Chapter 255: The Shape of Invisible
November 1976Gorakhpur — Lucknow
The envelope reached the Gorakhpur guest house on the morning of November 18th.
It had traveled from Washington D.C. through the Indian Embassy's diplomatic bag to the Ministry of External Affairs' liaison unit in New Delhi, and from there to the Defence Production liaison office, and from there by government courier to a contact at ISMC whose name appeared in no document associated with the envelope's journey. The contact had placed it on the desk in the guest house private study at six in the morning, before anyone else was in the building.
Karan arrived at the guest house at six-forty.
He found the envelope.
He read the twenty-two pages.
He read them the way he read everything that mattered — slowly, from the beginning, not skipping ahead to the conclusion because the conclusion without the reasoning was a statement and the reasoning was the thing worth understanding. He read with a pen, though he made no marks on the pages themselves — the pen was the habit of active reading, the instrument of engagement rather than the instrument of annotation.
By eight-fifteen he had finished.
He set the pages on the desk.
He looked at the window.
He was not alarmed.
He was something closer to the opposite of alarmed — he was, in the specific way of someone who has been watching something develop from a distance and who has now received confirmation that it is developing in exactly the way he expected, settled. The settling was not complacency. It was the particular calm of understanding.
He knew the F-117. He had known the F-117 in a way that nobody in this building or this country or this time knew it — not the specific classified engineering detail, but the trajectory. He knew what it was, what it became, what it was used for, what its limitations proved to be when those limitations were eventually and publicly exposed. He knew it the way a person knew a book they had read thirty years ago — not every sentence, but the shape of it, the direction of it, the specific ending.
The F-117 entered service in 1983 in the history he remembered. It was used operationally in Panama in 1989. It was used extensively in the Gulf War in 1991. It was, in those contexts, a powerful and effective weapon system.
It was also, in those contexts, aimed at the specific geopolitical environments of its time — the Soviet air defense network, the Iraqi integrated air defense system. The threat environments of the Middle East and eastern Europe.
But in this history — in the specific alternate present he was living in, the present that he had been shaping for six years in a country he had chosen to call home — the F-117 had a different trajectory. The Mauritius Crisis had changed the American military's specific understanding of its threat environment. The nineteen months of intensive programme acceleration that the Pentagon had been running since February 1975 had been framed, in every internal assessment he had seen, in the specific context of India. The stealth programme, the missile programme, the radar upgrade — all of it had been accelerated in response to what India had done in the Indian Ocean.
The F-117 was aimed at India.
Not exclusively. Not without the broader strategic context of the Cold War and the Soviet threat that was always the American military's primary planning basis. But specifically: a low-observable aircraft capable of penetrating air defenses was, in the American defence planning of 1976, a response to the demonstration that conventional aircraft penetrating Indian air defenses would face an adversary capable of engaging them at ranges and with precision that had not previously been part of the threat model.
He thought about this for a long time.
He was not frightened by it. He was, if anything, amused — not frivolously, not dismissively, but with the specific amusement of someone who had watched a chess game develop over many moves and who understood, from the position of the pieces, what both players were actually trying to do. India had built something real and capable. America had responded with seriousness and resources. The response was itself a form of acknowledgment.
But the response had a limitation that the document described precisely and that the document's author — whoever this remarkable engineer was — had not fully resolved in his own thinking.
The limitation was the approach.
He called Aditya at eight-thirty.
Aditya was already at the campus. He arrived at the guest house at eight-fifty and sat across from Karan with the expression he wore when he understood that something significant had happened but had not yet been told what.
Karan slid the twenty-two pages across the desk.
Aditya read.
He was faster than Karan — he always was, he read the way he did arithmetic, with a speed that was not superficiality but compression, the ability to process structured information at high velocity without losing content. He finished in eight minutes.
He looked up.
He said: "This is accurate."
It was not a question. He had been following the aerospace engineering literature and the defence publications and had developed, through that following, a model of where the American stealth programme was likely to be. The twenty-two pages matched his model.
Karan said: "Entirely. Every element."
Aditya said: "The inlet solution specifically."
Karan said: "The inlet solution is the most original part. The rest is derivable from known physics and known materials science by a sufficiently capable team. The inlet solution is the specific contribution of whoever wrote this."
Aditya said: "What do we do with it."
Karan said: "We don't copy it."
Aditya looked at him.
Karan said: "We build something better."
Aditya was quiet for a moment. He had learned, over four years of working with his brother, that this specific sentence — we build something better — was never rhetorical. It was always the beginning of a technical argument.
