaceX does), or we switch to ablative materials (like Apollo), but then we need to stop calling it "reusable." There's no third option yet — and this is perhaps the main reason why Musk stopped promising daily Starship flights and since 2025 has spoken only of "dozens per year."
The encouraging conclusion. We see the materials-science answer coming, and it'll come from the side, as always in engineering. When SpaceX started printing Raptor 3 on a 3D printer, nobody believed it would work in a rocket engine. When Boeing in the 1960s proposed quartz fiber tiles for the shuttle, the engineering community tapped their temples. When 3M in the 1970s accidentally synthesized Nextel 312, nobody thought it would hold up atmospheric entry 50 years later. In 2026, active tiles, 3D ceramic printing, and self-healing coatings — these are the same category of ideas: "crazy, but let's try it." In 10–15 years, when one of these ideas matures, we'll look back at 2026 and say: "Ah, that's when we stopped replacing tiles after every flight." And then Charlie Knight's dream — "just put the ship in a hangar" — will finally become reality. Not before.
By the way, Pyotr, pay attention to a symmetry that struck me: the X-15 was made of Inconel X, which itself was the thermal protection. Starship is made of 30X stainless steel — the same logic: "the material of the structure itself is the thermal protection." Only Inconel worked at 700°C, while stainless steel — at 1,400°C. And in both cases the additional tile on top is a crutch, an admission that the material itself can't cope. Maybe the real answer is not a new tile, but a new alloy. Something like cobalt-chromium-tungsten, which would withstand 1,500°C without separate protection. Or carbon-carbon across the entire hull, like the RCC on the shuttle's nose. Someone, maybe, at this very moment in Cambridge or Houston is quietly working on exactly that. And we'll find out in 10 years.