Chapter 176: A Billion Lights
2 January 1975 — Gorakhpur; Bombay; New York; Tokyo
The production line had run its first full shift on December 28th.
Not a test run. Not a pilot batch. A full production shift — eight hours, three reactors running simultaneously, the automated phosphor dispensing system operating at the throughput that the six-month scale-up programme had been building toward, the quality control station at the end of the line sampling every twentieth device and running it through the measurement protocol that SPEI's semiconductor division had developed over the previous four months.
The shift's output: 4,847 white LED devices.
The yield: 91.3%.
The average efficacy of the sampled devices: 84.2 lumens per watt.
The specification was 80 lumens per watt minimum for the industrial grade. 84.2 was above specification. The yield was above the 90% target. The throughput was within eight percent of the design rate.
The production line worked.
Dr. Ramesh Chandra had been in the production facility when the shift ended. He had been there since four in the morning, which was when the shift started, which was when he had arrived, which was when he had stood at the production line's input end and watched the first six-inch sapphire wafer go into the first reactor and had then spent eight hours moving between the three reactor stations and the phosphor dispensing system and the quality control station and the data acquisition computers, watching every stage of the process that SPEI had spent six months making reliable and repeatable and scalable.
When the shift ended, he had sat in the production manager's office at the end of the line and read the shift report.
He read it once.
He set it down.
He called Karan.
Karan's phone rang at twelve forty-seven in the afternoon on December 28th.
He was in the Gorakhpur design bureau with Ramanathan, reviewing the S-35's wing-body fairing data from the previous week's test flight. The S-35 programme was in an excellent phase — the fairing data was better than the design prediction, which was the specific quality of a result that confirmed the aerodynamicist's work and freed the test programme to proceed to the next envelope expansion.
He picked up.
"The first production shift is complete," Chandra said.
Karan was quiet.
"91.3% yield," Chandra said. "84.2 lumens per watt average. Above both specifications."
Karan looked at Ramanathan across the bureau table.
Ramanathan looked back with the expression of a man who understood that a phone call that caused Karan to become very still was usually a phone call about something significant.
"The line is operational," Karan said.
"Yes," Chandra said.
"Throughput?"
"4,847 devices in the first shift. Eight hours. Three reactors. The design rate is 5,200. We're within eight percent on the first full shift."
"When did the line run fully automated?" Karan said.
"The phosphor dispensing automation went live on December 22nd," Chandra said. "Six days of debugging before the first full shift. The automation is working."
"The SPEI team," Karan said.
"All seventeen people from the semiconductor division were in the building this morning," Chandra said. "Suresh ran the reactor operations team. The process integration work from the past four months—" He paused. "It worked."
Karan was quiet again.
Four months ago, on August 20th, 1974, he had given SPEI its first mission: close the gap between the laboratory white LED and the commercial production white LED. The specific mission parameters: a yield above 85% and a throughput that could meet the first production contracts by December 1974.
December 28th was inside December 1974.
The mission was complete.
"Chandra," he said.
"Yes," Chandra said.
"Tell the team," Karan said. "From me. Specifically and in person. Tell each of them."
"I will," Chandra said.
"And Chandra."
"Yes."
"You were right about the temperature uniformity," Karan said. "The SPEI solution — the susceptor geometry modification — that was the key."
"Suresh identified the specific interference pattern," Chandra said. "In week two of the SPEI programme. The three-week estimate you gave in August—"
"Was it three weeks?" Karan said.
"Three weeks and two days," Chandra said. "From the SPEI semiconductor division's first meeting to the yield improvement from 68% to 89%. And then two months to get it to 91%."
"SPEI works," Karan said.
"SPEI works," Chandra confirmed.
The production line had continued running through December.
December 28th, 29th, 30th, 31st. Four shifts. Total output: 19,388 devices. Average yield across all four shifts: 90.8%. Average efficacy: 83.9 lumens per watt.
The devices were packaged.
The packaging was the specific packaging that ISMC's electronics division had developed for the production LED — a standardised industrial format, the device in a ceramic mount with pre-attached gold wire bonds, the phosphor coat applied and cured, the dome encapsulant formed and verified, the whole assembly in a tape-and-reel format compatible with automated pick-and-place manufacturing equipment.
The packaged devices were placed in inventory.
The inventory was labelled: ISMC White LED, Industrial Grade, SPEI Production Series 001.
By January 2nd, 1975, there were 23,412 units in inventory.
The signing session on January 2nd began at ten in the morning.
It was not in a conference room.
It was in the main auditorium of the ISMC facility — the auditorium that the company used for large gatherings, which had a capacity of three hundred and which on this morning held approximately two hundred and forty people: the full ISMC research team, the full SPEI semiconductor division, the production line staff, the quality control team, the packaging team, the inventory team. Every person who had been part of the programme from the first reactor run in March 1973 to the production shift on December 28th was in the auditorium.
Suresh was in the front row. He had circles under his eyes from four days of production shifts that he had personally supervised at four in the morning. He was twenty-four years old and he looked twenty-four years old after four days of four AM shifts, which was to say: the circles under his eyes were significant but he was also sitting very straight with the quality of someone who had just done the thing they were there to do and who was still experiencing the specific alertness of that condition.
Arun Mehta was beside him. The youngest member of the original research team. He had called his mother after the December 28th shift. She had asked: is it real? He had said: it's real. She had been quiet for a long moment and then had said: put a white LED in one of the lights in your room tonight and tell me in the morning how it feels to wake up in a room lit by something you made. He had done this. He had written a note in his personal journal that morning: It feels different. I don't know how to describe how it feels different, but it does.