He said: "Tell me."
Karan said: "The document correctly identifies the faceted design philosophy's fundamental limitation. Against metric-wavelength radar — L-band, P-band, UHF — the facet geometry does not achieve meaningful stealth. The scattering physics at those wavelengths is different from the scattering physics at centimetric wavelengths."
Aditya said: "Yes. He says that explicitly."
Karan said: "What he does not say — what he may not fully understand, because he is working inside the design philosophy of his institution — is that faceting is not the only path to low radar cross-section. It is the path that is computationally tractable with 1976 American computing. It is not the path that produces the best possible stealth aircraft."
Aditya said: "What produces the best possible stealth aircraft?"
Karan said: "A curved surface. Not flat panels at calculated angles — a surface of continuously varying curvature, where every point on the surface is shaped to distribute the scattered radar energy across all angles simultaneously rather than deflecting it specularly in a specific direction. A surface that looks, from the outside, like nothing angular at all. Organic. Like a shape from nature rather than a shape from engineering."
Aditya said: "Why aren't the Americans doing that?"
Karan said: "Because computing a curved surface's radar cross-section is not analytically tractable. You cannot calculate it with a formula. You can only calculate it numerically, and the numerical computation scales with the cube of the surface-to-wavelength ratio. For an aircraft-sized surface at radar wavelengths, that computation is enormous."
Aditya was very still.
He said: "How enormous."
Karan said: "For a surface optimisation of reasonable resolution — finding the curved shape that minimises radar cross-section while maintaining flyable aerodynamics — you need to evaluate hundreds of candidate geometries. Each evaluation is a complete electromagnetic scattering computation. On the best American computing available in 1976, that computation takes something in the range of days per evaluation."
Aditya said: "Days per evaluation. Hundreds of evaluations."
"Yes," Karan said.
Aditya said: "On the Brahma-32."
Karan said: "How many hours per evaluation on forty parallel Brahma-32 units?"
Aditya picked up the pen from the desk. He wrote numbers for approximately ninety seconds.
He said: "Six to eight hours. Per full surface geometry evaluation. On forty units in parallel, assuming the computation can be decomposed across the units, which it can."
Karan said: "Three hundred evaluations at eight hours each, on forty parallel units."
Aditya wrote again.
He said: "Ten weeks of continuous computation."
Karan said: "Twenty weeks if you want to converge a joint optimisation — radar cross-section and aerodynamics iterated together."
Aditya set down the pen.
He looked at the twenty-two pages.
He looked at Karan.
He said: "We can find a shape they cannot find."
Karan said: "We can find a shape they cannot find because they do not have forty Brahma-32 units. We have forty-five production-qualified Brahma-32 units and fourteen in final test. We will have sixty by February."
Aditya said: "And the question is what we do with the shape once we find it."
Karan said: "We build it."
Aditya said: "An aircraft."
"A bomber," Karan said. "A stealth bomber. The kind of aircraft that, when it exists, makes the F-117 — which is optimised for ground attack in a specific threat environment — look like what it is, which is the first generation of an idea rather than the matured version of it."
Aditya was quiet for a long time.
He was the CFO of Shergill Industries. He thought in fiscal years and capital allocation and return on investment and the specific mathematics of committing resources to a programme that had an uncertain outcome. He did this not because he was unimaginative but because it was the specific discipline that kept the enterprise from destroying itself with its own ambitions.
He said: "Timeline."
Karan said: "1981."
Aditya said: "Five years."
"From today," Karan said.
Aditya said: "First flight?"
"1980," Karan said. "Operational delivery 1981."
Aditya said: "The same year the F-117 is projected to enter American service."
Karan said: "Yes."
Aditya looked at the ceiling for a moment.
He said: "The ambition of this is—"
Karan said: "Extraordinary."
Aditya said: "I was going to say insane."
Karan said: "Both."
Aditya said: "And you believe it is achievable."
Karan said: "I believe the computing problem is solved by the Brahma-32 cluster. I believe the materials problem is solvable by Seshadri's team in the time available. I believe the aerodynamics problem is solvable by Rao's team. I believe the avionics problem — the fly-by-wire control system for an inherently unstable aircraft — is solvable by Sundaram's team. Each of those beliefs is based on evidence." He paused. "Whether all four are solvable simultaneously, on the same programme, on the same timeline, without one of them failing in a way that stops everything — that I cannot guarantee."
Aditya said: "And if one of them fails."
