Chapter 250: The Distance Remaining
The Pentagon, Arlington, Virginia — October 1976
The room still had no windows.
This was the same room — the same compartmented conference space on the same corridor of the Pentagon where, twenty months earlier, James Schlesinger had stood at the head of the table and told twenty-three men that an Indian aerospace and electronics conglomerate had built something in Gorakhpur that had made the most expensive carrier group in the history of naval aviation turn around and go home. The room had not changed. The table was the same mahogany. The chairs were the same. The security detail outside was different personnel but the same process.
What had changed was everything else.
Secretary of Defense Donald Rumsfeld arrived at 0800, which was his habit. He had been in the job since November 1975, following Gerald Ford's elevation from Vice President and the subsequent cabinet reorganisation that had moved Schlesinger out and Rumsfeld in. Rumsfeld was forty-four years old, a former naval aviator turned congressman turned Ambassador to NATO turned White House chief of staff, and he brought to the job a quality that was different from Schlesinger's analytical economist approach — he was faster, more aggressive, less interested in comprehensive frameworks and more interested in specific answers to specific questions, and the specific question he had been asking every programme director in the defence establishment since his first week in office was: where are we and what is it going to take.
The answers, in October 1976, were considerably better than anyone in February 1975 had dared project.
He sat at the head of the table. He did not open with preamble.
"I read the twenty-month progress reports on all six programmes last week," he said. "I have questions. Some of the answers I already know and I'm asking anyway to see whether I get the same answer in this room that I got from the paper. Others I genuinely don't know." He looked at Dr. Malcolm Currie, who had survived the transition from Schlesinger's team and was still Director of Defense Research and Engineering, and who was therefore still, functionally, the most important scientist in the American defence establishment. "Currie. Programme Zero first. Tell me where we are."
Currie had been preparing for this meeting for three weeks. Not preparing in the sense of managing his presentation — he had been preparing in the sense of actually knowing the answer, which required going to the facilities himself rather than reading the programme manager's summaries, because he had learned in forty years of working in science and engineering that the gap between what programme managers wrote in progress reports and what was actually happening in the laboratory was frequently the most important information available.
He had visited Sunnyvale and Fort Monmouth and the classified fabrication facility in Arizona in the preceding three weeks. He had talked to the engineers rather than the managers. He had seen the equipment and the yields and the specific problems that were not in the progress reports because they had not yet been formally categorised as problems.
"Programme Zero," he said. "The semiconductor programme. We set the target in February 1975 as achieving 3-micron classified fabrication parity in 36 months and full-scale production in 48 months. We are at 20 months."
He paused, and for the first time in the meeting his posture carried something that looked, briefly, like satisfaction.
"We have beaten our own timeline substantially. We are currently fabricating production-qualified circuits at 2.8-micron with yields of fifty-eight percent — within striking distance of our sixty-five percent production target. We project full 3.0-micron production-scale qualification, with margin, by the second quarter of 1977. That is fifteen months ahead of the original 48-month target."
Rumsfeld looked up from his notes. "Fifteen months ahead."
"Fifteen months ahead," Currie confirmed. "The breakthrough was the contamination work I briefed eight months ago — the electrostatic precipitation filtration combined with positive-pressure clean room protocols. Once that was solved, the lithography and ion implantation processes, which had already been validated, scaled into production faster than we modelled. American industrial capacity, once the engineering problem is solved, scales extremely fast. That is, candidly, one of our genuine structural advantages over a smaller national industrial base."
"What does 2.8-micron in production buy us operationally?" Rumsfeld said.
"It means Programme One's phased array radar, which has been waiting on chip throughput, can begin full engineering integration in the first quarter of 1977 instead of late 1978," Currie said. "It means the AMRAAM seeker electronics can move to their final production architecture a year ahead of schedule. It means the F-15 Advanced Eagle's digital avionics integration, which Programme Three has been treating as a dependent variable, has a firm date rather than a contingent one." He paused. "Programme Zero was the bottleneck. Programme Zero is no longer the bottleneck at the rate we projected. It remains a constraint, but a loosening one."
"Where does that put us against India," Rumsfeld said.