Dr. Chandra was at the podium.
He did not usually use podiums. He was a scientist who was uncomfortable with ceremony and who had learned, in the twenty-two months since the programme began, that certain moments required ceremony even when ceremony was uncomfortable because the people who had been part of those moments deserved the ceremony's acknowledgement.
He looked at the auditorium.
He looked at the two hundred and forty people in it.
He said: "I want to say something before the business of today begins."
The auditorium was quiet.
"The Physical Review Letters paper was published on August 12th," he said. "The MIT confirmation was September 3rd. The Caltech confirmation was September 5th. The NPL Teddington certification was September 8th." He paused. "Those are the dates when the world confirmed that what we had built in the laboratory worked."
He paused.
"December 28th, 1974 — is the date when what we built in the laboratory became what can be built in a factory. That date," he said, "is the date that matters. The laboratory result was science. The production line result is what the science was always for."
He looked at Suresh in the front row.
"Suresh ran the first production shift," he said. "He has been awake since four AM on December 28th with intermissions. He is going home after this meeting and sleeping for thirty-six hours, which is a medical instruction." He paused. "Before he goes home, I want everyone in this room to understand that the temperature uniformity problem — the specific problem that reduced our six-inch wafer yield from 91% to 68% when we first scaled from three-inch — was identified and solved by Suresh in week two of the SPEI programme. The solution was a susceptor geometry modification. The modification took three weeks to design, fabricate, and test. It brought the yield from 68% to 89%. The subsequent two months of process optimisation brought it to 91%."
Suresh was looking at the floor. Not in embarrassment — in the specific condition of someone receiving acknowledgement that they had worked very hard to deserve.
"In this programme," Chandra said, "there are 217 people in this building who contributed something that I could describe specifically. I am not going to describe all 217 contributions. But I want everyone in this room to understand that what happened here — from the first wafer run in March 1973 to the production shift on December 28th — was not done by a few people at the top. It was done by every person who ran a reactor and recorded a data point and revised a process parameter and stayed until two in the morning because the measurement wasn't making sense and they needed to understand why."
He looked at the auditorium.
"India has never produced this," he said. "A semiconductor device that doesn't exist anywhere else in the world, produced in a factory, at scale, with a yield that would be competitive with any production facility in any country. Not a prototype. Not a pilot batch. A production line." He paused. "You built this."
He stepped away from the podium.
Karan came to the podium.
He came to it the way he came to most things that required him to speak to a large group: without performance of the occasion, with the specific quality of someone who had something specific to say and was going to say it.
He looked at the auditorium.
"The licensing agreements," he said.
He held up three folders.
"These are the signed licensing agreements for the ISMC white LED patents. I'm going to describe them, because the people in this room produced the technology that these agreements are about, and they deserve to know what is happening with it."
He set the folders on the podium.
"The first agreement," he said, "is with General Electric Company, USA, and Philips N.V., Netherlands, acting as a consortium." He picked up the first folder. "The American and European markets. The license covers the hydrogen passivation process patent, the multi-quantum-well architecture patent, and the white LED phosphor conversion patent. The term is fifteen years." He paused. "The terms: GE and Philips together pay seventy-five million US dollars upfront as a license fee, and five percent royalty on every unit sold under the license."
The auditorium was quiet in the way of rooms that are receiving numbers.
"The geographic scope," he said, "is the United States and Europe. GE handles the American market. Philips handles Europe. The exclusivity is territorial: GE cannot sell under the license into Philips' territory and vice versa." He paused. "Shergill Electronics retains the right to sell into both territories independently. We do not need the license to sell in their territories. But we chose not to compete directly for fifteen years in exchange for the upfront fee and the royalty structure."
He picked up the second folder.
"The second agreement," he said, "is with Sony Corporation and Toshiba Corporation, Japan, acting as a consortium." He looked at the auditorium. "Same patents. Fifteen years. The same seventy-five million US dollar upfront fee and five percent royalty." He paused. "The geographic scope: Japan, Hong Kong, and South Korea." He paused. "With two specific conditions that differ from the GE-Philips agreement. First: Shergill Electronics has the right to sell into Japan, Hong Kong, and South Korea on a non-exclusive basis — Sony and Toshiba have the license but we can compete in their territory. Second: the Sony-Toshiba consortium gets access to our production process documentation at a level that allows them to build their own production facilities. This was the specific concession they required that GE and Philips did not."
He looked at the auditorium.
"The reason Sony and Toshiba required the production process access," he said, "is that they understood something that GE and Philips also understood but did not say as directly: the technology is so far ahead of what anyone else has that the only way to build a commercially competitive position is to know not just the patent claims but the specific production knowledge that makes the patent work at scale. They were willing to pay the same fee for access to both." He paused. "We gave them the production process documentation because the documentation without the specific people who understand it — without the 217 people in this room — is useful but not complete. They will need two to three years to build production capacity using our documentation. By which time we will be two to three years further ahead."
He picked up the third folder.
"The third agreement," he said, "covers everything else. All of Asia excluding Japan, Hong Kong, and South Korea. The Middle East. Africa. South America. This territory is held exclusively by Shergill Electronics. No license. No royalty. We produce and sell into this territory ourselves."
He looked at the auditorium.
"India," he said. "The Indian market is ours. The Asian market is ours. The Middle Eastern market is ours." He paused. "The technology was developed here. The first commercial production happened here. The markets closest to here — the markets that are growing the fastest, the markets that have the greatest need for efficient lighting — those are our markets." He paused. "The Western companies can have the Western markets. We keep Asia."