Karan said: "Then the programme slips. We deliver in 1982 or 1983 instead of 1981. That is worse than 1981 but it is not a programme failure. The programme only fails if we don't start."
Aditya wrote something in the small notebook he always carried.
He said: "The fiscal cost."
Karan said: "Give me the estimate."
Aditya said: "For a programme of this ambition and this secrecy — no government procurement, no Ministry allocation, no public contract — the cost is entirely internal. The computation cluster is capital expense we have already made — the Brahma-32 units exist. The facility to house them is construction we can fund from the infrastructure budget without separate authorisation, if it is categorised correctly. The engineering team is internal. The aircraft fabrication is internal." He paused. "The materials development — the radar absorber — may require capital equipment I need to assess." He wrote. "My rough estimate for the full programme, over five years, including the first aircraft prototype and the flight test programme, is between six hundred and nine hundred crore."
Karan said: "The range is wide."
Aditya said: "The programme is unprecedented. The range reflects genuine uncertainty. If the computation converges cleanly and the materials work proceeds without a major setback, we are at the six hundred crore end. If we need to iterate on the design or restart the materials programme, we approach nine hundred." He paused. "At either end, it is the largest capital commitment in Shergill Industries' history by a factor of approximately two."
Karan said: "Can we carry it."
Aditya said: "Can we carry it financially — yes. The industrial expansion revenues cover it, the credit facilities cover it, the petroleum royalty compact allocation covers it. It is not comfortable but it is within the envelope." He paused. "Can we carry it operationally — allocating that level of engineering talent to a secret programme over five years while maintaining the rest of the company's development commitments — that is the harder question."
Karan said: "What does it require."
Aditya said: "Ring-fencing. The project team cannot be borrowed from when other programmes hit difficulty. The project schedule cannot be subordinated to commercial deadlines. The people who are on this programme are on this programme, and that is protected status." He paused. "I need the Air Force to contribute to that ring-fencing — not funding necessarily, but operational priority commitment. If the Air Force has formally committed to receive this aircraft and has made its own planning commitments based on that, then the programme has the status that protects it from the internal pressures that would otherwise dilute it."
Karan said: "We will have that commitment."
Aditya said: "You have spoken to the Air Chief."
Karan said: "I am speaking to him tomorrow."
Aditya looked at him.
He said: "Air Marshal Moolgankar."
Karan said: "Yes."
Aditya said: "He will either embrace this completely or refuse it completely. There is no middle position available for a programme this ambitious."
Karan said: "I know."
He said it with the quiet certainty of someone who had spent six years learning how to present extraordinary things to serious people in a way that produced commitment rather than skepticism. He had learned that the correct approach was not to soften the ambition — softening the ambition produced polite interest, not commitment. The correct approach was to demonstrate the basis for the ambition so precisely and completely that the serious person could evaluate it on its own terms.
He had the basis. He had the Brahma-32 units and the process technology and the aerodynamics team and the materials programme. He had, in the twenty-two pages from Burbank, the precise specification of what the competing programme was doing and what its limitations were.
He had enough.
Air Marshal Hrushikesh Moolgankar arrived in Gorakhpur on November 22nd.
He had been the Chief of Air Staff since February 1975 — fourteen months, the specific period in which the IAF's strategic thinking had been fundamentally reoriented by the Mauritius Crisis and its aftermath. He was fifty-eight years old, from Pune, had commanded fighter squadrons and transport wings and had sat in the cockpit of every major aircraft type the IAF had operated since independence, and had the specific quality of a man who had spent his professional life around machines of extraordinary capability and who therefore had a calibrated sense of what was extraordinary and what was merely ambitious.
He had fought in three wars. He had planned operations of genuine strategic complexity. He had the particular quality of senior officers who had commanded at the highest levels: he was completely comfortable with the possibility of failure and completely committed to trying regardless.
He read the three-page summary Karan had prepared.
He read the twenty-two pages.
He sat back.
He said: "The Americans are building this specifically because of us."
It was the first thing he said, after six minutes of reading and four minutes of sitting with his eyes focused on the middle distance in the way of someone who was thinking rather than resting.
Karan said: "Yes."
Moolgankar said: "The Mauritius incident. The missile range. The radar engagement geometry. They drew the same conclusions we drew from the encounter, but they drew them from the opposite side of the engagement."
"Yes," Karan said.
"They are building an aircraft that can penetrate our air defense network at the level of sophistication it currently represents," Moolgankar said. "The specific technical parameters — the radar cross-section target, the altitude envelope, the attack profile — those are designed for the threat environment we represent, not the Soviet threat environment."
