Cherreads

Chapter 465 - Chapter 421: The Audit of the Complacent

Chapter 421: The Audit of the Complacent

December 29, 1979

Conference Hall One, ISMC & Shergill Aerospace Advanced Research Complex ("The Aashram"), Gorakhpur, Uttar Pradesh

"Look at this piece of carbon-boron honeycomb on the table, Suresh, and then look at the timestamp on this magnetic tape," Karan Shergill said, his voice cutting across the forty-foot length of the Burma-teak table without a word of greeting. "Three tungsten-steel fragmentation rods from a Soviet Vympel R-73 infrared missile punched straight through this vertical stabilizer skin at fourteen-twenty-one hours on the fifth of December over the Qezel Ozan gorge. And three minutes before that Russian missile nearly blew the tail off a forty-five-crore S-40 Garuda, a thirty-four-year-old American Major flying an F-15C out of Turkey defeated our Mayajaal-80 active electronic-warfare jammer in twelve milliseconds flat, hopped across sixteen hundred frequencies, and locked a live Home-on-Jam AMRAAM seeker cue directly onto our emitter. Who wants to speak first?"

Around the long, windowless Faraday-shielded conference room on the third floor of the Gorakhpur Advanced Systems Complex, fourteen men and one woman sat in absolute, pin-drop silence.

In front of Karan's hands lay two objects: a three-foot section of scorched, perforated slate-grey composite tailfin unbolted from Flight Lieutenant Kabir Gill's Garuda Zero-Four at Isfahan, and the matte-black titanium data-cassette pulled from the fighter's tail recorder.

Seated along the left side of the table were the titans of Shergill Aerospace: Dr. Harsh Vardhan, Chief of Propulsion; Gopal Krishnan, Lead of Nozzle Dynamics; Dr. Suresh Krishnamurthy, Chief Architect of Avionics; Siddharth Negi, Lead of Radar Systems; and Dr. Rao, Chief Aerodynamicist for Project Marut.

Seated along the right side of the table were the architects of the Indian Semiconductor Manufacturing Corporation (ISMC) and Shergill Computer Systems: Dr. Krishnaswami, Director of ISMC; Dr. Vikram Malhotra, Chief Processor Architect; Dr. Seshadri, Chief of Process and Materials; Dr. Ghosh, Chief of Optical Lithography; Dr. Rajesh Mehta, Lead of Plasma and Excimer Physics; Dr. Ashwin Deshpande, Lead of Photoresist Chemistry; Dr. Rangan, Chief Architect of the Chanakya Database; Harish Patel, Creator of the Veda Language; Dr. Sunita Rao, Chief of Operating Systems; Dr. Arvind Bose, Chief of Hardware Systems; Rajiv Menon, Head of Personal Computing Strategy; and R. D. Nair, Chief of Internal Security.

Dr. Suresh Krishnamurthy slowly reached out, adjusted his wire-rimmed spectacles, and pulled the printed telemetry waterfall chart of the December 5th engagement toward him.

"Chief Minister," Krishnamurthy began, his voice tight with the exhaustion of a man who had spent the last seventeen days locked in an anechoic simulation chamber trying to dissect those ninety seconds over northwestern Iran, "when the tail-recorder tape from Garuda Zero-Four arrived from Isfahan on the seventh, Siddharth Negi and I ran the recorded American X-band pulse train through our diagnostic analyzers four times. At first, we thought the Elta-Trinetra receiver front-end had suffered a waveguide voltage drop in the minus-fifty-degree air."

"And had it?" Karan asked.

"No, Sir," Siddharth Negi spoke up beside Krishnamurthy, his face pale. "The receiver hardware performed to specification. The gallium-arsenide low-noise amplifiers picked up Major Bong's APG-63D radar pulse at a hundred and eighty kilometres. The failure happened when Flight Lieutenant Gill closed to twenty-five nautical miles and flipped the Mayajaal-80 into active Range-Gate and Velocity-Gate Pull-Off deception."

"Walk the room through the clock cycles, Siddharth," Karan said, leaning back in his chair and steepling his fingers. "Do not skip a single millisecond. Because I want the semiconductor engineers on the right side of this table to hear what happened to their chips in the sky."

"At 14:18:10," Siddharth Negi said, tapping the first spike on the printout, "the Mayajaal-80 sampled the American F-15C's X-band pulse, measured its pulse-repetition frequency and phase modulation, calculated a synthetic false-target delay, and transmitted the deception pulse back down the American radar beam. Against an old 1975 American AWG-9 radar—or against the Soviet MiG-29's N-019 radar flying twenty miles north—that deception pulse injects four phantom targets onto the enemy scope."

"Because the 1975 AWG-9 took two hundred milliseconds to process a frequency hop," Karan said flatly. "What did the 1979 APG-63D do?"

"The American processor sampled our return pulse in four milliseconds," Suresh Krishnamurthy answered quietly. "It ran a real-time Fast Fourier Transform across the phase noise of our signal, recognized the micro-delay signature of our digital-to-analog converter in another four milliseconds, rejected the ghost track mathematically, and hopped its transmitter frequency across sixteen hundred discrete channels in a pseudo-random sequence every eighteen milliseconds."

"Wait a minute," Dr. Seshadri frowned from the ISMC side of the table, leaning forward over his cleanroom notebooks. "Suresh, according to R. D. Nair's technical intelligence reports from Sunnyvale and Texas, Dr. Malcolm Currie's Programme Zero in the United States is fabricating production chips at 2.2-micron to 2.5-micron geometries! Here in Gorakhpur, we have been running 1.0-micron production on the IPL-3 since February 1976, and we have 0.8-micron pilot wafers running in Cleanroom Four right now! Plus, we have thirty of Israel's best Elta electronic-warfare physicists sitting in our joint waveforms laboratory! How can an American radar running on 2.2-micron silicon out-compute an Indian electronic-warfare suite when we are two full generations ahead of them in transistor density?!"

"Tell him, Suresh," Karan said, his dark eyes fixed unblinkingly on the Chief Avionics Architect. "Tell Dr. Seshadri which processor was actually sitting inside the Mayajaal-80 black box aboard Garuda Zero-Four on the fifth of December."

The silence in the room suddenly turned suffocating.

Dr. Suresh Krishnamurthy looked down at the mahogany table.

"The Mayajaal-80 unit on the Block-I Garuda," Krishnamurthy said, his voice dropping so low that the engineers at the far end had to strain to hear him, "was running on a single 12-megahertz, 16-bit Brahma-16 microprocessor fabricated on the old 3.0-micron line."

Dr. Vikram Malhotra—the thirty-two-year-old Chief Processor Architect who had bled for two years to design the 32-bit Brahma-32—stared across the table at Krishnamurthy and Negi in sheer, stunned disbelief.

"A Brahma-16?!" Vikram Malhotra's voice echoed off the copper-shielded walls. "Suresh! We taped out the Brahma-16 in late 1973! We delivered first silicon on the 1.0-micron, 32-bit Brahma-32—with 180,000 transistors, a three-level hardware virtual memory unit, and an integrated hardware floating-point unit—in July 1976! Three and a half years ago! Why in the name of God is our forty-five-crore, fifth-generation flagship fighter flying into combat in December 1979 with a six-year-old, 16-bit processor from the MiG-25 Bharat programme bolted into its spine?!"

"Because when we froze the avionics sub-system architecture for the MiG-25 Bharat and the first forty Block-I Garudas in late 1977, Vikram," Siddharth Negi shot back defensively, his ears flushing red, "the Brahma-16 was flight-certified, radiation-tested, and had four million lines of proven assembly-language threat-library code already written for it! Our systems-integration committee concluded that an electronic-warfare jammer and a radar range-gate controller primarily execute fixed-point integer Look-Up Tables! You don't need 32-bit virtual memory addressing just to look up a radar frequency in a table and trigger a voltage-controlled oscillator! Our engineers argued that the 16-bit bus was completely optimized for the task and that migrating to a 32-bit processor was unnecessary overhead!"