He stepped away from the podium slightly.
"The mathematics," he said. "Seventy-five million dollars from GE and Philips. Seventy-five million dollars from Sony and Toshiba. One hundred and fifty million dollars in upfront licensing fees." He paused. "At five percent royalty on sales. In fifteen years, when the LED lighting market reaches its first mature commercial phase, the royalty income will exceed the upfront fee by a factor of ten or more." He paused. "We do not need the royalty income. We are building a manufacturing business. The royalty income is the return on the intellectual property. The manufacturing business is the return on the production capability." He paused. "These are different revenue streams from the same underlying investment. The same 22 months of work produced both."
He looked at the auditorium.
"Aditya has the financial model," he said. "He has been projecting the royalty income for six months. I am not going to quote his projections because his projections are specific and I disagree with three of his assumptions and we have been arguing about them since October." He paused. "What I will say is that both the number I think is correct and the number he thinks is correct are large enough that the specific difference between us is interesting rather than consequential."
A sound from the auditorium — not quite a laugh, the specific sound of a large group of people who have just been given permission to find something slightly funny and who are taking it.
Aditya, who was sitting in the second row, made a note in his notebook. It was not visible what the note said but his expression suggested it was the expression of a man who was noting the disagreement for the record.
"The production line," Karan said. He came back to this. "The 23,412 devices in inventory. The agreements specify the first commercial delivery: GE receives 50,000 devices in February. Philips receives 50,000 in March. Sony and Toshiba split 100,000 devices across Q1 and Q2 of 1975." He paused. "Every device that leaves this building carries the ISMC brand mark. Not the license holder's brand. Ours. The license holders sell under their own brands in their markets. But the devices they sell in 1975 and 1976 and 1977 — until they build their own production lines — will be manufactured here, in Gorakhpur, by the people in this room."
He looked at the auditorium one last time.
"The world is going to light itself with LEDs," he said. "Not in five years. In thirty years. The transition from incandescent and fluorescent to LED will happen over thirty years, progressively, as production costs fall and product quality improves and the economics become undeniable in every application. When that transition is complete — when the last incandescent light bulb is replaced — every light that replaced it will carry the genetic marker of what was done in this building." He paused. "The hydrogen passivation solution. The multi-quantum-well architecture. The phosphor conversion formula. These are the foundation. Every LED that exists will trace back to what was figured out here."
He was quiet for a moment.
"That is what you built," he said.
He stepped down from the podium.
The signing happened at eleven.
The actual signings had occurred over the previous two weeks — GE and Philips on December 15th in New York, Sony and Toshiba on December 19th in Tokyo. The agreements were signed. The upfront payments had been received. The 150 million US dollars was in ISMC's account at the State Bank of India's international division.
What happened at eleven on January 2nd was the ceremonial signing — the copies for the record, the photographs, the specific public acknowledgement of what had been agreed.
Vikram Sharma, the patent lawyer from New York, was present. He had flown from New York to Bombay to Gorakhpur specifically for this. He was a man who had been doing patent law for fifteen years and who had seen many significant technology licensing agreements, and who had described this agreement to his partner in New York as: the largest single technology licensing transaction in the history of the semiconductor optoelectronics field, and probably in the history of any licensing deal originating from an Indian company.
He stood at the side of the auditorium's stage with the composed expression of a man who was at peace with the work he had done.
Aditya stood beside Karan at the table where the copies were laid out.
The copies were elegant documents — not the functional commercial contracts, which were locked in files at various law firms, but the ceremonial copies, printed on the specific paper that significant agreements were printed on, the paper that communicated: this document will be framed and remembered.
Karan signed.
Aditya countersigned.
Chandra signed as the programme director.
Vikram Sharma notarised.
The auditorium watched.
At twelve-thirty, lunch.
The ISMC canteen had been operating since six in the morning preparing for the lunch, which was the canteen's contribution to the ceremony — a full meal for two hundred and forty people, with the specific quality of food that the canteen produced when it was given time and resources, which was the quality of food that ISMC's canteen always produced because the canteen's head cook had strong opinions about food and no one had ever successfully argued him down.
The lunch produced the specific atmosphere that meals after significant events produced: the release of tension that had been held through the formal morning, the shift into the register of ordinary human interactions that followed extraordinary human events.
Suresh ate the largest plate of food that anyone at his table had seen him eat in the four months of the SPEI programme.
Arun sat beside him and said: "When did you last eat properly?"
"December 27th," Suresh said.
"When are you going home?" Arun said.
"After lunch," Suresh said. "Chandra issued a medical instruction."
"He's right," Arun said.
"I know," Suresh said. "I'm going. After this daal."
He ate the daal.
The daal from the ISMC canteen was — Suresh had been eating this daal for four months and had formed a settled opinion of it. The opinion was that it was better than his mother's daal, which he had not told his mother and would not tell his mother.
"The Sony visit," Arun said.
"What about it?" Suresh said.
"When they come to see the production line," Arun said. "The production process documentation visit. They're sending a team of twenty engineers."
"I know," Suresh said.
"You'll have to explain the susceptor modification," Arun said.
"I'll explain what there is to explain," Suresh said. "The modification is not complicated. The insight was in identifying that temperature uniformity was the problem. The modification itself is a standard susceptor geometry change."
"The insight," Arun said. "How did you identify it?"
Suresh thought about this.