Karan said: "That is my assessment."
Moolgankar said: "Which means the timeline matters. If this aircraft enters American service before we have an answer to it, we have a window of vulnerability." He paused. "How long is the window?"
Karan said: "The document describes a programme that the assessment puts at operational deployment in the early 1980s. Call it 1981 to 1983."
Moolgankar said: "And the answer."
Karan said: "We build one first. Or at the same time. And we build a better one."
Moolgankar looked at him.
He said: "You are describing building India's first stealth aircraft — more sophisticated than the American design — and having it operational before or concurrent with the American system."
Karan said: "1981. September 1981, operational delivery. First flight December 1980."
Moolgankar was very quiet.
He was doing the calculation that military men did when presented with a programme — not the engineering calculation, though he understood engineering well enough to follow the broad strokes, but the operational calculation. What it meant. What it changed. What it made possible that was not currently possible.
A low-observable bomber with the range and payload specifications Karan had described. An aircraft that could penetrate any radar network currently deployed by any adversary India faced. An aircraft whose existence, once known, would change the strategic calculations of every nation planning against India, because none of them had a proven answer to it.
He said: "Walk me through what the computation advantage means. Specifically. Why does it produce a better aircraft than the Americans are building."
Karan gestured to Rao, who was sitting at the side of the room. The aerodynamicist had been at the meeting since its beginning, silent, in the role of someone who was present to answer specific questions rather than to participate in the broader discussion.
Rao said: "The American programme is using flat, angled surfaces — facets — to deflect radar energy. The approach works because the scattered field from a flat surface can be calculated analytically, without a computer of great power. The facets are angled so that the radar energy is deflected away from the transmitter — like a mirror reflecting light in a direction other than back at the source." He paused. "The limitation is geometric. You cannot make an aircraft with good aerodynamic performance out of flat panels at discrete angles. The aerodynamic compromise is significant — the aircraft is not particularly agile, its aerodynamic efficiency is reduced, and it has specific instability characteristics that require sophisticated electronics to compensate for."
"And your approach," Moolgankar said.
"A curved surface," Rao said. "Instead of deflecting radar energy in specific directions, a curved surface of the right geometry distributes the scattered energy across all angles simultaneously. The peak scatter in any single direction is very small — too small for a radar receiver to detect above the noise threshold — even though the total scattered energy is not zero." He paused. "The advantage is that a curved surface can be aerodynamically efficient. There is no fundamental conflict between a curved shape that scatters radar well and a curved shape that flies well. The difficulty is finding the specific curved shape that does both simultaneously."
Moolgankar said: "Which requires the computing."
"Which requires the computing," Rao confirmed. "A complete electromagnetic scattering calculation for a curved surface of the complexity we are discussing takes approximately six to eight hours on forty parallel Brahma-32 processors. The Americans cannot do this calculation in a reasonable time because they do not have the computing to do it. We can."
Moolgankar said: "And the result is an aircraft that is stealthier and aerodynamically better than the F-117."
Rao said: "Stealthier across a wider frequency range. The American design is optimised for X-band radar — the centimetric wavelength tracking radars that are the most immediate engagement threat. It is less effective against L-band and UHF search radars. Our optimisation includes both frequency bands explicitly. Against a long-range search radar — the kind that provides the early warning that enables an organized defense — our aircraft is significantly more difficult to detect than theirs."
Moolgankar said: "Because we can find the surface geometry that is optimal at both frequencies simultaneously."
"Because the computer can find it," Rao said. "Humans cannot find it analytically. The computer navigates a six-thousand-dimensional parameter space to find the specific surface shape where both objectives are satisfied simultaneously. That requires evaluating hundreds of candidate geometries. Each evaluation requires a complete electromagnetic computation. The Americans cannot afford those evaluations. We can."
Moolgankar was quiet for several minutes.
The guest house study was warm — the November morning had the specific quality of north Indian November, the first real cold of the year making the indoors welcome in a way that the previous months had not. Outside, the ISMC campus was in its working morning rhythm, the ventilation running, the clean room in its continuous operation, the engineering offices filling.
He said: "What is the aircraft."
Karan said: "Tell me what you need it to be."
This was the correct question — not this is what we are building, will you accept it but what do you need, and let us see whether what we can build matches it. Moolgankar recognised the difference.