This was the question Currie had been bracing for, and he answered it with the specific discipline of a man who had decided, three weeks earlier during his facility visits, that he was not going to let the briefing's good news soften the honesty of the assessment that followed it.
"The February 1975 intelligence placed ISMC at production-scale 3-micron with the volume capacity to supply their aerospace and computing programmes," he said. "Our community's working assumption since then has been that they have continued advancing at a steady pace and are now operating somewhere in the 2.5-micron range — eighteen months ahead of where we project ourselves reaching 3-micron." He paused. "I want to flag, on the record, that this working assumption may itself be conservative, and I want to explain why before I give you the number."
He had the room's full attention now.
"The Siddharth-2 personal computer's published specifications, when we ran them against every known model of 2.5-micron-class performance, do not fit," he said. "We have had two independent teams at Lincoln Laboratory and at NSA's technical division attempt to reverse-engineer the implied transistor density from the published clock speed, memory bandwidth, and thermal dissipation figures. Both teams, independently, arrived at density figures that are more consistent with a process node in the 1.0 to 1.2-micron range than with 2.5-micron." He paused. "I want to be precise about what that means. It does not mean we are certain ISMC is at 1-micron. It means that our 2.5-micron assumption, when tested rigorously against the only hard public data we have, does not survive the test. The honest range we can defend is somewhere between 1.2-micron and 2.0-micron, with our best single estimate closer to the lower end of that range than the community has previously been willing to state."
The room had gone very quiet.
"If that is accurate," Rumsfeld said slowly, "they have not just maintained an eighteen-month lead. They have extended it."
"If that is accurate," Currie said, "the lead in leading-edge process technology may be closer to four years than eighteen months, even accounting for our own acceleration. I want to caveat this heavily — we do not have confirmed intelligence of a 1-micron Indian fabrication line. We have a model mismatch that two independent technical teams cannot resolve at 2.5-micron, and which resolves cleanly only if the process node is significantly more advanced than our working assumption. That is suggestive, not confirmed." He paused. "But I would be doing this room a disservice if I presented the comfortable number instead of the number the data actually supports."
Rumsfeld was quiet for a long moment.
"So we have beaten our own clock by fifteen months," he said, "and the target has moved further away in the same period."
"That is an accurate, if uncomfortable, summary," Currie said. "I want to add one element of genuine good news that should not be lost in this. Our acceleration has been real and it compounds. The contamination control breakthrough, the lithography yield work, the ion implantation precision — these are not one-time gains. They establish an institutional capability and a rate of learning inside the American semiconductor-defence complex that did not exist in February 1975. We are not just closer to a fixed target. We are now capable of closing distance faster than we were eighteen months ago. The rate of our own improvement has itself improved." He paused. "Whether that rate of improvement in our own capability exceeds whatever rate of improvement is occurring inside ISMC is the question this room cannot yet answer with confidence."
The men around the table in October 1976 were not exactly the same as the men who had been in the room in February 1975.
Ruben Robertson of Hughes Aircraft was still present. He was fifty-three now and had the specific quality of a man who had spent twenty months in a state of productive alarm that had, in the last several months, begun to resolve into something closer to grim confidence — the quality of an excellent engineer who has been given a genuinely hard problem, has been given the resources to address it, and has begun, finally, to see daylight.
Robert Everett of Westinghouse was present, and had aged in a way that was not merely chronological — though in his case the aging had recently been accompanied by the specific energy of a man whose team had just delivered something he was proud of.
D.B. Kipps of Raytheon was present, with the expression of someone who had made very significant progress on a problem and had every reason to be confident about it.
Sanford McDonnell of McDonnell Douglas was present, and had brought with him a young engineer named John Capuano, in his early thirties, sitting at McDonnell's left hand with the alert stillness of someone who understood that he was in this room to provide specific technical detail when McDonnell required it and to be invisible the rest of the time.
Ben Rich of Lockheed's Skunk Works was present, carrying the specific quality of a man who has been doing the most important work of his career for twenty months and knows it and cannot say so in any public context.
Bill Gunston of Pratt and Whitney was present. Brian Rowe of General Electric was present.