"The yield data," he said. "The pattern of the yield failures. The failures weren't random. They were spatially distributed on the wafer — the peripheral devices were consistently worse than the centre devices. That's a temperature gradient signature. Once you know it's temperature, the susceptor geometry is the obvious variable."
"Obvious in retrospect," Arun said.
"Obvious with the data," Suresh said. "The data was the key. We had comprehensive yield mapping data because the SPEI protocol required it. Without the spatial yield maps, you might spend months chasing other variables." He paused. "The SPEI methodology — the structured problem analysis, the complete data collection before diagnosis — that's what made it findable in three weeks."
Arun was quiet for a moment.
"SPEI works," he said.
"SPEI works," Suresh said. He finished his daal. "I'm going home."
He stood.
He looked at the auditorium — at the two hundred and forty people eating lunch on January 2nd, 1975, in the building that had produced something that the world had not had before and that was now, as of December 28th, a production line.
He picked up his bag.
He went home.
He slept for thirty-nine hours.
In New York, at the offices of General Electric's Corporate Research and Development headquarters in Schenectady, Dr. Harold Thompson had been the primary technical evaluator for the GE-Philips licensing deal. He was fifty-one years old, a solid-state physicist with twenty-two years at GE, the person who had been dispatched in September 1974 to receive the device samples from ISMC and run GE's own internal characterisation.
He had run the characterisation.
The numbers had come back at 10.1 percent EQE and 83 lumens per watt, consistent with the MIT and NPL confirmations.
He had written an internal report.
The internal report had said, in its summary section:
The ISMC device represents a legitimate breakthrough in semiconductor optoelectronics. The efficiency figures are real, confirmed by our own measurements. The production process, as described in the Physical Review Letters paper and in the patent claims, is technically coherent and reproducible by any facility with MOCVD capability and the specific process knowledge described. The key unknowns are: (1) whether ISMC can maintain these efficiency figures at production scale, and (2) when competitors can independently develop the hydrogen passivation solution from the published claims.
He had also written, in a personal note to his department director: We need this license. The specific term of fifteen years is the time window within which LED lighting transitions from niche to mainstream. If we're inside the technology through the license, we're positioned for the transition. If we're outside it, we're selling incandescent bulbs to a market that will stop wanting them.
The license had been approved.
The seventy-five million dollars — half the total fee, shared with Philips — had been paid.
On January 2nd, Thompson received a cable from ISMC.
The cable said: First production shift completed December 28th. Yield 91.3%, average efficacy 84.2 lumens per watt. 23,412 units in inventory. Initial delivery of 50,000 units to GE scheduled for February 1975 as per agreement terms.
Thompson read the cable.
He read the yield figure.
91.3%.
He set the cable down.
He picked up the phone and called his department director.
"The ISMC production line is running," he said.
"What yield?" the director said.
"91.3%," Thompson said.
A silence on the phone.
"That's better than our gallium arsenide LED production line," the director said.
"Yes," Thompson said.
"In the first month," the director said.
"In the first shift," Thompson said.
Another silence.
"Harold," the director said.
"Yes," Thompson said.
"Did we pay enough?" the director said.
Thompson was quiet for a moment.
"We paid the price they asked," he said.
"That's not what I asked," the director said.
"No," Thompson said. "In retrospect, probably not."
"How far ahead of us are they?" the director said.
"Production yield at 91% on their first shift," Thompson said. "Our best LED production line — the red and green LEDs we've been making since the 1960s — runs at 83% on a technology we've had for a decade." He paused. "They're producing a brand new technology at a yield we haven't achieved on our mature technology."
"That's a process engineering capability," the director said.
"Yes," Thompson said. "They built an institute specifically for process engineering. SPEI — the Shergill Process Engineering Institute. We've been tracking it since it was announced in August. The specific mission of the institute is industrial scale-up — translating laboratory results into production. The LED production line is their first major output."
"And the institute started in August," the director said.
"Yes," Thompson said. "Four months from founding to a working production line at 91% yield."
The director was quiet.
"What do they do next?" the director said.
"The institute's second project," Thompson said, "according to our intelligence, is the LED packaging automation. Making the device packaging faster and cheaper. After that, the driver electronics integration — putting the power control circuitry into the LED package so it becomes a direct replacement for incandescent bulbs without external ballast."
"When?" the director said.
"If they move at the same pace," Thompson said, "eighteen months to two years."
"For direct incandescent replacement," the director said.
"Yes," Thompson said.
"Harold," the director said.
"Yes," Thompson said.
"Do we need to accelerate our own production timeline?" the director said.
"We've had this conversation with the R&D leadership," Thompson said. "The consensus is yes. We need to build production capacity now, using the licensed process, so that when the market transitions we have scale. If we wait for the market to transition and then build capacity, we'll be two years behind ISMC's own production and they'll have supply agreements in place that we'll struggle to displace."
"The first delivery is February," the director said. "Fifty thousand units."
"Fifty thousand units from ISMC's production," Thompson said. "We will sell them under GE's brand in the American market. The margin on the first batches will be thin — we're paying the production cost plus the royalty on top of the license. The margin improves as our own production capacity comes online." He paused. "The business case is about the royalty flow going the other direction eventually. When we're producing our own LEDs at volume under the license, the royalty is a cost of doing business. The current arrangement is a transition."
"The transition timeline," the director said.
"Two years," Thompson said. "If we move immediately."
"Move immediately," the director said.
In Tokyo, the Sony headquarters meeting on January 2nd had a different character.