He said: "A bomber. Not a ground attack aircraft — a strategic bomber. The difference is range and payload. I need it to reach targets at two thousand kilometres radius without refuelling. I need it to carry a meaningful payload — four thousand kilograms minimum. I need it to operate at altitude, not low-level — low-level stealth penetration is one doctrine but it has costs in fuel and range and crew fatigue that I do not want to accept if the stealth is good enough to make altitude penetration viable." He paused. "And I need two pilots. Not one. A mission of that range and that criticality requires two pilots."
Rao was writing.
He said: "Two thousand kilometres radius, four thousand kilograms payload, high altitude cruise, two crew." He looked at the numbers. "That is an aircraft in the thirty-five to forty-five tonne gross weight range. Achievable with our engine technology. The wing geometry for that weight class at high altitude cruise is significantly different from a ground attack aircraft — larger wingspan, higher aspect ratio. The surface area is larger."
Karan said: "Does larger surface area make the stealth problem harder?"
Rao said: "Not directly. The radar cross-section is determined by the surface geometry, not the surface area, when the geometry is optimised correctly. A larger aircraft with an optimised curved surface is not more detectable than a smaller aircraft with the same surface quality." He paused. "The computation is more expensive — more control points, more evaluations needed for convergence. I need to revise the timeline."
Karan said: "Revise it."
Rao wrote for a moment.
He said: "For the aircraft in this weight class, the surface optimisation requires approximately fifty percent more computation than my earlier estimate. On forty parallel Brahma-32 units, the convergence time extends from twenty weeks to thirty weeks. The design freeze moves from mid-1978 to late 1978."
Moolgankar said: "And the operational delivery."
Rao said: "First flight: December 1980. Operational delivery: September 1981." He paused. "That timeline is achievable if the computation begins by April 1977. If it begins later, the timeline slips accordingly."
Moolgankar said: "April 1977. That is five months from today."
Karan said: "The computation tools need to be built, the cluster needs to be installed, the design needs to be sufficiently refined that the computation starts in the right part of the solution space. Five months is not comfortable. It is sufficient."
Moolgankar looked at the timeline.
He was quiet for a long moment.
He said: "Tell me what success looks like. Not just the aircraft — what success looks like strategically."
Karan said: "In September 1981, India takes delivery of a low-observable bomber that the American programme — the most advanced stealth programme in the world — cannot detect with the radar systems it has deployed, because our aircraft's radar cross-section is optimised against the frequency bands that American search and tracking radar uses. The Americans know we have a stealth bomber because we have demonstrated it. They do not know its specific performance characteristics." He paused. "The strategic consequence is that any adversary planning an air campaign against India must now account for an aircraft that their air defense network cannot reliably detect and engage. Not just the aircraft that is already known — the S-27, the S-35, the missile systems. A new capability that was not in any adversary's planning model until the moment it is demonstrated."
Moolgankar said: "And the Americans. They are building the F-117 in response to what India has. What is their response when India demonstrates a stealth bomber?"
Karan said: "They build a better stealth aircraft." He paused. "That is already happening. The B-2 programme — the American stealth bomber programme that follows the F-117 — is already in early research. What we are doing accelerates their timeline because it demonstrates the strategic requirement for the second-generation capability. But by the time their second-generation capability enters service, we will have the knowledge to build a third generation."
Moolgankar said: "We are in a technology race."
Karan said: "We have been in a technology race since 1975. What we are deciding today is whether we are running it or watching it."
Moolgankar looked at the aerodynamicist, who had returned to his seat.
He said: "Rao. What is the single most likely reason this programme does not meet its 1981 deadline."
Rao thought honestly.
He said: "The computation does not converge. The optimisation problem is genuinely hard, and there is a possibility — I would put it at fifteen to twenty percent — that the algorithm finds a local minimum rather than the global optimum, and the resulting shape, while better than a faceted design, does not achieve the radar cross-section target Pillai has specified. In that case, we need to restart the computation with a different approach, which adds six to twelve months."
"Fifteen to twenty percent probability of a six-to-twelve-month slip," Moolgankar said.
"Yes," Rao said.
"And the probability of the programme failing entirely — of not producing an operational aircraft by 1983," Moolgankar said.
Rao thought again. He said: "Very low. Less than five percent. The worst failure mode is a computation that never converges to a viable solution, which would force us to fall back to a modified faceted design — still better than the American design in several respects, but not the curved surface optimum we are targeting. Even in that failure mode, we have built significant technical knowledge. We would not have nothing."
"But the curved surface optimum is the target," Moolgankar said.
"Yes," Rao said.
"And you believe the probability of achieving it is eighty to eighty-five percent on the 1981 timeline."