In place of Lieutenant Commander Christopher Vance — recovering from injuries sustained in a carrier-approach hydraulic failure the previous month — was Commander Patricia Nguyen, who had spent eighteen months at the Naval Air Systems Command studying the original encounter and had written the classified assessment of its implications that had been more widely read inside the Pentagon than any unclassified publication in the preceding two years.
She sat at the end near the door, legal pad and pencil in front of her, with the expression of someone paying very close attention to the gap between what people were saying and what they meant.
"Programme One," Rumsfeld said. "Radar. Robertson."
Robertson stood, and for the first time in twenty months his posture was that of a man delivering good news rather than managing expectations.
"The AWG-9 signal processor upgrade is in the fleet," he said. "Eighteen-month commitment. We delivered at twenty months — two months over our internal target, but I want to put that in context. We discovered during integration that our original upgrade design, while adequate against the February 1975 threat picture, would not have been adequate against a threat that had continued to evolve. We redesigned the signal processor architecture mid-programme rather than ship something we assessed as already obsolescent."
"Tell me what we have now," Rumsfeld said.
"A signal processor running at three times the original AWG-9's clock rate, built around Programme Zero's first production-run chips," Robertson said. "Adaptive frequency-hop cycle time down from approximately 200 milliseconds to 40 milliseconds — and our latest bench testing, completed three weeks ago using the next chip revision Currie's team delivered early, shows we can push that to 25 milliseconds in the production configuration shipping in 1977." He paused. "Against the jamming profile Commander Vance and Lieutenant Murphy encountered and documented, the upgraded AWG-9 does not merely perform better. It defeats that specific jamming approach outright. We have run the recorded jamming signature against the new processor in simulation eleven times. It breaks lock in under four seconds, every time."
"Against the system as it existed in February 1975," Rumsfeld said.
"Against the system as it existed in February 1975," Robertson confirmed. "Twenty months have passed and we have no confirmed intelligence on whether the Indian system has been upgraded. Given everything we are hearing about their chip trajectory, I would assume they have. But I want this room to understand: we are no longer simply chasing a snapshot from 1975. We have built a genuinely capable system that defeats the specific threat we know about, with margin to spare against moderate improvements to that threat."
"The phased array radar," Rumsfeld said. "Long-term programme."
"Original estimate seven years from programme start," Robertson said. "We are revising that downward. With Programme Zero's acceleration, full engineering integration of the transmit-receive modules begins in Q1 1977 instead of late 1978 — Currie's fifteen-month gain transfers almost directly onto our schedule, because the phased array's performance ceiling was always set by chip throughput, not by our architecture work, which has been mature for over a year." He paused. "Revised estimate: operational phased array radar in fleet testing by 1981, a year ahead of the original seven-year mark."
"You sound confident," Rumsfeld said.
"I am confident in the engineering," Robertson said. "I am less confident that 1981 is fast enough, given what Dr. Currie just told this room about where their process technology may actually be."
Kipps of Raytheon stood for Programme Two, and unlike twenty months earlier, he did not need to manage the room's expectations downward before he began.
"The missile designation is AIM-120," Kipps said. "AMRAAM — Advanced Medium-Range Air-to-Air Missile. I want to start with where we actually are today, because it has moved substantially since our last formal report to this committee three months ago."
"Go," Rumsfeld said.
"The production-design seeker — three generations ahead of the original test article — completed its first captive-carry flight trials in August," Kipps said. "Free-flight testing began last month. We have flown six engagements against manoeuvring drone targets. Five successful intercepts. The sixth was a controlled abort due to a telemetry fault unrelated to the seeker or guidance package." He paused. "Range on the production design, measured in these trials: 140 kilometres against a manoeuvring target, with a non-manoeuvring engagement envelope extending to 165 kilometres."
Rumsfeld looked up. "You committed to 150 kilometres interim in four years. You're at twenty months."
"We are fourteen kilometres under our four-year interim target at less than half the programme timeline," Kipps said. "The seeker sensitivity problem that was limiting us to the 75-kilometre test article range has been solved — a new gallium arsenide front-end that Programme Zero's compound semiconductor work made possible six months ahead of when we expected to have it. Fire-and-forget capability is confirmed across the full test envelope. The launching aircraft can break lock and manoeuvre immediately after launch with no degradation to terminal accuracy."