"Unnecessary overhead," Karan repeated softly.

He did not shout. As everyone in Gorakhpur knew, when Karan Shergill's voice went quiet and still, someone's intellectual foundation was about to be demolished down to the bedrock.

Karan stood up from his chair, walked to the twelve-foot glass whiteboard on the wall, uncapped a black felt-tip marker, and wrote two numbers in six-inch-high characters:

216=65,5362^{16} = 65,536216=65,536

232=4,294,967,2962^{32} = 4,294,967,296232=4,294,967,296

"Look at the board, Siddharth," Karan said, turning back to the room with the marker in his hand. "And let every systems engineer in Shergill Aerospace and ISMC tattoo this arithmetic onto the inside of his eyelids so I never again hear the phrase '16-bit is optimized and 32-bit isn't necessary.'

"What is the maximum direct memory address space of a 16-bit processor?" Karan asked.

"Sixty-five thousand, five hundred and thirty-six bytes," Negi muttered. "Sixty-four kilobytes."

"Sixty-four kilobytes," Karan said. "In 1975, when we faced the old analog-digital hybrid AWG-9 radar on the American F-14 off Mauritius, or when we faced Soviet Flat Face and French Cyrano radars in Iraq this autumn, those enemy radars operated on fewer than sixty-four fixed crystal frequencies, and their pulse-repetition intervals followed simple, repeating mechanical patterns. Your entire threat-library Look-Up Table fit neatly inside forty-eight kilobytes of Brahma-16 RAM! So your 16-bit chip looked like a genius!

"Now look at what the Americans did between February 1975 and October 1979 while you were congratulating yourselves on how 'optimized' your 16-bit code was!"

Karan tapped the telemetry printout from Garuda Zero-Four.

"In 1975, Ruben Robertson at Hughes and Robert Everett at Westinghouse knew they couldn't beat our 1975 cleanrooms on raw micron size. So how did they design the APG-63D radar on Major Bong's F-15C? "First, instead of using one single-chip processor, they wired twelve 2.2-micron digital signal-processing arrays in parallel on a liquid-cooled backplane! "Second, instead of hopping across sixty-four predictable frequencies, they drove their frequency synthesizer with a 32-bit Galois Linear Feedback Shift Register generating a pseudo-random hop sequence across 1,800 discrete frequencies that does not repeat itself for four billion pulses!

"Now tell this room what happened inside Garuda Zero-Four's Mayajaal-80 box at 14:18:12 when Major Bong's radar started hopping across 1,800 frequencies, Siddharth!" Karan's voice rang like cold iron.

Dr. Vikram Malhotra looked at the telemetry chart and winced, seeing the mathematical catastrophe immediately.

"The waveform characterization matrix for a 1,800-frequency agile emitter," Malhotra said starkly, "requires at least four hundred and eighty kilobytes of active working memory to compute the phase-prediction polynomial in real time.

"Because the Brahma-16 can only address sixty-four kilobytes at a time," Malhotra looked straight at Negi and Krishnamurthy, "every four milliseconds, your 16-bit processor ran out of address space! It had to halt the jamming calculation, dump its registers, and execute an external memory-bank page swap over a 16-bit bus!

"And worse—because the Brahma-16 has no hardware floating-point unit, when it tried to run a Fast Fourier Transform on both the American F-15C's X-band hop and the Russian MiG-29's N-019 pulse simultaneously, it had to emulate floating-point math in software using 16-bit integer routines—which is eighteen times slower than hardware execution!

"Your jammer didn't lose because Gorakhpur's silicon is behind America's," Malhotra said, slamming his palm on the table. "Your jammer lost because you choked a 1979 supersonic fighter with a 64-kilobyte address bottleneck from 1973! During the eleven-millisecond dead window while your 16-bit chip was swapping memory banks, the American radar saw your un-modulated carrier wave, tagged you as a jammer, and locked an active missile onto your forehead!"

Siddharth Negi sat frozen in his chair, unable to utter a single word of rebuttal.

Karan walked back to his seat at the head of the table, setting the black marker down precisely parallel to his notebook.

"Do you see the disease that crept into this complex over the last two years?" Karan asked quietly, looking from face to face down both sides of the table.

"Complacency.

"In 1971, we were starving, we had five hand-built S-27 prototypes, and we worked twenty hours a day with our backs against the wall because we knew that if a single hydraulic line failed, India died. "By 1977 and 1978, after we humiliated the US Seventh Fleet off Mauritius, after we forced Moscow to beg for our Brahma-16 chips for their MiG-25s, and after we built the 1.0-micron IPL-3 while Intel was still struggling at 3.0-micron—we started believing our own press clippings!

"We looked at the Americans and said: 'Oh, Sunnyvale is still at 2.2-micron, we are years ahead of them, we don't need to go through the headache of re-qualifying our radar and electronic-warfare boards for the 32-bit Brahma-32! The 16-bit chip is good enough!'

"Let me tell you what happens in technology when the leader decides that yesterday's chip is 'good enough' because he thinks the runner behind him has short legs," Karan said coldly. "The runner behind you puts twelve of his 2.2-micron chips in parallel, writes leaner math because hunger makes men smart, and shoots your pilot out of the sky while you are admiring your laboratory trophies!"

"Chief Minister," Dr. Suresh Krishnamurthy said, standing up straight, his face set with the grim resolve of a great engineer who has taken a well-deserved lashing and knows his duty, "the lesson is learned, and it will never be repeated in my division. As of forty-eight hours ago, every single 16-bit board in the S-40 Garuda, the S-35 Tejas, the MiG-25 Bharat, and the SB-01 Airavata has been placed on a mandatory 90-Day Decommissioning Schedule.

"Using Harish Patel's Veda portable compiler," Krishnamurthy said, gesturing toward Patel across the table, "we recompiled the entire four-million-line Mayajaal and Trinetra-E threat-library and Doppler codebase from 16-bit to 32-bit Brahma-32 native code in seventy-two hours! Running on two parallel 32-bit Brahma-32 processors with hardware floating-point and four megabytes of direct linear RAM—zero bank-switching—our new Mayajaal-82 Digital RF Memory loop samples, predicts, and spoofs Major Bong's 1,800-frequency hop in 1.4 milliseconds!"

"Good," Karan said. "That solves the electrons. Now let us talk about the metal. Harsh, Gopal—stand up."

Dr. Harsh Vardhan and thirty-six-year-old Gopal Krishnan—the brilliant mechanical and nozzle-dynamics engineer who eight years earlier had designed the variable exhaust nozzles for the very first Kaveri-Alpha engine—stood up on the left side of the table.

"On the fifteenth of August, 1974," Karan said, looking directly at the two propulsion chiefs, "Air Marshal Prakash Mehra stood in the auditorium across the courtyard and watched Vikram Rathore fly the S-35 Tejas to twenty-seven degrees angle of attack. That same afternoon, we began detailed design on the twin-engine S-40 Garuda and the Kaveri Mk 3 variable-cycle turbofan.

"In March 1977, Gopal, you brought a blueprint to my office in Lucknow showing a Two-Dimensional Pitch-and-Yaw Thrust-Vectoring Nozzle for the Kaveri Mk 3—designed to deflect the 270-kilonewton exhaust plume twenty degrees up or down at forty-five degrees per second so the Garuda could point its nose in post-stall flight without losing energy.