Matsumoto Akira had been the head of Sony's optical semiconductor research division since the licensing negotiations in September, and he had been the person who had flown to Gorakhpur in August to see the ISMC facility before the commercial agreements were signed. He had walked through the research programme. He had seen the production line under construction. He had looked at the six-inch wafer reactors and the phosphor dispensing system and the quality control station and had spoken with Suresh for forty-five minutes about the susceptor geometry and the temperature uniformity and the yield trajectory.
He had come back to Tokyo and written a report that was eleven pages long.
The report had described the ISMC facility as: two to three generations ahead of any comparable facility in Japan in terms of process capability and equipment sophistication. The SPEI process engineering approach is a specific institutional innovation — the systematic application of scale-up engineering knowledge to production problems — that has produced results in LED manufacturing that we would estimate require three to five years of independent development to replicate.
The report had also said: We should acquire the license and treat the production process documentation as the primary asset. The patents themselves protect the core technology. The production process documentation tells us how to catch up.
The license had been signed.
The seventy-five million dollars had been paid.
The production process documentation had been received — a 400-page technical document covering every aspect of the production process from reactor conditions to packaging specifications. Matsumoto had three teams of engineers reading it.
On January 2nd, the cable from ISMC arrived.
First production shift. 91.3% yield. 84.2 lumens per watt.
Matsumoto read it.
He walked to the window of his office on the forty-second floor of Sony's Tokyo headquarters building. Outside: Tokyo in January, the grey-blue quality of a Japanese winter morning, the city extending to the visible horizon in every direction with the specific density of the most economically active metropolitan area in Asia.
He thought about the transistor radio.
He thought about Ibuka-san flying to New York in 1955 to license the transistor. He thought about what Sony had built from that license — the TR-55, the first transistor radio, the beginning of a consumer electronics empire that had defined what Japanese industrial development could achieve.
He thought about what Sony was going to build from the LED license.
The applications were clear and in his view underappreciated by the Western licensing analysis. The Western companies — GE, Philips — were thinking about LED lighting. The replacement of the incandescent light bulb. The efficiency improvement. The energy saving.
Matsumoto was also thinking about lighting.
But he was thinking about something else.
He was thinking about displays.
A white LED was a light source that could be made very small and very precise and very controllable. A display — a television screen, a computer monitor, a future device that didn't yet exist but that the logic of the technology pointed toward — required an array of very small, very precise, very controllable light sources. The LED was the foundation of the next generation of display technology in a way that the Western companies had not fully articulated in their competitive analysis.
He was thinking about the back-lighting of displays. The thin flat light source that could be placed behind a liquid crystal panel — a technology that was itself in early development, that Sony was investing in — to create a flat display device.
An LED-backlit flat panel display.
He wrote this in his notebook. Not as a product specification. As a direction.
The direction required: efficient white LEDs in very small form factors, at high brightness, controllable in brightness. All of these requirements pointed at ISMC's technology as the enabler.
He looked at the cable again.
91.3% yield on the first production shift.
He wrote another note in his notebook: Invite ISMC production team to Tokyo for knowledge exchange. Reciprocal basis — send our process engineers to Gorakhpur. Build the relationship directly.
He called his assistant.
"Draft a letter to Dr. Chandra at ISMC Gorakhpur," he said. "Congratulations on the production milestone. A request for a technical exchange visit at his convenience. Emphasise that we want to build a long-term technical relationship, not just a licensing relationship."
"Yes, Matsumoto-san," his assistant said.
He looked at Tokyo.
He thought: Ibuka-san was right. The companies that identify the transitions and move quickly are the ones that define the next era.
He thought: ISMC has already moved.
He thought: we need to move faster.
Back in Gorakhpur.
The afternoon session was not in the auditorium.
It was in the ISMC production facility — specifically, at the production line, which was running its fifth full shift. The auditorium event had been the morning. The afternoon was the work.
The production line ran.
The three reactors processed their wafers with the automated precision that four months of SPEI process engineering had made possible. The phosphor dispensing system applied the yttrium aluminium garnet phosphor with the consistency that the automated calibration maintained. The quality control station sampled and measured.
Chandra walked the line with Karan.
They walked it the way Karan walked things that mattered — with complete attention, without performing the attention. They stopped at each station. Chandra explained what was happening. Karan asked specific questions.
At the reactor station:
"The temperature uniformity," Karan said. "Post-Suresh-modification. The current gradient across the six-inch wafer?"
"Plus or minus 1.2 degrees Celsius," Chandra said. "The pre-modification gradient was plus or minus four degrees. The susceptor modification brought it to 1.8 degrees in the first week. The subsequent two months of optimisation — the rotation speed, the gas flow distribution — brought it to 1.2."
"Target was 1.5," Karan said.
"SPEI's target was 1.5," Chandra said. "The team improved on it."
Karan looked at the reactor.
"The six-to-eight-inch wafer transition," he said. "When?"
"The next SPEI project phase," Chandra said. "The eight-inch reactor design is on the drawing board. The SPEI semiconductor division's assessment is fifteen months from design to validated production."
"Fifteen months," Karan said.
"The six-inch transition took six months," Chandra said. "The eight-inch is more complex — the susceptor design challenge scales non-linearly with wafer area. But the methodology is now established. The knowledge base has the six-inch solution in detail. The eight-inch solution starts from a better place."
"SPEI's second project," Karan said.
"After the eight-inch transition," Chandra said, "the packaging automation is the third project. The current packaging is semi-automated — the wire bonding and the dome encapsulant application have manual steps that limit throughput. Full automation is the difference between 5,000 devices per shift and 15,000 devices per shift."