"Strategic target," Rumsfeld said. "200 kilometres, six years."
"We will beat that," Kipps said, with the directness of an engineer who has done the arithmetic and trusts it. "Production design first flight is scheduled for early 1977 — we have effectively already flown it in the trials I just described. IOC is now projected for 1978, a full year ahead of the original 1979 fleet-service target. The 200-kilometre strategic variant, using the next-generation seeker architecture currently in detailed design, is on track for 1980, a year ahead of the six-year mark."
Currie added, from the side of the room: "I want to flag something for context, given what I said earlier about the chip trajectory. The Mauritius engagement geometry placed the Indian missile's effective range at a minimum of 210 kilometres, with our analysis suggesting it could be substantially more. Our 1980 target of 200 kilometres closes most of that gap as it stood in February 1975. It does not close a gap that has continued to widen if their missile programme has advanced at anything like the rate we now believe their semiconductor programme has advanced."
Rumsfeld absorbed this. "So the missile programme itself is ahead of schedule and performing exceptionally well."
"Yes," Kipps said.
"And it may still not be enough."
"That is the honest caveat," Kipps said. "I am not going to apologize for AMRAAM. It is, today, the best air-to-air missile programme in the world outside of whatever India is building. I want this room to hold both things at once: this is genuinely excellent, and it may not be sufficient."
The F-15 discussion was Sanford McDonnell's, with periodic clarification from Capuano.
"The F-15A has been in fleet service since January at Langley," McDonnell said. "Twenty-one months in, the operational data confirms everything the test programme promised. Thrust-to-weight at combat weight exceeds 1.2-to-one. The APG-63 radar gives us look-down/shoot-down capability the F-14's AWG-9 never had. We have flown the F-15 against every NATO and Warsaw Pact aircraft we have access to in controlled exercises, and it wins decisively in every visual-range engagement."
"Against the Indian profile," Rumsfeld said.
"That is the harder question, and I am going to give you the honest version," McDonnell said. "We do not have confirmed specifications on the Indian aircraft. What we have, from the Mauritius encounter, is a profile: supercruise without afterburner, a radar lock at 113 nautical miles, a fire-and-forget missile, and sustained 7-G manoeuvring while supersonic." He paused. "In a beyond-visual-range engagement initiated by an aircraft with that profile, even our upgraded F-15 — with the Programme One radar improvement Robertson just described — is at a disadvantage, because their detection and engagement range still appears to exceed ours."
"The Advanced Eagle programme," Rumsfeld said.
"Ahead of schedule, thanks to Programme Zero," McDonnell said. "Engine thrust upgrade — the F100 derivative with fifteen percent additional thrust — completes flight testing in early 1977, six months ahead of our original mark. Structural modifications for extended fuel capacity are in fabrication now. The digital avionics integration, which was our dependent variable, now has a firm 1978 date instead of a contingent post-1981 date, because Currie's team delivered the production chips fifteen months early." He paused. "When the full Advanced Eagle configuration enters service in 1979, we will field an aircraft that is, in every respect we can presently characterise, competitive with the Indian aircraft as it existed in February 1975. Whether it remains competitive with whatever the Indian aircraft has become by 1979 is the open question this entire room keeps returning to."
Capuano had been writing on his notepad. He looked up.
"Mr. McDonnell," he said quietly.
McDonnell looked at him. "Go ahead. This is the work I mentioned."
"The supercruise question," Capuano said. "I've spent four months on the propulsion-airframe integration question, and I want to present something the team has actually made real progress on, not just a diagnosis."
"Go ahead," Rumsfeld said.
"The standard explanation for why our aircraft don't supercruise efficiently is the transonic wave drag peak," Capuano said. "Sustaining supersonic flight through that peak without afterburner requires either very high thrust or very low wave drag. We've always solved it with thrust. The Indian aircraft, based on everything we can reconstruct, solves it with drag — an area-ruling approach that integrates the inlet geometry, the fuselage cross-section, the wing-fuselage junction, and the tail configuration as a single optimised system rather than as separately area-ruled components." He paused. "For four months I believed this was achievable only through computational fluid dynamics modelling at a resolution we could not match. I want to revise that assessment."