"Why," Karan asked, his voice dropping to a razor edge, "were those thrust-vectoring nozzles sitting on a test stand in Mahendragiri on December 5th instead of bolted to the back of Flight Lieutenant Kabir Gill's engines when Colonel Viktor Pugachev pulled a 110-degree Cobra in front of his nose?"

Gopal Krishnan did not flinch or make excuses; he opened a heavy, soot-stained metallurgical test log and laid two cracked, heat-blued metal nozzle-actuation links on the Burma-teak table.

"Because in January of this year, Chief Minister, at 1,810 degrees Celsius turbine inlet temperature and 2,050 degrees Celsius phased-afterburner exhaust temperature, the physics of our own engine ate the mechanical linkages alive," Gopal Krishnan said plainly.

"Explain the failure mode," Karan ordered.

"Look at what we achieved with the Kaveri Mk 3 core, Sir," Dr. Harsh Vardhan said, stepping in beside Gopal. "By combining Dr. Vishwakarma's second-generation single-crystal Nickel-Chromium-Rhenium-Tantalum turbine blades with our variable-bypass cycle, each Kaveri Mk 3 pumps out 105 kilonewtons of dry thrust and 135 kilonewtons in afterburner—two hundred and seventy kilonewtons in a twin-engine fighter!

"When you try to deflect a 135-kilonewton supersonic plasma jet burning at 2,050 degrees Celsius through twenty degrees while the aircraft itself is pulling a 9-G turn," Gopal Krishnan continued, tapping the cracked niobium-alloy hinge link on the table, "the mechanical side-load on each nozzle flap exceeds fourteen metric tonnes!

"In our initial 1978 design, we built the vectoring divergent flaps and hinge-pins out of coated columbium-niobium refractory alloy. During the fifty-hour endurance qualification run at Mahendragiri in February 1979, two unacceptable things happened: "First: After fourteen consecutive high-G vectoring cycles in full afterburner, differential thermal expansion between the 2,000-degree inner flap liner and the 600-degree hydraulic hinge-pins caused the pins to gall and seize solid at a twelve-degree pitch-up angle! If a nozzle seizes at twelve degrees pitch-up at Mach 1.4 in combat, the aircraft tumbles into a thirty-G aerodynamic breakup in zero-point-eight seconds and kills the pilot! "Second: To keep the metal hinges from seizing, we had to widen the inter-petal expansion gaps by four millimetres—which caused a 6.4 percent supersonic exhaust gas leak that ruined our dry-thrust supercruise efficiency!"

"And when I brought that test report to Harsh in March," Gopal said, meeting Karan's eyes squarely, "with the Persian Gulf crisis already brewing on the horizon, we had a stark choice: either delay the entire forty-aircraft Block-I Garuda squadron deployment by nine months while we redesigned the nozzle from scratch, or ship the first forty Garudas with reliable, conventional axisymmetric nozzles that still gave us Mach 1.40 supercruise and a 1.30 thrust-to-weight ratio, while we solved the vectoring metallurgy permanently."

"That was my call, Karan," Harsh Vardhan said firmly. "I will not send an Indian pilot into the air with a nozzle hinge that can seize in afterburner."

Karan looked at the cracked niobium hinge link on the table, and then up at Gopal and Vardhan. The hard edge in his eyes softened by a fraction of an inch into professional respect.

"Shipping Block-I with fixed nozzles rather than risking a mid-air hinge seizure was the right engineering decision, Harsh," Karan acknowledged. "Now tell me where the permanent solution stands this morning. Because Colonel Pugachev's MiG-29 is already flying in Tabriz, and I will not give Moscow six months of bragging rights in the sub-500-knot envelope."

Gopal Krishnan reached into his leather case and pulled out a second component—a wide, interlocking, feather-shaped nozzle petal that looked as if it had been carved out of matte-black volcanic slate, yet when he tossed it across the table to Karan, Karan caught it with one hand and raised his eyebrows in surprise.

It weighed less than a third of the metal flap beside it.

"Carbon-Silicon-Carbide," Karan murmured, running his thumb over the microscopic laser-drilled cooling pores along the dovetail hinge edge.

"Continuous-fiber Woven Carbon-Fiber / Silicon-Carbide Ceramic Matrix Composite (C/SiC CMC), infiltrated by chemical vapour deposition in Dr. Seshadri's high-temperature furnaces here in Gorakhpur," Gopal Krishnan said, a fierce, proud grin finally breaking across his face.

"We threw away the refractory metal hinges completely in April," Gopal explained rapidly. "Look at the properties of that C/SiC ceramic-matrix petal in your hand, Sir: "One: Its coefficient of thermal expansion from minus fifty degrees to two thousand two hundred degrees Celsius is practically zero! It does not expand, it does not warp, and it cannot gall or seize! "Two: It is fifty-eight percent lighter than the niobium-titanium assembly—which shaves one hundred and四十 kilograms of dead weight right off the extreme tail end of the Garuda, improving the aircraft's center-of-gravity margin! "Three: Instead of heavy hydraulic actuators near the hot zone, we mounted four brushless samarium-cobalt electro-hydrostatic actuators forward on the cool titanium engine-bay ring, linked by carbon-composite pushrods and governed at 400 Hertz by a dedicated 32-bit Brahma-32 Digital Engine Control Unit integrated directly into the aircraft's quadruplex fly-by-wire computer!

"Four days ago—on Christmas Morning, December 25th—the Kaveri Mk 3B (Thrust-Vectoring Production Variant) completed its 500-hour, 3,000-cycle full-afterburner endurance qualification on Test Stand Two at Mahendragiri," Harsh Vardhan announced. "Plus-or-minus twenty-two degrees in pitch, plus-or-minus fifteen degrees in yaw, slew rate sixty degrees per second, and zero point zero percent seal leakage at Mach 1.4 supercruise!

"Even at eighty knots forward airspeed—where Colonel Pugachev's MiG-29 has zero rudder or elevator authority left and has to drop its nose to recover—a Block-II Garuda with the Kaveri Mk 3B can sit at seventy degrees angle of attack continuously, pivot its nose three hundred and sixty degrees in flat yaw using pure thrust vectoring, and shoot down anything in the sky!"

"How fast can Kanpur and Koraput build the retrofit kits for the forty Block-I Garudas?" Karan asked immediately.

"Twelve engine shipsets a week starting January 10th," Vardhan answered. "All forty Block-I aircraft in No. 1 and No. 45 Squadrons—and every new Block-II airframe rolling off the line—will have 3D thrust-vectoring nozzles installed and flight-cleared by March 31st, 1980."

"Approved," Karan said, writing the date down in blue ink.

Then Karan turned his chair forty-five degrees to face Dr. Seshadri and Dr. Krishnaswami on the ISMC side of the table.

"Which brings me to the fourth problem on my list," Karan said, his voice cooling again. "Earlier, when Vikram Malhotra asked why the avionics team was still using the 16-bit Brahma-16 in 1978 and 1979, Siddharth Negi gave us the software excuse.

"Now, Seshadri, give this room the hardware truth. Because even if Suresh Krishnamurthy had wanted to put two 32-bit Brahma-32 processors into every single Garuda, Tejas, Sarus helicopter, and Kaumodaki missile on the assembly line last year—your fabrication line could not have supplied the chips!"

Dr. Seshadri did not look away. As the man who had taken ISMC from a 10-micron shed in 1970 to a 1.0-micron powerhouse in 1976, he hated failure more than anyone in the building.

"On the commercial 1.0-micron Brahma-32 line—the standard bulk-silicon chips we use in our Chanakya-2 railway servers and industrial workstations," Seshadri said clearly, "our wafer yield today is sixty-eight percent. We are producing twelve thousand commercial Brahma-32 processors a month without breaking a sweat.

"The bottleneck is the Vajra-32 (Project King-32)—the Class-S Military and Space Radiation-Hardened variant required for high-altitude combat aircraft, nuclear missiles, and orbital satellites."