"Aditya wants 15,000 by Q3," Karan said.
"Aditya's projections," Chandra said, with the specific quality of someone who had been on the receiving end of Aditya's projections, "are thorough."
"He's usually right," Karan said.
"He's usually right," Chandra conceded. "Q3 is achievable for the packaging automation. Q2 is aggressive."
"He said Q2," Karan said.
"I know," Chandra said.
They walked to the quality control station.
The QC station was the line's final checkpoint — the measurement of the sampled devices before they went to packaging. The measurement system was a custom-built photometric instrument that ISMC's measurement team had developed specifically for the production line, calibrated to the NPL Teddington standards that the third-party certification had established as the reference.
The current reading on the QC station's display: efficacy 84.6 lumens per watt. Forward voltage 3.2V. Colour temperature 5,580K. CRI 76.
"Above specification on all parameters," Chandra said.
"The CRI," Karan said.
"76," Chandra said. "Minimum specification is 75."
"The improvement pathway," Karan said.
"The phosphor blend," Chandra said. "The single YAG:Ce phosphor gives CRI in the 75-80 range. Adding a second phosphor — a red-emitting europium phosphor — can bring CRI above 85. The trade-off is efficacy — the red phosphor absorption reduces output by approximately 8%."
"When?" Karan said.
"The ISMC research programme has the dual-phosphor blend in late-stage development," Chandra said. "First quarter prototype. The production scale-up — again through SPEI — probably Q3 or Q4."
"CRI above 85," Karan said.
"Consistently," Chandra said.
"That opens the residential market," Karan said.
Chandra looked at him.
"Residential application requirements are stricter," Chandra said. "CRI above 80 for most residential applications. The current device at CRI 76 is acceptable for industrial and commercial — warehouses, offices, factories. The residential market wants CRI 80 or above because the colour rendering at lower CRI makes skin tones look wrong and food look unappetising."
"Aditya's residential market entry is 1977," Karan said.
"1977 is achievable," Chandra said. "With the dual-phosphor blend on schedule, yes."
They walked to the end of the production line.
The packaged devices coming off the line were going into inventory in the specific format — tape-and-reel, 500 devices per reel — that the GE and Philips specifications required. The inventory was growing at the rate of approximately 5,000 devices per shift. At current production pace, the 50,000-unit GE February delivery was achievable with margin.
Karan looked at the inventory shelves.
At the reels of packaged LEDs — each reel 500 devices, each device an 86 lumens per watt white light source, produced in Gorakhpur, carrying the ISMC brand, about to be shipped to the United States where they would be installed in their first commercial applications.
He thought about the first application.
He had discussed this with the GE technical team in November. The first commercial application of the GE-licensed ISMC white LEDs would be in industrial indicator lights at a manufacturing facility in Ohio. Not glamorous. Not visible. Forty indicator lights in a machine control panel, replacing forty red incandescent bulbs with forty white LED indicators that would last fifty times as long and use a tenth of the electricity.
Forty indicator lights in Ohio.
The beginning.
"Chandra," he said.
"Yes," Chandra said.
"The production mission," Karan said. "The one I gave in August. December 1974, 85% yield, commercial throughput."
"Yes," Chandra said.
"December 28th," Karan said.
"Yes," Chandra said.
"Two days before the deadline," Karan said.
"Three days before the deadline," Chandra said. "December 28th. The deadline was December 31st."
"SPEI's first mission," Karan said.
"Completed," Chandra said.
They stood at the end of the production line with the inventory shelves in front of them and the line running behind them and the specific quality of a moment that was the conclusion of something that had begun twenty-two months ago with a reactor run that produced a device at 1.3% efficiency.
"SPEI's second mission," Karan said.
"Eight-inch wafer transition," Chandra said. "Fifteen months."
"Twelve," Karan said.
Chandra looked at him.
"The SPEI methodology is established," Karan said. "The team knows how to do this now. The knowledge base has the six-inch solution. The starting position for the eight-inch is better. Twelve months."
"Suresh needs a week off first," Chandra said.
"Suresh gets a month," Karan said. "He earned it. The twelve-month clock starts when he's back."
"He'll be back in a week," Chandra said.
"He has a month," Karan said.
"I'll tell him," Chandra said. "He'll be back in a week."
Aditya's review.
At four in the afternoon, Karan and Aditya sat in the ISMC financial director's office — a room that had been established as the de facto financial management centre for the LED programme since the licensing negotiations began — and went through the numbers.
Aditya had his notebook and his financial model, which was a 47-sheet spreadsheet that had been built and revised over six months and that he kept in a locked drawer and did not share with anyone except the specific auditors who needed to see it for the legal documentation.
"The GE-Philips upfront," he said.
"75 million US dollars," Karan said.
"Received December 15th," Aditya said. "Converted to rupees at the commercial rate on December 16th. The rupee amount is—" he checked "—672 crore rupees at the December 16th rate."
"672 crore," Karan said.
"The Sony-Toshiba upfront," Aditya said. "75 million US dollars. Received December 19th. 671 crore rupees at the December 19th rate."
"The rate moved," Karan said.
"The dollar strengthened slightly on December 18th," Aditya said. "Irrelevant in the context of the total, but noted."
"Total upfront," Karan said.
"1,343 crore rupees," Aditya said. "Equivalent to approximately 150 million US dollars."
"And the royalty projections," Karan said. "The ones we've been arguing about."
Aditya looked at him.
"My projection or your projection?" Aditya said.
"Both," Karan said. "Start with yours."
Aditya turned to a specific sheet in the model.