The room's attention sharpened.
"Three weeks ago, working with the new computing allocation Programme Zero's accelerated chip output gave us access to, my team ran a full-aircraft CFD model at four times the resolution we were using in February. We found an inlet-fuselage integration geometry that reduces our own F-15 derivative's transonic wave drag by approximately eleven percent without any structural weight penalty." He paused. "It is not the Indian solution. We don't know their exact solution. But it demonstrates that the computational approach is now within our reach, not five years away from our reach. The chip gain Currie described this morning didn't just speed up the radar and the missile. It gave my team the computing power to start closing the aerodynamic design gap directly."
Gunston from Pratt and Whitney leaned forward. "Eleven percent wave drag reduction. What does that do to supercruise endurance on the current engine?"
"On the F100 derivative entering testing in 1977, it extends sustained supersonic cruise range by approximately eighteen percent at the same fuel fraction," Capuano said. "It does not give us limitless supercruise. It gives us a meaningfully more capable aircraft using an engine we already have, and a design methodology we can keep applying."
Currie, listening from the side of the room, said: "This is exactly the kind of compounding effect I want this committee to register. Programme Zero is not only a radar and missile enabler. It is now directly enabling aerodynamic design gains that we did not anticipate when we set the programme's original objectives. The chip advantage cuts across every programme in this room simultaneously."
McDonnell said: "Which tells us something important about how the other side may be operating, if their aerospace design teams have access to computing of the kind their chip programme appears capable of producing. It's not only that their components are individually better. It is that their entire design process may be operating with a tool advantage we have only just begun to acquire ourselves."
Rumsfeld had been listening to this exchange closely. "What do we do with that?"
"We keep doing exactly what we just did," Capuano said. "Every increment of chip capability Programme Zero delivers, we push immediately into design tools, not just into finished hardware. That is new as of this year. It wasn't happening systematically before."
Ben Rich's turn came after lunch.
He had spent the morning listening more than was his habit — Rich was a man who talked, who explained, who drew diagrams on whatever surface was available. But the morning's exchanges had given him something to think with, and when he stood, he did not preamble.
"Programme Four," he said. "The low-observable demonstrator. We are at 20 months. The demonstrator has flown, and I want to open with the headline: it works better than our design predictions said it would."
The room went very still, in a different register than it had for Currie's caveats — this was the stillness of genuinely good news.
"First flight was in June," Rich said. "We have accumulated 31 hours of flight time across 24 sorties as of last week. Ground-based radar tracking measurements have confirmed the design predictions to within eight percent — tighter than our own conservative engineering margin assumed. At the worst-case aspect angle — direct frontal, optimum frequency for this target class — we are measuring a radar cross-section reduction of approximately 250-times relative to a conventional fighter, not the 100-times we projected in our original design baseline."
"That's better than what you told us a year ago," Rumsfeld said.
"It is," Rich said. "The improvement came from two places. First, the radar-absorbing material chemistry problem I flagged in our last full briefing — the delamination issue — is essentially solved. Our materials team found a polymer bonding agent that eliminates the thermal-cycling failure mode entirely. We have had zero delamination incidents in the last fourteen sorties. Second, and this is the part I want to spend time on: we applied Mr. Capuano's inlet integration approach, which McDonnell's team shared with us informally two months ago, to our own engine inlet geometry. The inlet was always our single largest radar reflector. The redesign Capuano's CFD work made possible cut our inlet's contribution to total RCS by more than half, with no thrust penalty, because the geometry we found was simultaneously better for radar return and better for mass flow efficiency."
"So the programmes are now reinforcing each other," Rumsfeld said.
"For the first time in twenty months, yes," Rich said. "That is the single most encouraging development in this entire briefing, in my view. We are no longer six isolated programmes converging by coincidence on a common chip dependency. We have started actively sharing design methodology across programme boundaries, and it is compounding."
"The third problem you flagged a year ago," Rumsfeld said. "India."
Rich was quiet for a moment — the pause of a man who has thought carefully about how to frame something difficult without undercutting genuinely good news.