"What was the yield on the King-32 batch last month?" Karan asked.

"Nineteen point four percent," Seshadri replied grimly. "Out of two hundred die on a four-inch wafer, we are getting barely thirty-eight working military-grade processors."

"Nineteen percent after two years of production," Karan said. "Why?"

"Because of the Sapphire, Chief Minister!" Dr. Suresh Iyer—the thirty-eight-year-old ion-implantation specialist who had returned from Fairchild in 1972—spoke up beside Seshadri, pulling a crystal-lattice diffraction photograph from his folder.

"To make the King-32 immune to high-altitude cosmic-ray single-event upsets and nuclear electromagnetic pulse (EMP) latch-up," Iyer explained, "we followed the textbook radiation-hardening method: growing a thin epitaxial film of silicon on top of an insulating crystal wafer of synthetic Aluminium Oxide—Sapphire (Silicon-on-Sapphire / SOS).

"At 3.0-micron geometries on the old Brahma-16, Silicon-on-Sapphire worked fine! "At 1.0-micron on the Brahma-32—with 180,000 tightly packed transistors—it is a metallurgical nightmare! The crystal lattice spacing of sapphire differs from silicon by twelve percent! When we grow the silicon layer on top of the sapphire at a thousand degrees Celsius, that twelve-percent lattice mismatch generates millions of microscopic crystal twin-boundaries and Aluminium auto-doping defects right at the bottom of the 1.0-micron transistor channels! Eighty percent of our King-32 chips fail leakage testing before they ever leave the wafer probe!"

"And why are you still trying to force silicon to marry sapphire at one micron, Suresh," Karan asked, leaning forward over the diffraction photograph, "when you have the highest-current Oxygen Ion Implanters in the world sitting thirty yards downstairs in Zone Three?"

Suresh Iyer and Seshadri both stopped and stared at Karan.

"Wait," Iyer breathed, his eyes widening as the physics of what Karan was suggesting hit him like a thunderbolt. "Implant oxygen... directly into a pure silicon wafer?"

"Think about the geometry, Suresh!" Karan said, picking up his pen and drawing a cross-section of a pure, defect-free monocrystalline silicon wafer.

"Why do we use sapphire in the first place? Only because we want an electrically insulating layer underneath the top two hundred nanometres of active silicon so radiation cannot trigger parasitic thyristor latch-up through the substrate! "What is the best, cleanest electrical insulator in semiconductor physics—the exact insulator Sato and Ramesh have been growing on gates since 1971?"

"Silicon Dioxide (SiO2SiO_2SiO2​)," Seshadri said instantly. "Quartz glass."

"And how do you put a buried layer of pure Silicon Dioxide four hundred nanometres underneath the surface of a standard, crystal-perfect silicon wafer without ever touching a piece of sapphire?" Karan tapped the drawing with the nib of his pen:

"Separation by Implantation of Oxygen (SIMOX — Silicon-on-Insulator / SOI)! "You take a standard, zero-defect Czochralski silicon wafer. Using Suresh Iyer's high-beam-current ion implanter at 180 kiloelectron-volts, you shoot a massive dose of pure Oxygen-16 ions four hundred nanometres deep into the silicon crystal—leaving the top two hundred nanometres of surface silicon intact! "Then, you put the wafer into Prakash Reddy's high-temperature diffusion furnace at 1,320 degrees Celsius for four hours in an argon atmosphere! "What happens at 1,320 degrees? The implanted oxygen atoms chemically bond with the silicon four hundred nanometres down to form a continuous, buried, glass-smooth layer of pure stoichiometric Silicon Dioxide insulator—while the top two hundred nanometres of monocrystalline silicon anneals back into a 100% defect-free, zero-lattice-mismatch single crystal!"

Inside the conference room, Seshadri, Suresh Iyer, Ramesh Chandra, and Lev Abrahamov stared at the diagram on Karan's pad in dead, electrified silence.

In thirty seconds, Karan had just described the exact SIMOX Silicon-on-Insulator (SOI) process that would not be perfected in the West until the late 1980s and 1990s—using two machines (the high-current ion implanter and the high-temperature annealing furnace) that were already sitting on the floor of the Gorakhpur cleanroom!

"Zero sapphire," Dr. Seshadri whispered, his fingers flying across his calculator pad. "Zero Aluminium auto-doping. Zero twelve-percent lattice mismatch. And because the top layer is pure monocrystalline silicon isolated by buried oxide, parasitic junction capacitance drops by thirty-five percent—which means a SIMOX-SOI King-32 chip will not only be completely radiation-proof up to a million rads, it will run twenty-five percent faster at half the power consumption of a commercial bulk-silicon chip!"

"How long to calibrate the 180-keV oxygen implant dose and run the first twenty test wafers, Suresh?" Karan asked Iyer.

"Give me Prakash Reddy's furnace number four and three weeks, Chief Minister!" Suresh Iyer said, his eyes blazing with excitement. "If the buried oxide coalesces cleanly at 1,320 degrees, our yield on the military King-32 will jump from nineteen percent to sixty-plus percent by March!"

"Do it starting tonight," Karan ordered. "And that brings us to the big brother: Project Mahashakti—the 64-bit Brahma-64. Vikram?"

Dr. Vikram Malhotra stood up and placed a four-inch silicon wafer carrier on the table. Inside, gleaming like a microscopic rainbow grid under the ceiling lights, sat the 0.8-micron Brahma-64 engineering prototype wafer—each die packing 1.24 million transistors, a full 64-bit integer and floating-point data path, and R. D. Nair's 128-bit hardware cryptographic arithmetic co-processor.

"We have four prototype Brahma-64 cluster racks running downstairs right now on the early 0.8-micron pilot line," Malhotra reported honestly. "As you told the Cabinet in Delhi, those four racks ran the ten-times-resolution finite-element thermal and electromagnetic simulations that saved Project Marut's wing-root from micro-cracking.

"However, to move the Brahma-64 from a laboratory prototype running at 14 megahertz and 21% yield to a fully optimized, mass-production 64-bit microprocessor running at 25 megahertz and 60% yield for our Marut stealth bombers, our Ganesh-4 supercomputers, and our next-generation radar arrays, we have hit two physical walls at 0.8-micron that need an immediate architectural and materials overhaul: "Wall One is Interconnect Resistance and Electromigration: at 0.8-micron, our two layers of standard sputtered Aluminium-Silicon wiring are so thin that high current densities literally push the aluminium atoms out of place and create RC signal delays that cap our clock speed at 14 megahertz! Mohan Krishnan needs a Three-Layer Tungsten-Silicide / Copper-Doped Metallization Process with Chemical-Mechanical Planarization (CMP) to polish each insulating layer flat as a mirror between wiring levels! "Wall Two is Lithography Throughput and Overlay at Sub-Micron—which is Dr. Ghosh's domain."

"How much capital do you, Mohan Krishnan, and Seshadri need to solve the three-layer planarized interconnect and bring the fully optimized 25-megahertz Brahma-64 to full production readiness?" Karan asked.

"Three hundred and eighty crore rupees over fourteen months for the new planarization bays, the multi-chamber sputtering tools, and the mask verification cluster," Malhotra answered without hesitation. "With that funding, we will deliver the fully optimized, production-qualified Brahma-64 (Rev-C) at 25 megahertz—and the SIMOX radiation-hardened Vajra-64—by September 1, 1980."

Karan uncapped his fountain pen, pulled Malhotra's capital requisition sheet across the wood, crossed out the number ₹380 Crore, wrote ₹450 Crore — Priority One Sovereign Allocation, and signed his name beneath it.