"Yes," Rao said. He said it without qualification — not hedging, not softening, the specific directness of an engineer who had done the calculation and was reporting the result.
Moolgankar looked at Karan.
He said: "Eighty to eighty-five percent probability of delivering a stealth bomber — the most capable stealth aircraft in the world — in September 1981, concurrent with the F-117's entry into American service."
Karan said: "Yes."
Moolgankar said: "What do you need from the Air Force."
The conversation that followed lasted ninety minutes.
What Moolgankar needed to give was specific. Not money — the programme was internal to Shergill Industries and the IAF had no mechanism to fund a black programme through commercial accounts without leaving the kind of paper trail that a black programme could not afford. What he could give was commitment, which was in some ways more valuable than money.
He committed: that when the aircraft was ready for flight test, the IAF would provide test facilities, the airspace allocation, and the flight test pilots, without those arrangements appearing in any standard IAF test and evaluation record. That the test pilots would be drawn from a specific pool of experienced fighter and bomber pilots who could be briefed at the compartmented level and who had the specific flying qualities the programme required. That the IAF's operational planning would allocate a permanent slot for this aircraft's eventual unit, beginning in 1979, so that the basing, maintenance doctrine, and training programme were developed in parallel with the aircraft rather than after it.
He committed that Air Vice Marshal Vijayananda — the officer responsible for the IAF's operational requirements and capability development — would be read into the programme within forty-eight hours, because Vijayananda was the person who could most specifically define the operational requirement against which the aircraft was being designed, and defining it precisely in November 1976 was worth more than iterating toward it during a design review in 1979.
He committed that no reference to the programme would appear in any IAF document, internal or external, until he personally authorised it. The existence of the programme, the name of the programme, the description of the aircraft — all of these were controlled by him.
He said: "The name."
Karan said: "Project Marut."
Moolgankar was still for a moment.
He said: "Marut. The wind god."
Karan said: "The thing that moves everywhere and is seen nowhere."
Moolgankar looked at him.
He said: "Marut." He said it once more as if establishing something. "Yes."
He stood.
He said, before leaving: "I want to say something. Not as the Chief of Air Staff — as a pilot."
He said: "I have flown every major aircraft India has operated since 1952. I have flown the Gnat, the MiG-21, the Ajeet, the Marut — the fighter aircraft named Marut, the original Marut, which was our first attempt at an indigenous fighter, and which I believed in when almost nobody else did." He paused. "The original Marut was a good aircraft that was undermined by engine problems we could not solve because we did not have the engine technology. The original Marut failed not because the design was wrong but because one critical system was not ready." He looked at Karan. "I want to be certain, before I commit the Air Force's planning to Project Marut, that you have thought about the equivalent of the engine problem in this programme. The one critical system that could make the aircraft impossible to fly if it fails."
Karan said: "The flight control computer."
"Yes," Moolgankar said.
Karan said: "The aircraft we are designing will be aerodynamically unstable. It cannot be flown without the digital flight control system providing continuous stability augmentation. If the flight control computer fails in flight, the aircraft departs controlled flight within seconds." He paused. "This is not a unique problem — the F-16, which the Americans are already operating, is aerodynamically unstable and depends on an equivalent system. But it is a critical dependency."
Moolgankar said: "Your mitigation."
Karan said: "Four redundant flight control computers — quadruplex redundancy. The system continues operating correctly with any two of the four failed. The probability of all four failing simultaneously is low enough that it is not the limiting reliability factor — the probability of catastrophic loss from other causes exceeds it." He paused. "The flight control computer is based on the Brahma-32. The Brahma-32 will be qualified for flight-critical applications beginning in January. The qualification programme runs eighteen months. It will be complete before the aircraft needs the qualified hardware."
Moolgankar looked at him steadily.
He said: "You have thought about the engine problem."
Karan said: "I have thought about the engine problem."
Moolgankar nodded once.
He left.
The first working group meeting of Project Marut was on November 26th.
Nine people. The guest house conference room, which was smaller than the ISMC conference rooms and was chosen for that reason — fewer chairs meant fewer people, and fewer people was the correct number.
Krishnaswami. Malhotra. Seshadri. Rao. Nair from computational fluid dynamics. Pillai from radar electromagnetics. Sundaram from avionics. And Air Vice Marshal Vijayananda, who had been read into the programme two days earlier and who arrived with the expression of a man who had been told something remarkable and who was still calibrating his response to it.