"That problem has not gone away, and I want to be precise about it rather than let the good news this morning soften it," he said. "Our 250-times RCS reduction is a real, measured, flight-tested achievement. The question remains whether it is sufficient against a radar system we cannot fully characterise — and after what Dr. Currie told this room this morning about the likely process node, I am more concerned than I was in our last formal review, not less." He paused. "A radar built on chips at 1.2-micron or below has detection sensitivity that may be improving faster than our RCS reduction is improving. I do not know the crossover point. Nobody in this room knows the crossover point."
"What would it take to know?" Rumsfeld said.
"Either much better intelligence on their radar architecture, or a margin large enough that the question stops mattering," Rich said. "I previously told this committee that a 1,000-times reduction was achievable but came with a meaningful performance penalty — roughly twelve percent thrust loss from the inlet redesign required. I want to revise that, in light of the inlet work Capuano's team and my team have now done jointly. We believe we can reach 600-times to 700-times RCS reduction with the penalty closer to four to five percent thrust loss, by combining the cross-programme inlet geometry with a second-generation absorbing material currently in chemistry validation." He paused. "That is a genuinely better trade than I could offer this committee a year ago. It is still not the 1,000-times margin that would let me stop worrying about their radar trajectory entirely."
"Recommendation," Rumsfeld said.
"Pursue the 600-to-700-times design as the production baseline," Rich said. "Keep a dedicated research thread on the absorbing material chemistry pushing toward 1,000-times for the following aircraft generation. And — I want to say this plainly — keep pushing the cross-programme sharing that produced this quarter's gains. The single biggest improvement in this programme in twenty months came from talking to McDonnell's propulsion team, not from anything my own people did in isolation."
Commander Nguyen had filled fourteen pages of her legal pad by the time Rumsfeld turned to her at the end of the programme reviews. She set her pen down and looked at the room with the composure of someone who has reached a genuinely more complicated set of conclusions than the morning's good news alone would suggest.
"Commander," Rumsfeld said. "You wrote the classified assessment a year ago. Tell me what's changed in the integrated picture."
"What's changed is substantial, and I want to start with what should not be understated," she said. "Twenty months ago I would not have predicted this level of programme convergence. Programme Zero beat its own production timeline by fifteen months. AMRAAM is fourteen kilometres under its four-year interim target at less than half the timeline. The radar upgrade defeats the documented threat outright. The stealth demonstrator is outperforming its own design baseline. And as of this morning, for the first time, the programmes are actively cross-pollinating — McDonnell's CFD work improving Lockheed's inlet, Programme Zero's chip output feeding directly into aerodynamic design tools rather than only into finished avionics." She paused. "This is, without qualification, the most effective twenty months of concentrated defence development I am aware of in the post-war period. I want that on the record before I say anything else."
"Go ahead and say the anything else," Rumsfeld said.
"The integrated picture also shows that the target has moved," she said. "Dr. Currie's revised process-node estimate this morning — if it holds, and I think this room should treat it as the working assumption rather than the cautious outlier — means the semiconductor gap we believed was eighteen months in February may currently be closer to three to four years, even after our own fifteen-month acceleration. The missile gap, even at AMRAAM's excellent 140-kilometre demonstrated range, may still be sixty to seventy kilometres short of where their system already was in 1975, before accounting for any improvement on their side since." She paused. "And Mr. Capuano's inlet integration work, while a genuine and important American achievement, replicates roughly half of what we infer the Indian design already achieved nearly two years ago — and it took us four months of dedicated work with newly available computing power to get there."
She looked at Currie.
"None of this contradicts the good news," she said. "It contextualises it. We are running faster than we have ever run. The track may also be longer than we thought."
The room held this for a long moment.
"What do you recommend," Rumsfeld said.
"Continue and accelerate the cross-programme integration that started organically this quarter," she said. "Formalise it. Don't wait for McDonnell's team to informally share a CFD result with Lockheed over lunch — build the structure that makes that sharing automatic and continuous. The single largest gain this committee heard about today came from two defence contractors talking to each other without being told to. Institutionalise that, and I believe the rate of our own improvement increases again." She paused. "And I want a standing instruction, not a one-time data pull, to revisit the semiconductor process-node estimate every quarter rather than annually. If the number moved this much between our last full review and today, we cannot afford to operate on a stale assumption for twelve months at a time."