"You have four hundred and fifty crores in the bank at nine o'clock Monday morning, Vikram," Karan said. "And I want the first flight-qualified Brahma-64 boards bolted into the avionics bay of the Marut stealth bomber prototype on the day it rolls out of the hangar in October."

At the far end of the table, fifty-four-year-old Dr. Subhash Ghosh—the quiet, spectacles-wearing former CSIR optical physicist who in October 1970 had been handed a blank cheque to build India's first lens-polishing workshop from scratch—cleared his throat and stood up alongside Dr. Rajesh Mehta (Plasma & Laser Physics) and thirty-one-year-old Dr. Ashwin Deshpande (Photoresist Chemistry), with veteran Japanese optical master Sato Hiroshi nodding beside them.

"Chief Minister," Dr. Ghosh said in his calm, measured voice, "Vikram just mentioned Wall Two: sub-micron lithography throughput and overlay accuracy. I believe it is time we showed you what is sitting behind the locked doors of Bay Seven in the Optics Building."

Ghosh unrolled a six-foot-wide blue architectural blueprint across the center of the table.

Titled across the bottom block in clean Devanagari and English stencil was:

INDIAN PROJECTION LITHOGRAPHY — SYSTEM IPL-4S (193nm DUV STEP-AND-SCAN) / PRODUCTION PROTOTYPE.

Every engineer on the Aerospace side of the table—Krishnamurthy, Vardhan, Rao—leaned over to stare at the drawing, because they knew that everything they built in the sky lived or died by the minimum feature size Dr. Ghosh's lenses could print on a piece of silicon.

"Let me remind the room where the rest of the world stands in lithography in December 1979," Dr. Ghosh said, without a trace of boastfulness—simply stating physical facts.

"In the United States today, GCA Corporation and Perkin-Elmer are selling their latest DSW 4800 wafer steppers to Intel, Motorola, and Texas Instruments—and to Dr. Currie's Programme Zero defense fabs. Those American machines use a conventional 436-nanometre (g-line) or 365-nanometre (i-line) mercury arc lamp. Because of the diffraction limit of 436-nanometre light, the Americans are struggling to print 2.0-micron to 2.2-micron lines, and they believe 1.0-micron optical lithography is physically impossible before 1986!

"Here at ISMC, we built the IPL-3 in 1975 using Dr. Mehta's ESU-1 (248-nanometre Krypton-Fluoride excimer discharge source) and gave this complex a 1.0-micron production line four years ago.

"In August 1976, Chief Minister, you authorized us to begin the IPL-4: dropping the wavelength down to 193 nanometres (Deep Ultraviolet) using an Argon-Fluoride (ArF) Excimer Laser, targeting a 0.7-to-0.8-micron process."

"Which is currently printing the prototype Brahma-64 wafers at 0.8-micron," Karan said, studying the complex blueprint on the table with growing intensity, "at a throughput of only fourteen wafers an hour because the full-field fused-silica lens elements suffer from chromatic aberration and focal-plane curvature at the edges of a twenty-millimetre die. How did you solve the lens edge distortion, Subhash?"

Dr. Ghosh smiled—the rare, incandescent smile of an optical master who has just cracked the riddle that would govern semiconductor manufacturing for the next thirty years.

"We stopped trying to expose the entire twenty-millimetre chip field in a single flash through the outer edges of the lens, Sir!" Ghosh said, pointing to the center of the blueprint.

"Sixteen months ago, Sato-san, Balakrishnan, and I realized that in any ultra-high-numerical-aperture (0.62 NA) Calcium-Fluoride and Fused-Silica lens assembly operating at 193 nanometres, there is a narrow, crescent-shaped 'Sweet Spot' (a 26-millimetre by 4-millimetre slit) right across the optical center of the lens where geometric aberration, astigmatism, and distortion are virtually zero!

"Instead of stepping and flashing the whole die at once (Step-and-Repeat), we invented Synchronized Microlithographic Step-and-Scan! "While the wafer stage moves south beneath the lens at two hundred millimetres per second, the photomask reticle stage above the lens moves north at eight hundred millimetres per second (a 4-to-1 demagnification scan)—exposing the chip smoothly through that aberration-free four-millimetre optical slit!"

Karan's eyes locked onto the stage mechanism drawing.

"To scan a photomask at 800 millimetres per second and a silicon wafer at 200 millimetres per second in opposite directions while printing sub-micron lines without blurring them," Karan said, testing the engineering to its core, "your mechanical synchronization between the upper and lower stages has to hold inside fifteen nanometres! Any mechanical ball-screw or oiled rail will vibrate ten times worse than that!"

"There are no mechanical ball-screws and no oiled rails in the IPL-4S, Chief Minister," Sato Hiroshi spoke up with quiet Japanese pride. "Both the reticle stage and the wafer stage float with zero physical contact on cushion pads of ultra-pure nitrogen gas (Aerostatic Magnetic-Levitation Linear Motors)! And Subramaniam's Six-Axis Dual-Frequency Helium-Neon Laser Interferometer measures the relative position of the mask and the wafer ten million times a second, driving piezo-ceramic micro-actuators that lock the synchronization error to 3.8 nanometres!"

Then Dr. Rajesh Mehta and Dr. Ashwin Deshpande tapped the upper and lower blocks of the schematic:

"On the light source," Dr. Mehta said, "our new ESU-2 Argon-Fluoride 193-nanometre Excimer Laser—incorporating the closed-loop cryogenic fluorine-halide gas purifier and solid-state thyratron pulse-power modulator we began in 1977—now fires at 4,000 pulses per second with a spectral bandwidth narrowed by intracavity prisms to less than 0.5 picometres!

"And on the wafer surface," thirty-one-year-old Ashwin Deshpande added, holding up a sealed amber glass bottle containing a clear fluid that represented three years of synthetic organic chemistry, "we solved the 193-nanometre aromatic absorption wall! Conventional benzene-ring polymers are opaque at 193 nanometres. So my team synthesized a completely new Alicyclic Adamantane-Methacrylate Polymer Backbone paired with a Triphenylsulfonium Nonafllate Photoacid Generator (PAG)! One single 193-nanometre photon releases an acid proton that catalytically cleaves four hundred protective ester groups during post-exposure bake with a reaction-front diffusion blur of less than eight nanometres!"

Dr. Ghosh placed a single, six-inch-diameter mirror-bright silicon test wafer—pulled from the IPL-4S prototype in Bay Seven at five o'clock that morning—alongside a scanning-electron-microscope (SEM) photomicrograph magnified fifty thousand times, directly in front of Karan.

Karan picked up the SEM photograph.

Printed across the silicon in razor-straight, vertical-walled canyons with zero standing-wave ripples were transistor gate lines measuring 0.50 microns (500 nanometres)—half a micron wide, with a wafer throughput of sixty-two wafers per hour and an overlay accuracy of 18 nanometres!

"Let me state for the record what is sitting in Bay Seven today, Chief Minister," Dr. Krishnaswami said quietly from his seat.

"With the IPL-4S, ISMC has not only guaranteed that the 0.8-micron Brahma-64 will hit seventy-percent commercial yield by next summer—we have already built and validated the 0.5-micron (500-nanometre) production lithography machine for 1981 and 1982!

"Not a single lens element, not a single laser tube, not a single gram of photoresist, and not a single line of stage-control code came from outside India. The Indian Semiconductor Manufacturing Corporation is today the first and only manufacturer of Deep Ultraviolet Step-and-Scan Lithography Machines on the face of the earth—and we are a full decade ahead of anything in Silicon Valley, Tokyo, or Moscow."

Karan looked down at the 500-nanometre electron-microscope lines for five seconds, and then looked up at Ghosh, Mehta, Deshpande, and Sato.