Karan said: "This is the most ambitious thing Shergill Industries and the IAF have attempted. I want that stated plainly at the beginning so that nobody in this room is surprised by what the ambition requires."
He said: "We are designing and building a stealth bomber. Not a stealth fighter — a bomber, with the range and payload that Moolgankar has specified. We are doing it using a design approach that has never been applied to an aircraft anywhere in the world, because the computing capability required to find the optimal design shape has not previously been available anywhere in the world." He paused. "It is now available here, because of the Brahma-32 and the fabrication capability behind it. The computing advantage is the foundation of the programme. Everything else is the application of engineering capability that this team has built over the previous six years."
He paused again.
"The F-117 is being built to defeat India's air defense systems as they exist today," he said. "It is entering American service in 1981 on the current programme assessment. Project Marut also enters operational service in 1981. The aircraft that enters service against India enters service against an air force that already has the answer to it." He looked around the table. "That is the strategic purpose of the timeline. Not to beat the Americans to a capability — to ensure that the capability they are building does not create a window of vulnerability for India."
He said: "I will say one more thing, and I want it to be clearly heard by everyone in this room, because it affects how I want this team to work."
He said: "The twenty-two pages that initiated this programme came from an engineer who sent them because he loves this country more than he has ever seen it. He is a first-generation American whose grandfather left Rajkot in 1921 and who has never visited India and who sat at his desk with a brass Ganesh on his left side and decided, for reasons that are his own and that I respect completely, that what he knew should serve India." He paused. "He told us about the faceted approach because he thought that was the best available answer to the problem. He does not know that we have the computing capability to find a better answer. He gave us what he had and we will use it to go further than he could go. That is exactly what his gift deserves — not to be a ceiling, but to be a foundation."
He looked at the room.
He said: "We begin the mathematical formulation today. I want the problem fully specified before December 15th. The computation facility construction begins this week — the east secondary building, which has been cleared. The cluster installation begins in January. First computation run begins April 1st."
He looked at Rao.
Rao said: "I will have the mathematical formulation document on your desk by December 12th."
Karan said: "December 10th."
Rao said: "December 10th."
Karan said: "Good."
He looked at the room.
He thought about the shape that the computation would find — the specific curved surface, somewhere in the six thousand-dimensional parameter space, that scattered radar energy across all angles simultaneously and flew at high altitude for two thousand kilometres and carried four thousand kilograms of payload and had, in its complete three-dimensional form, never existed before in any manufactured object anywhere on earth.
He thought: the shape exists. It is in the mathematics, waiting for the algorithm to find it. The algorithm needs the computation. The computation needs the cluster. The cluster is being built.
He thought about November 1976 and about September 1981 and about the sixty months between them in which the shape would go from a mathematical optimum to a physical aircraft to an operational weapon.
Sixty months was a long time.
Sixty months was also exactly the distance between today and the moment when the F-117 was projected to enter American service.
He thought: the timeline is not arbitrary. The timeline is the strategic requirement dressed in months.
He said: "Any questions."
Vijayananda raised his hand. He was a measured man, a staff officer's precision in his questions.
He said: "The programme does not exist. How do we manage the procurement of materials and components that are not manufactured internally? Radar absorbers, composite structures, specialised avionics components — some of these require suppliers outside the Shergill ecosystem."
Karan looked at Aditya.
Aditya said: "Every external procurement for this programme goes through a front company we are establishing. Gorakhpur Advanced Research Applications Private Limited — GARAPI. Registered in Delhi, with a genuine commercial profile in advanced industrial materials research. Its suppliers see a materials research company with reasonable order sizes. They do not see an aircraft programme." He paused. "The front company is legally legitimate. The description of its business is accurate — it genuinely does advanced materials research. The research it conducts is for this specific application, but the research itself is real."
Vijayananda considered this.
He said: "Acceptable."
Karan said: "Anything else."
Seshadri said: "The radar absorber. I need to start the chemistry programme in January. I need a specification."
Pillai said: "I can give you a preliminary specification by December 15th. X-band and L-band simultaneously, twenty decibels reduction target. Graded layer design — the electrical properties need to vary through the thickness of the coating."
Seshadri said: "Graded layer chemistry for a compound-curved surface application. You know what that means for the application process."
Pillai said: "I know."
Seshadri said: "I will need at least one materials engineer from the process development team assigned to me full time from January. Non-negotiable."
Karan said: "Name the person."
Seshadri said: "Deshpande."
Krishnaswami said: "Deshpande is on the 0.8-micron process development."
Seshadri said: "Then you need another engineer for the 0.8-micron process, because I need Deshpande for this."