Rumsfeld nodded slowly. "Noted. Both recommendations stand as directives, effective today."
The afternoon session addressed the intelligence directive.
The collection had produced extensive overhead imagery of the Gorakhpur industrial complex, which had grown significantly since the 1975 baseline — new buildings, expanded facilities, construction activity consistent with substantial additional manufacturing capacity. The imagery did not show what was inside the buildings. This remained the fundamental limit of overhead collection.
The human intelligence collection had produced three contacts and one reliable asset in twenty months — a supplier-side employee with delivery access to the facility. His reports confirmed continued expansion, personnel growth from roughly 800 in 1975 to approximately 1,400, and a significant increase in security protocols over the preceding six months: vehicle searches at the perimeter, mandatory surrender of personal effects for all visitors.
"The security increase began roughly eight months ago," the intelligence officer briefing this section said. "We assess it as a response to awareness of our collection activity."
"Do they know we have an asset inside?" Rumsfeld asked.
"We assess no," the officer said. "His access is through a commercial supply relationship, a category that would be difficult to identify without dedicated counterintelligence effort against supplier networks. The new protocols focused on personnel access and document control, not supply chain relationships. His access is also limited — loading docks and receiving areas, not production floors."
"What would it take to get onto the production floors?" Rumsfeld said.
"Direct recruitment of engineering personnel," the officer said. "We have made five attempts in twenty months. Three did not progress past initial approach. One individual agreed to a second meeting and then ceased contact. One provided two reports and then went silent and subsequently became unreachable. We cannot determine whether that was voluntary withdrawal or a counterintelligence response on their side."
"What is the Indian intelligence establishment's apparent sophistication here?" Rumsfeld asked.
"Higher than our 1975 baseline assumed," the officer said. "India's external intelligence service has expanded its technical collection capability substantially in the past five years, and the facility-level security response we are observing shows pattern-recognition and counterintelligence tradecraft that is, frankly, more disciplined than we initially modelled for an industrial security apparatus of this kind."
Currie said, from the side of the room: "I'd note that this tracks with everything else we've discussed today. An organisation capable of accelerating semiconductor process development at the rate we now believe is occurring, and capable of the systems-integration discipline Commander Nguyen described across their aircraft and missile and radar programmes, is also the kind of organisation capable of building a serious internal security function. These are not separate observations. They're the same underlying institutional capability expressing itself in different domains."
Rumsfeld looked at the three-page summary of Shergill Industries' commercial and industrial portfolio that had been prepared for the briefing — the aircraft programmes, the naval construction, the electronics, the Nobel Prize for the LED work, the public-domain release of the core patents that had accelerated the global market even as the proprietary engineering and manufacturing processes remained closely held.
He set it down.
"Twenty months ago this room was told we faced a wake-up call," he said. "I want to be precise about where I think we actually stand today, because I don't think 'we are still behind' is the complete or even the most useful way to characterise it." He looked around the table. "We have, in twenty months, built a missile that beats its own four-year target at less than half the timeline. We have built a radar upgrade that defeats a documented real-world threat outright. We have a stealth aircraft outperforming its own design baseline by a factor of two and a half. We have a semiconductor programme that beat its production schedule by fifteen months. This is not the record of an organisation that is failing to compete. This is the record of the most capable defence-industrial mobilisation this country has produced since the Manhattan Project, and I want that understood plainly before anyone in this room leaves and characterises today's briefing as discouraging."
He paused.
"What I also heard today is that the adversary's own trajectory may be steeper than our most recent working assumption, specifically on process technology, and that closing an eighteen-month gap that may actually be a three-to-four-year gap requires us to keep accelerating our own rate of acceleration, not simply hold our current pace." He looked at Currie. "Both things are true simultaneously. We are winning the execution race against our own targets. We do not yet know if we are winning the race against theirs, because we do not have a complete picture of theirs."
He looked at the intelligence officer.