"The royalty income is a function of: the volume of LED devices sold under license in each territory, multiplied by the sale price per device, multiplied by the five percent rate." He paused. "The specific assumptions that drive the divergence between your projection and mine are the adoption rate and the market size."
"Tell me your assumptions," Karan said.
"My assumption is that LED lighting achieves 15% market penetration in the American and European lighting markets by 1985," Aditya said. "The current global lighting market is approximately 8 billion US dollars annually. Growing at 4% per year, the 1985 market is approximately 11.8 billion. 15% penetration is 1.77 billion in LED sales in the American and European markets. At five percent royalty, that's 88 million US dollars in royalty income from the GE-Philips license alone, in 1985."
"Per year," Karan said.
"Per year, in 1985," Aditya said. "Growing from zero in 1975 to 88 million in 1985 as adoption accelerates."
"The total over fifteen years," Karan said.
Aditya checked the model. "Using a ramp curve — slow adoption in the first five years, accelerating in years 6-10, mainstream by years 11-15 — total royalty income from GE-Philips alone over fifteen years: approximately 620 million US dollars."
"And the Sony-Toshiba license," Karan said.
"The Japanese and Korean markets adopt faster," Aditya said. "Japan's industrial adoption curve is steeper than America's historically. My projection for Sony-Toshiba royalty income over fifteen years: approximately 480 million US dollars."
"Total royalty income, both licenses, fifteen years," Karan said.
"Approximately 1.1 billion US dollars," Aditya said. "In addition to the 150 million upfront."
"Your number," Karan said. "Now tell me why I think it's wrong."
Aditya looked at him.
"You think my adoption rate assumption is too conservative," Aditya said.
"15% by 1985," Karan said. "LED adoption in the residential and commercial sectors. I think you're underestimating the energy cost pressure that's going to drive adoption."
"The energy cost pressure," Aditya said.
"The 1973 oil crisis," Karan said. "The specific response of Western governments to the energy supply shock. In the United States, Congress is already discussing energy efficiency mandates. The specific trajectory — efficiency requirements, energy price increases, the economics of LED versus incandescent shifting as production costs fall — pushes adoption faster than your 15% by 1985."
"What's your number?" Aditya said.
"25% by 1985," Karan said.
Aditya typed something into the model.
He looked at the result.
"At 25% penetration," he said, "the GE-Philips fifteen-year royalty income is approximately 1 billion US dollars."
"And Sony-Toshiba?" Karan said.
"At proportionally higher penetration for Japan—" Aditya typed again. "About 800 million."
"Total royalty income at my assumption," Karan said.
"1.8 billion US dollars," Aditya said. "Plus the 150 million upfront. Just under two billion US dollars over fifteen years from two licensing agreements."
He looked at what he had typed.
He looked at Karan.
"You might be right about the adoption rate," he said.
"I know," Karan said.
"If you're right," Aditya said, "the royalty income alone from these two licenses over fifteen years is roughly equal to the petroleum division's annual revenue at current production levels."
"Yes," Karan said.
"From a programme that cost—" Aditya checked "—the total programme cost from March 1973 to December 1974 was 47 crore rupees. The SPEI scale-up programme from August to December 1974 was an additional 8 crore rupees."
"55 crore rupees total," Karan said.
"Return on investment," Aditya said. He was not asking a question. He was completing a calculation.
"Don't do that calculation aloud," Karan said.
Aditya looked at him.
"Why not?" Aditya said.
"Because the number will make us think about the money rather than about the technology," Karan said. "The money is the consequence. The technology is the cause. We should continue thinking about the cause."
Aditya was quiet for a moment.
"The cause," he said.
"The next SPEI mission," Karan said. "The eight-inch wafer transition. The packaging automation. The dual-phosphor development. The driver electronics integration." He paused. "The thing that makes the residential market entry possible in 1977. That is what the 55 crore rupee programme produced — not the licensing income. The capability to continue producing things that the world needs."
Aditya looked at the notebook.
He wrote something.
He showed it to Karan.
The note said: The money is the consequence. The technology is the cause.
In the evening, Sakshi read the newspaper.
She was in the sitting room at the Gorakhpur villa — the same sitting room where, a year ago, Karan had been writing in his notebook at the desk after the mall launch while she was reading, and where the specific quality of the room had the accumulated character of a space that had been lived in rather than merely occupied.
The newspaper was the January 2nd Hindustan Times.
The story was on the front page, below the fold: ISMC White LED Production Begins: India's Historic Semiconductor Achievement Enters Commercial Phase.
She read it.
She read it the way she read things that were about Karan's work: with the complete attention that came from caring about the work rather than about the publicity of the work. She was interested in what had happened, not in how it was being described.
She read the production numbers.
She read about the GE and Philips and Sony and Toshiba licenses.
She read about the 150 million dollar upfront payment.
She looked up from the newspaper.
Karan was at the desk.
He was writing in the notebook.
"150 million dollars," she said.
"Yes," he said, without looking up.
"In your account," she said.
"In ISMC's account," he said.
"Which is your account?" she said.
"Which is the programme's account," he said.
She looked at the newspaper.
"What are you going to do with it?" she said.
"SPEI's next phase," he said. "The eight-inch wafer transition. The packaging automation. The research programme for the next generation device." He paused. "The LED driver electronics programme."
"The money goes back into the programme," she said.
"The money is the programme's," he said.
She was quiet for a moment.
"Is there anything left over?" she said.
"There is a reasonable amount left over," he said.
"I want a new library," she said.
He looked up.