Karan looked at Krishnaswami.
Krishnaswami looked at the table for a moment.
He said: "The 0.8-micron programme slips by approximately four months."
Karan said: "Accept the slip."
Krishnaswami said: "Noted."
Seshadri said nothing further. He had the thing he needed.
When the meeting ended and the people filed out into the November morning — the ISMC campus busy in the midday hour, the fabrication facility in its continuous cycle, the engineering buildings with their specific internal light — Karan stayed for a few minutes in the conference room alone.
He thought about the original Marut. The fighter aircraft that had been India's first attempt at indigenous combat aircraft design, in the 1960s, which Moolgankar had mentioned. The aircraft had flown. It had been a real aircraft, designed and built by Indians, with genuine capability. It had been limited by the engine — the British engines it used were underpowered for the airframe's potential, and the indigenous engine programme could not fill the gap.
The engine problem had stopped the original Marut.
He thought about the current programme's equivalent of the engine problem — the flight control computer, which he had named in response to Moolgankar's question. The flight control computer that needed the Brahma-32 qualification. Eighteen months. The qualification needed to complete before the aircraft was ready for first flight.
He thought: the engine problem in 1969 was unsolvable because India did not have the turbine technology. The flight control computer problem in 1976 is solvable because we built the Brahma-32 ourselves, on our own process, on our own substrate, qualified by our own programme.
The difference between the original Marut and Project Marut was the semiconductor programme. Not directly — not as an obvious causal chain — but as the accumulated technical capability that made the critical system achievable where it had previously been unachievable.
He thought: everything is connected.
He thought about six years of building — the industrial programme, the semiconductor programme, the aeronautics programme, the radar programme, the materials programme — and how each of them had been built in a direction that made the next thing possible. Not because each programme had been planned in service of this specific outcome. Because the capability accumulated regardless of the specific application, and when a new application appeared, the capability was there to serve it.
This was the specific thing that distinguished building from buying. You did not choose what you needed before you built it. You built capability and the capability revealed what it made possible.
He thought about September 1981.
He thought about the shape that existed in the mathematics, waiting to be found.
He thought: we are going to find it.
He stood.
He picked up the folder.
He went back to work.
End of Chapter 255
Project Marut — Programme SummaryClassification: Compartmented — Named Access OnlyInitiated: November 1976
Strategic context: F-117 programme identified as specifically designed to defeat Indian air defense systems. American operational deployment projected 1981. Project Marut initiated to ensure Indian operational response at equivalent or earlier date.
Technical differentiation from F-117:
Curved surface design vs. American faceted design Optimised against both X-band and L-band radar simultaneously (American design optimised primarily for X-band) Superior aerodynamic efficiency enabled by curved geometry Enabled by Brahma-32 parallel computation cluster (40+ units) unavailable to American programme in 1976
Computation approach: 6,000-variable surface geometry optimisation (joint radar cross-section + aerodynamic efficiency) using differential evolution algorithm on 40-unit Brahma-32 parallel cluster. Estimated 30 weeks computation for full convergence. Computation start: April 1977.
Aircraft specification:
Weight class: 35–45 tonnes gross Radius of action: 2,000 km Payload: 4,000 kg Crew: 2 pilots Altitude: High-altitude cruise penetration Propulsion: Existing engine technology programme
Key programme risks:
Computation convergence failure (15–20% probability of 6–12 month delay; <5% probability of programme failure) Radar absorber development (graded layer chemistry for compound-curved surface; critical path) Brahma-32 flight-critical avionics qualification (18 months from January 1977)
Timeline:
Nov–Dec 1976: Mathematical formulation, programme structure Jan–Mar 1977: Computation cluster construction, tool development, initial shape families Apr 1977–Oct 1977: Optimisation computation Nov 1977–Dec 1977: Scale model validation (wind tunnel + anechoic chamber) Jan 1978–Sep 1979: Structural design, materials, systems integration Oct–Nov 1979: Ground testing Dec 1980: First flight Sep 1981: Operational delivery, IAF
IAF commitment: Air Marshal Moolgankar (Chief of Air Staff), Air Vice Marshal Vijayananda (operational requirements). Basing, maintenance doctrine and training programme development begins 1979. Quadruplex redundant Brahma-32 flight control computers.
Commercial cover: Gorakhpur Advanced Research Applications Private Limited (GARAPI) — front company for external material and component procurement.
Programme name meaning: Marut — Vedic god of wind. That which moves everywhere and is seen nowhere.