"I want a standing quarterly review of the process-node estimate, as Commander Nguyen recommended. I want the cross-programme integration that produced this quarter's best results formalised into a standing structure, not left to chance lunch conversations between contractor teams." He paused. "And I want an honest fifteen-year trajectory assessment — not of where the Indian programme is today, but of where it is going, set against where our own accelerating programmes are going. Low confidence in an honest projection is more useful to me than high confidence in a comfortable one."
He stood and put on his coat.
"Same time next year," he said. "I want to know if we closed more distance than they did."
He walked out.
In the room, the people remaining looked at the table and at each other and at the mahogany surface and at the closed door. The security detail began collecting the classified materials.
Commander Nguyen was the last to pick up her legal pad. She had filled eighteen pages. She looked at the final line she had written during Rumsfeld's closing remarks.
It read: We are running faster than we ever have. We still do not know if the track is getting shorter.
She thought about the morning's genuine successes — the missile that had beaten its own timeline, the radar that defeated a real threat outright, the stealth aircraft outperforming its design baseline, the chip programme that had clawed back fifteen months through sheer engineering discipline and American industrial scale. These were not small things. She had spent eighteen months studying an adversary whose capabilities had seemed, in February 1975, almost unanswerable, and she had just watched six American programmes answer it, one component at a time, with real hardware and real flight hours and real intercepts.
She also thought about Currie's revised process-node estimate, and what it implied if it held.
She thought: the American system, once it commits its full industrial and engineering weight to a problem, moves with a speed that nothing else in the world can fully match. That had been true in 1942 and it appeared to still be true in 1976.
She thought: the question was never whether the United States could build excellent things quickly. It always could. The open question was whether it was racing against a fixed point or a moving one — and after today, she believed it was racing against a moving one, and that the only sustainable strategy was to keep moving faster than the thing they were chasing, indefinitely, without ever assuming the gap was closed.
She closed the legal pad.
She walked out.
The room had no windows.
Outside the Pentagon, the October afternoon was clear, and the Potomac was visible from the parking lot, and Washington looked the same as it always looked — the monuments, the marble, the specific weight of a city that understood itself as the centre of something enormous.
She stood for a moment in the October air.
She thought about silicon, and about missiles that already flew further than anyone had told her was possible eighteen months earlier, and about an aerospace design problem that her own country's engineers had cracked, independently, in four months once they had the computing power to try.
Then she got in her car and drove back to the Naval Air Systems Command, where she had work to do.
There was always work to do.
End of Chapter 245
Programme Status Summary — October 1976
Programme Zero (Semiconductor): Production-qualified 2.8-micron, 58% yield. Full 3.0-micron production scale projected Q2 1977 — fifteen months ahead of original 48-month target. Adversary process-node working estimate revised from 2.5-micron to a defensible range of 1.2–2.0 micron based on independent reverse-engineering of commercial product specifications.
Programme One (Radar): Upgraded AWG-9 in fleet service, defeats documented February 1975 jamming threat outright in simulation (11/11). Phased array radar operational testing revised to 1981, one year ahead of original seven-year mark.
Programme Two (Missile/AMRAAM): Production-design seeker flight-tested, 140km demonstrated range against manoeuvring targets (5/6 successful intercepts). IOC revised to 1978. 200km strategic variant on track for 1980, one year ahead of six-year mark.
Programme Three (F-15 Advanced Eagle): In fleet service since January 1976. Engine thrust upgrade testing early 1977. Digital avionics integration date firmed to 1978 (from contingent post-1981). Full Advanced Eagle configuration: 1979.
Programme Four (Low-Observable Demonstrator): 31 flight hours, 24 sorties. RCS reduction measured at 250x baseline (vs. 100x design target). Delamination failure mode resolved. Cross-programme inlet redesign (with Programme Three's CFD work) projects 600–700x reduction achievable at 4–5% thrust penalty for production baseline.
Cross-Programme Integration: First documented instance of organic cross-contractor design sharing (McDonnell Douglas CFD methodology → Lockheed inlet redesign). Directive issued to formalise as standing structure.
Intelligence: Quarterly (from annual) process-node reassessment directive issued. Fifteen-year trajectory assessment commissioned.