"At the Gorakhpur technical institute," she said. "The student library. It's adequate. It could be much better." She looked at the newspaper. "The LED programme employed 217 people. Most of them were trained somewhere. The training happened in facilities that were adequate. The next generation of ISMC engineers needs a better starting point."
"A new library," he said.
"A proper one," she said. "With the journals. The Physical Review Letters back catalogue. The semiconductor engineering texts. The specific materials that the students need and that are not currently in the adequate library."
"How much?" he said.
"I've been researching," she said.
He looked at her.
"Since when?" he said.
"Since October," she said.
He was quiet.
"How much?" he said again.
"Two crore rupees for the building and equipment," she said. "The collection is another eighty lakh. The annual maintenance is fifteen lakh per year." She paused. "The fifteen-year cost is approximately five crore rupees total."
"Out of 1,343 crore rupees," he said.
"Yes," she said.
"That's—" he started.
"Correct," she said. "It's a small fraction of a large number. But the students who use that library are the ones who will run the next generation of ISMC. The fraction is well spent."
He looked at her.
"Done," he said.
She nodded.
She picked up the newspaper again.
"The Sony and Toshiba engineers who are coming to visit," she said. "For the production process documentation exchange."
"Yes," he said.
"When?" she said.
"March," he said. "A team of twenty."
"They'll need accommodating," she said.
"The institute has guest facilities," he said.
"Adequate ones," she said.
He looked at her.
"I'll speak with Meera," he said.
"Good," she said. She returned to the newspaper. "The library building can start this month. The collection takes longer to acquire."
He looked at her for a moment.
"Sakshi," he said.
"Yes," she said.
"You've been running the philanthropic programme of this company without a title or a budget for four years," he said.
"I've been making suggestions," she said. "You implement them."
"The medicine drawer," he said.
"Your mother's idea," she said.
"The library," he said.
"The students' need," she said.
"The worker housing school," he said.
"Obvious requirement," she said.
He was quiet.
"You should have a title," he said.
She looked at him over the newspaper.
"No," she said.
"Why not?" he said.
"Because titles make you attend meetings," she said. "And meetings take you away from the actual work." She returned to the newspaper. "I'll keep making suggestions. You keep implementing them."
"Fair," he laughed.
He went back to the notebook.
He was writing the SPEI second mission parameters — the eight-inch wafer transition specification, the timeline, the team structure, the knowledge base requirements that the six-inch programme had generated and that would inform the eight-inch's starting point.
The year was two days old.
The production line was running.
The licenses were signed.
The library would be built.
The work continued.
It always continued.
Outside, Gorakhpur was doing what Gorakhpur did at eight in the evening in January — the winter night, the specific cold of a north Indian January that was clean and clear, the stars visible in the way they were visible in a city that had a Shergill captive power grid and therefore fewer grid outages and therefore more consistent street lighting and therefore a specific quality of visibility in the clear winter air that was different from what had been here five years ago.
The factory was running the evening shift.
The production line was producing white LEDs at 91% yield.
In the sitting room of the villa, Sakshi was reading about what had been produced.
At his desk, Karan was planning what came next.
This was how it worked.
End of Chapter 176
ISMC White LED Programme — Status, 2 January 1975
Production Line:
Location: ISMC Facility, Gorakhpur Status: Operational from December 28, 1974 Reactors: 3 (6-inch wafer, MOCVD) Throughput: ~5,000 devices/shift (design: 5,200) Yield: 91.3% (first shift average across four shifts: 90.8%) Efficacy: 84.2 lumens per watt average (specification: ≥80 lm/W) Inventory (January 2, 1975): 23,412 units
SPEI Semiconductor Division Contribution:
Mission assigned: August 20, 1974 Problem solved: Temperature uniformity on 6-inch wafer (yield improvement 68% → 91%) Solution: Susceptor geometry modification (Suresh, Week 2, SPEI Programme) Time from assignment to yield specification: 3 weeks Time from assignment to production line commissioning: 4 months
Licensing Agreements:
GE-Philips Consortium (USA/Europe):
Upfront: $75 million USD Royalty: 5% on sales Term: 15 years Territory: USA (GE) and Europe (Philips) Shergill Electronics rights: Non-exclusive sales in all territories
Sony-Toshiba Consortium (Japan/Hong Kong/Korea):
Upfront: $75 million USD Royalty: 5% on sales Term: 15 years Territory: Japan, Hong Kong, South Korea Additional: Production process documentation access Shergill Electronics rights: Non-exclusive sales in Sony-Toshiba territory
Shergill Electronics exclusive territory: All of Asia (excluding Japan, Hong Kong, South Korea), Middle East, Africa, South America
Financial:
Total upfront received: $150 million USD / ₹1,343 crore Projected 15-year royalty income (conservative): ~$1.1 billion USD Projected 15-year royalty income (Karan's estimate): ~$1.8 billion USD Total programme cost (March 1973 - December 1974): ₹55 crore
Next Phase:
Eight-inch wafer transition (SPEI Mission 2, 12-month target) Packaging automation (SPEI Mission 3, Q3 1975) Dual-phosphor development (Research, CRI >85) Driver electronics integration (Research, residential market enabler)
Gorakhpur Technical Institute Library: Approved January 2, 1975. Budget: ₹2 crore construction, ₹80 lakh collection. Annual maintenance: ₹15 lakh. Proposed by Sakshi Shergill.
First commercial delivery: 50,000 units to GE, February 1975. Application: Industrial indicator lights, manufacturing facility, Ohio, USA. The beginning.