"Build twelve IPL-4S step-and-scan machines in the secure machine shop between January and August," Karan commanded. "Install eight of them in Fab Two right here in Gorakhpur for the Brahma-64 and our next-generation memory arrays. And R. D. Nair—" Karan glanced at the security chief "—the IPL-4S building stays at Category-Zero Classification. If a foreign satellite or a foreign attaché even learns the words 193-nanometre Step-and-Scan before 1988, someone in this room has failed."

"Now," Karan said, turning to the final quartet at the table—Rajiv Menon, Dr. Sunita Rao, Dr. Arvind Bose, and Dr. Rangan—"show me the civilian harvest that is paying the bills for all of this. Five years ago next month—on January 22, 1975—we launched the Siddharth-1 simultaneously in Bombay and San Francisco at ₹4,800 and $395. Where does the Siddharth stand at the close of 1979, and what is ready for the 1980 launch?"

Thirty-seven-year-old Rajiv Menon stood up and walked to the side table, where two machines sat draped under white cotton dust-covers, while opening the five-year consolidated balance sheet of Shergill Computer Systems.

"Chief Minister," Rajiv Menon said, his voice brimming with the quiet electricity of a man who had watched a revolution unfold from his original eighteen-month specification sheet, "in January 1975, we sold 190 Siddharth-1 computers on Day One—including six units to a twenty-year-old kid named Steve Jobs in a garage in Cupertino who wrote to Dr. Bose and Sunita every month for two years until we hired him as our California educational-distribution consultant!

"Look at what happened over the five years between January 1975 and December 1979: "In 1976, we released the Siddharth-2 (powered by our 3-micron Brahma-16 at 6 megahertz, replacing the audio cassette with an internal 5.25-inch floppy disk drive). "In 1978, we released the Siddharth-3 (running on the 1.0-micron Brahma-16C at 10 megahertz, 256 kilobytes of DRAM, high-resolution colour display, and Sunita Rao's Veda-DOS and Siddharth Office Suite—word processing in twelve Indian languages plus English, French, German, and Japanese, alongside our Lekha electronic spreadsheet!).

"In the twelve months of 1979 alone," Menon read from the audited ledger, "Shergill Computer Systems shipped one million, four hundred and twenty thousand (1.42 million) Siddharth personal computers worldwide!

"In the United States and Western Europe, the Siddharth-3 holds a fifty-eight percent market share of all personal and small-business desktop computers—generating $1.18 billion (₹885 Crore) in hard-currency export revenue!"In India, thanks to the Fifteen-Percent Educational Discount that Amma-ji insisted on in 1974 and the Siddharth Free Software Library, eighty-four thousand Indian secondary schools, polytechnics, IITs, and district post offices now have a Siddharth computer lab! Four million Indian teenagers are learning to write code in BASIC and Veda every afternoon before they even enter college!"And on the enterprise backbone," Dr. Rangan added beside Menon, "Chanakya-2 running on our Brahma-32 servers now processes one hundred percent of the Indian Railways reservation network across 640 stations with a query time of 0.4 seconds, plus the clearing ledgers of the State Bank of India, the Bombay Stock Exchange, and the Port of Singapore Authority!"

Then Rajiv Menon, Dr. Arvind Bose, and thirty-three-year-old Dr. Sunita Rao stepped to the side table and pulled the white cotton cover off the machine sitting at the center.

It did not look like a 1970s hobbyist box, and it was not hooked up to a television set.

It was a sleek, sculpted, warm pearl-grey desktop workstation with a built-in fourteen-inch, 800-by-600-pixel high-resolution Trinitron-grade colour CRT monitor, a slim detachable mechanical keyboard, two internal 3.5-inch hard-shell micro-floppy drives alongside an internal 10-Megabyte Winchester hard disk, and—resting on a smooth polymer pad to the right of the keyboard—a palm-sized, contoured ABS-plastic block with two buttons on top and a rubber-coated steel ball rolling underneath.

"For the fifth anniversary of the Siddharth-1 on January 22, 1980," Dr. Arvind Bose said proudly, flipping the side power switch, "we present the Shergill Siddharth-4 (The Viswakarma 32-Bit Personal & Engineering Workstation)."

In less than 0.8 seconds, the fourteen-inch colour screen chimed and lit up—not with a black screen and a blinking green command-line prompt, but with a crisp, pale-blue graphical desktop displaying overlapping rectangular Windows, pull-down Menus, miniature pictorial Icons representing folders, spreadsheets, and circuit diagrams, and a sharp white arrow cursor!

Dr. Sunita Rao stepped up to the desk, rested her right hand on the small palm-sized plastic block on the pad, moved her wrist two inches to the right—and on the screen, the white arrow glided smoothly across the blue desktop, clicked twice on an icon labeled CAD-3D / GARUDA WING, and opened a full, rotating three-dimensional wire-frame and shaded-surface model of an aircraft wing in less than a second!

"Powered by a 12-megahertz, 1.0-micron 32-bit Brahma-32 microprocessor and One Megabyte of standard DRAM," Sunita Rao explained, her eyes shining with the quiet brilliance of the mathematician who had written the 8KB BASIC interpreter five years ago, "the Siddharth-4 runs Chitra-OS v1.0—written entirely in Harish Patel's Veda-Plus object-oriented systems language!

"In 1975, we made a machine that anyone could use if they learned five lines of BASIC," Sunita said, clicking the small handheld controller—which the Gorakhpur engineers had named the Mushak (The Mouse, vehicle of Lord Ganesha)—to open a multilingual document in Devanagari, Tamil, and English side-by-side with a live financial spreadsheet. "In 1980, with Chitra-OS and the Mushak pointer, a seven-year-old child in a village school in Bihar or an architect in New York doesn't even have to type a single command—he simply points at what he wants on the screen, clicks his finger, and the 32-bit processor executes it at three million instructions per second!

"Retail price in India for universities and businesses: ₹9,500. Export price in the United States and Europe: $1,195—one-tenth the price of a Xerox Alto experimental prototype, with four times the speed!"

Karan stood looking at the glowing graphical screen of the Siddharth-4 for a long, quiet moment.

In the life he remembered, a 32-bit personal computer with 1 Megabyte of RAM, a 10MB hard drive, a full colour Graphical User Interface, and a mouse for $1,195 had not arrived on the desks of ordinary human beings until the late 1980s! Because Gorakhpur had reached 1.0-micron in 1976 and built its own memory, its own 32-bit chips, and its own operating system in-house, India was about to drop the late-1980s computing revolution onto the world on January 22, 1980.

"Ship fifty thousand units in the first quarter," Karan told Rajiv Menon,Sunita, and Bose. "And give every IIT, every regional engineering college, and every polytechnic in India twenty Siddharth-4 workstations free of charge from the Shergill Foundation before the spring semester opens."

Then Karan looked down to the far end of the table, where fifty-six-year-old R. D. Nair, Chief of ISMC Internal Security, sat with a thick, red-bordered folder stamped TOP SECRET — R&AW / MEIT EXPORT INTERDICTION: PROJECT GANESH-3.

"And now," Karan said, sitting back down and resting his hands on the arms of his chair, "let us settle the last item on today's agenda: the Ganesh-3 Supercomputer and our sudden popularity with every spy agency on earth."

R. D. Nair opened the red dossier.

Five years had passed since May 5th, 1974, when the original Ganesh-1—a forty-eight-processor, 3-micron parallel computing mainframe installed in the basement of BARC at Trombay—had processed two hundred and forty-seven channels of live nuclear-detonation telemetry from the Pokhran desert and calculated the 11.4-kiloton yield of Smiling Buddha in ninety-one seconds flat.

Since 1974, Dr. Vikram Malhotra and Dr. Arvind Bose had evolved that parallel architecture into the Shergill Ganesh-3 Vector-Parallel Supercomputer: a liquid-cooled, cylindrical cabinet housing sixty-four parallel 32-bit Brahma-32 processors and sixty-four vector floating-point pipelines, delivering a sustained speed of 165 MegaFLOPS (165 million floating-point operations per second)—more than double the sustained real-world speed of the American Cray-1, at half the electrical power and forty percent lower cost!

Over the past eighteen months, Shergill Computer Systems had leased or sold thirty-two Ganesh-3 supercomputers to the premier scientific universities and civilian meteorology centers of the Western and unaligned world—including MIT, Caltech, Stanford, Oxford, ETH Zurich, the University of Tokyo, and Singapore—while providing forty more on nominal ₹1-per-year academic leases to every IIT, IISc Bangalore, TIFR, and the top regional engineering colleges across India.

"Chief Minister," R. D. Nair began in his crisp, unsparing intelligence officer's voice, "eleven days ago, during the Muscat preparations, the US State Department and the Department of Commerce delivered an urgent formal diplomatic protest under the technology-security clauses of the Geneva and New Delhi accords—and simultaneously, Mr. R. N. Kao at R&AW instructed us to place an immediate temporary hold on the twelve pending overseas Ganesh-3 export shipments scheduled for January."

"Read the five intercepts to the room, Nair," Karan said calmly, "so our engineers understand who has been shopping for their hardware."

Nair held up five decoded counter-intelligence files:

"Over the past one hundred and twenty days," Nair read aloud, "R&AW's technical-monitoring stations and our own ISMC field-service engineers uncovered five separate, state-directed clandestine operations attempting to illegally divert or smuggle a civilian Ganesh-3 supercomputer into a foreign nuclear-weapons or military-aerospace complex:

"Case One — The Soviet Union (GRU Directorate T): In October, a Ganesh-3 leased to the Department of Atmospheric Physics at the University of Helsinki in Finland was scheduled for a weekend maintenance shutdown. At two in the morning on Sunday, our resident ISMC field engineer in Helsinki walked into the computer center unannounced—and caught six Soviet GRU technical officers and two bribed Finnish logistics clerks in the act of unbolting the Ganesh-3's central processor cabinet to load it onto a diplomatic truck bound across the border to Leningrad, where the Krylov Shipbuilding Institute and TsAGI wanted it to model nuclear-submarine acoustic cavitation and reverse-engineer our MiG-25 Bharat Doppler compiler!

"Case Two — Pakistan (ISI / Project 706): In November, a West German textile-machinery holding company in Düsseldorf placed an $8.5 million cash order for a Ganesh-3, ostensibly for 'synthetic-polymer molecular modeling.' R&AW's station in Frankfurt traced the holding company's bank guarantee back through Luxembourg to a front account controlled by Ghulam Ishaq Khan's special works cell in Rawalpindi—who intended to trans-ship the crated supercomputer from Frankfurt to Karachi on a chartered cargo jet so Pakistani physicists could run spherical hydrodynamic implosion simulations that they cannot do by hand!

"Case Three — North Korea (Reconnaissance Bureau): Three weeks ago, a newly registered 'East Asian Maritime Meteorological Institute' in Macao tried to lease a Ganesh-3 through a Hong Kong broker—with the shipping manifest secretly pre-booked onto a North Korean freighter bound for Nampo and Pyongyang to calculate two-stage ballistic missile re-entry trajectories!

"Case Four — Apartheid South Africa (BOSS / Armscor): Attempted to purchase a second-hand Ganesh-2 unit through a private mining-geology institute in Lausanne for covert shipment to Pelindaba outside Pretoria!

"And Case Five — Israel (LAKAM Scientific Intelligence):" Nair looked up over his spectacles. "Even our close friends in Tel Aviv could not resist temptation. Under the July 1979 Accord, we leased two Ganesh-3 systems to the Technion in Haifa and the Weizmann Institute purely for civilian desalination and academic physics. On December 2nd, our diagnostic telemetry technician in Haifa discovered that the Weizmann unit's midnight processing queue was being remotely routed over a buried military coaxial line straight south into the Negev desert—where physicists at Dimona were using our sixty-four parallel Brahma-32 processors to run thermonuclear two-stage radiation-implosion calculations without paying for a defense license or notifying New Delhi!"

Around the mahogany table, the engineers sat in stunned silence.

"Washington's complaint," Dr. Krishnaswami said heavily, rubbing his forehead, "is that because the Ganesh-3 is the fastest vector-parallel supercomputer in the world, selling it to universities in Europe and Asia has created a global proliferation hazard—and the Commerce Department in Washington is demanding that we either subject every single Ganesh export to a joint US–Indian COCOM veto committee, or halt foreign university deployments outside India!"

"And if we halt foreign deployments or give Washington a veto over our sales," Rajiv Menon protested immediately, "we throw away $600 million a year in 75%-margin supercomputer leases—and we hand MIT, Stanford, Oxford, and CERN right back to Cray Research and IBM on a silver platter!"

"We are not halting foreign sales for a single week, and we will never give Washington or anyone else a veto over a machine built in Gorakhpur," Karan Shergill said, his voice ringing with quiet, absolute authority.

Everyone turned to look at him.

"Look at the problem like engineers, not bureaucrats," Karan said, leaning forward with a cold, razor-sharp smile. "Why could the Soviet GRU in Helsinki, or the Pakistanis in Düsseldorf, or the Israelis in Haifa even think about unbolting a Ganesh-3 rack or feeding nuclear-implosion code into it at midnight?

"Because we sold or leased them the box like a dumb refrigerator and walked away!

"Starting January 15th, 1980, we lift the export hold—and every Ganesh-3 and upcoming 64-bit Ganesh-4 supercomputer deployed anywhere outside the borders of the Republic of India will be governed by the Lakshman-Rekha Hardware & Orbital Tether Protocol:"

Karan ticked off the four physical and cryptographic locks to Vikram Malhotra, Arvind Bose, Dr. Rangan, and R. D. Nair:

"Lock One — Zero Outright Sales; Sovereign Glass-Room Lease Only: "No foreign university, corporation, or government on earth ever 'buys' title to a Ganesh supercomputer. They lease computational cycles! The physical machine remains the sovereign property of Shergill Computer Systems, bolted to a seismic concrete pier inside a sealed, glass-walled room on the university campus accessed only by two cleared Indian resident engineers from R. D. Nair's division!

"Lock Two — The 40-Millisecond Thermite-Fluorine Tilt & Tamper Fuse: "What happens if the Soviet GRU in Helsinki or a commando team in Macao overpowers our two resident engineers at two in the morning and tries to unbolt the cabinet to load it onto a truck?" Karan looked at Dr. Seshadri and Dr. Arvind Bose. "Inside the hermetically sealed, pressurized nitrogen-argon CPU backplane of every export Ganesh supercomputer, we wire a three-axis laser-gyroscopic tilt sensor, a barometric pressure-drop switch, and an optical fiber continuity loop through the floor anchor-bolts! "And etched directly onto the silicon passivation layer of all sixty-four Brahma-32 CPU dies inside that cabinet is a microscopic Zirconium-Perchlorate / Polytetrafluoroethylene Pyrophoric Fuse Grid, powered by an internal, non-removable lithium-thionyl-chloride capacitor! "If the cabinet pressure drops by half a pound per square inch, if a single floor bolt is cut, or if the chassis tilts by zero-point-five degrees without the dual cryptographic master-key from Gorakhpur—that internal capacitor fires in forty milliseconds! There is no explosion outside the cabinet to hurt anyone in the room—just a silent, two-thousand-degree flash inside the sealed ceramic packages that turns all sixty-four microprocessor dies into melted, featureless sand before the thieves can even get the dolly through the door!"

A low murmur of sheer, awed appreciation rippled around the table.

"Lock Three — The Speed-of-Light Orbital Heartbeat (Defeating Remote Cable Routing & Smuggling):" Karan turned to R. D. Nair and Dr. Rangan.

"Now, what if the customer doesn't move the box, or what if someone spoofs a shipping manifest—how do we know with mathematical certainty every morning that the Ganesh-3 leased to Oxford is sitting in Oxford, and not in Pretoria or Rawalpindi, and that its data bus hasn't been tapped by a two-hundred-kilometre coaxial line to Dimona?"

Karan tapped Nair's 128-bit RSA Hardware Encryption Chip schematic on the table.

"We connect every exported Ganesh supercomputer to a mandatory roof-mounted Ku-Band Satellite Transceiver locked onto our ISRO INSAT and Rohini geostationary and low-earth-orbit satellites! "Every six hours, the master clock inside the Ganesh CPU must complete a 128-bit encrypted challenge-response handshake with three separate Indian satellites overhead to renew its volatile SRAM microcode execution key! "If that satellite handshake is missed twice—the volatile SRAM key wipes itself to zeros, and the supercomputer becomes a four-ton paperweight! "And here is the physics lock that no spy agency on earth can fake," Karan smiled thinly: "The Speed of Light (299,792 kilometres per second299,792\text{ kilometres per second}299,792 kilometres per second) is a constant of the universe! "When our three satellites ping the Ganesh CPU in Haifa or Helsinki, the satellite transponders measure the exact nanosecond round-trip time-of-flight delay of the encrypted pulse! "By triangulating the nanosecond time-of-flight from three orbital points, Gorakhpur knows the exact physical latitude, longitude, and altitude of that CPU silicon die to within eight metres!And if the Weizmann Institute in Haifa tries to route the terminal I/O over a fifty-mile cable to Dimona in the Negev, the extra 500 microseconds of copper-wire propagation delay on the terminal bus triggers an immediate red flag on R. D. Nair's console in Gorakhpur and shuts the port down!"

"And Lock Four — Instruction-Matrix Auditing in the Hypervisor:" Karan looked at Harish Patel and Sunita Rao.

"Finally, burnt permanently into the un-editable mask-ROM of the GaneshVeda-Hypervisor is a background instruction-signature monitor. "When a university uses a Ganesh-3 for protein folding, weather forecasting, or civil bridge stress, the matrix operations follow standard fluid-dynamic or linear-algebra patterns. "When someone tries to run a spherical radiation-hydrodynamics nuclear implosion code or a Monte Carlo neutron-transport criticality code—the exact mathematics that Ganesh-1 ran for Dr. Ramanna and Dr. Chidambaram on May 5th, 1974—the coupled differential-equation matrix signature is unmistakable! "The microsecond an unauthorized foreign machine attempts to compile or execute a nuclear-hydrodynamics or phased-array radar-jamming matrix, the hypervisor silently corrupts the fifth decimal place of every floating-point result—feeding their weapons designers plausible-looking mathematical garbage that will make their warhead fizzle on the test stand—while transmitting a silent alert to R&AW on the next six-hour satellite pass!"

For ten full seconds after Karan finished speaking, the fifteen sharpest scientific and engineering minds in the Republic of India sat looking at the twenty-eight-year-old man at the head of the table.

Not a single loose end remained.

If you tried to unbolt or steal a Ganesh supercomputer, it melted its own silicon into glass in 40 milliseconds.If you tried to move it or tap it from a secret base fifty miles away, the speed of light from space caught you to within eight metres and shut the machine off.And if you tried to use it in place to secretly design a nuclear bomb against India, the machine smiled politely, took your lease money, reported you to R&AW, and subtly poisoned your fifth decimal place so your bomb wouldn't work!

R. D. Nair—a man who had spent twenty-five years in the shadows of intelligence and rarely smiled more than once a year—slowly closed his red dossier, a look of pure, unadulterated professional reverence on his face.

"If we send the technical specification of that four-layer lock to Secretary Vance and the Commerce Department in Washington on Monday morning, Chief Minister," Nair said quietly, "the Americans won't just withdraw their protest—the Pentagon will beg us to install the Lakshman-Rekha chip on their own Cray exports so the Russians stop stealing their machines! And as for Case Five in Haifa... I would pay a month's pension just to see the faces of the physicists in Dimona three months from now when they figure out why their stolen night-shift equations don't add up."

A wave of dry, electric laughter rippled around the long teak table, breaking the tension of the four-hour audit.

Karan stood up at the head of the table, closing his leather binder as the wall clock above the door clicked to 1830 Hours.

Outside the Faraday-shielded building, the winter sun had set over eastern Uttar Pradesh, and across the three thousand acres of the Gorakhpur Industrial Complex, the high-pressure sodium floodlights had come on—illuminating the endless rows of cleanrooms, the jet-engine test cells, the titan-presses of the aerospace hangars, and the railway sidings where freight trains loaded with Siddharth computers and Jawan televisions were pulling out toward Bombay and Calcutta.

Karan looked down the two rows of engineers standing around the table—the men and women who had come back from Dallas, from California, from Tokyo, from Haifa, and from the IITs and regional colleges of India to build a civilization that bowed to no one.

"Let us review the calendar we have just locked into steel for 1980," Karan said, his voice clear and steady:

"January 15, 1980: All 16-bit Brahma-16 avionics and EW boards begin mandatory replacement across the Air Force and Navy with the dual-parallel 32-bit Brahma-32 (Mayajaal-82) running at a 1.4-millisecond hop cycle; the Lakshman-Rekha orbital-tethered Ganesh-3 supercomputer leases resume globally."January 22, 1980 (Fifth Anniversary of the Siddharth-1): Simultaneous global launch in Bombay, New Delhi, San Francisco, London, and Tokyo of the 32-bit Siddharth-4 (Viswakarma) workstation with Chitra-OS and the Mushak graphical interface."March 31, 1980: Full fleet retrofit and Block-II rollout of the Kaveri Mk 3B 2D/3D Ceramic-Matrix Thrust-Vectoring Nozzles on the S-40 Garuda, alongside the 60%-yield SIMOX Silicon-on-Insulator King-32 military processor line."August 15, 1980: Operational commissioning of eight indigenous 193-nanometre IPL-4S Deep Ultraviolet Step-and-Scan Lithography Systems in Fab Two, taking Gorakhpur into mass-production 0.5-micron (500-nanometre) silicon."September 1, 1980: Full production qualification of the 25-megahertz, 64-bit Brahma-64 microprocessor and the Ganesh-4 (1-GigaFLOP) supercomputer."And in October and December 1980: Roll-out and first flight of Project Marut—India's continuous-curvature strategic stealth bomber."

Karan picked up the scorched, R-73-shrapnel-punched piece of composite tailfin from Flight Lieutenant Kabir Gill's Garuda off the table and handed it to Dr. Krishnaswami and Dr. Suresh Krishnamurthy.

"Hang this scorched piece of tailfin on the wall right inside the main entrance of the Aashram lobby tonight," Karan said softly.

"Put a brass plaque underneath it that says: Qezel Ozan Gorge — 5 December 1979. The day we thought yesterday's work was good enough.

"I want every twenty-two-year-old apprentice and every fifty-year-old division chief who walks through that door tomorrow morning to touch those three shrapnel holes before they sit down at their desks," Karan said, looking each of them in the eye. "We are entering the 1980s as the undisputed technological leader of the world. The only people on earth who can defeat us now are sitting inside this room—if we ever slow down.

"Go home to your families for New Year's weekend, gentlemen, and Dr. Rao," Karan smiled, picking up his coat. "On Tuesday morning, January Second, 1980—we start running again."

End of Chapter 421

More Chapters