With the CPU decided, the focus shifted to the most critical component.
Yuji Naka leaned forward, bracing his arms on the desk.
"President, the CPU is the easy part. The graphics chip is the real challenge." Naka looked at Takuya Nakayama. "Ken Kutaragi has been talking big, saying he'll develop a microprocessor that will crush existing architectures in raw computing power. We need external help for our graphics chip."
"Who do you have in mind?"
"Jensen Huang," Naka said. "When Sega invested in NVIDIA, you signed an agreement with him. NVIDIA is supposed to help Sega customize the GPU for our next-generation console. Now that pre-research is underway, we need to go to North America and finalize this in person."
Naka's enthusiasm grew as he spoke.
"And it's not just the graphics chip. The network module is absolutely crucial for the next-generation console. Single-player games have a ceiling—not every game can achieve the success of Final Fantasy VII. The future trend is definitely online multiplayer and digital distribution. While we're in North America, we should also visit Silicon Valley Online. Let's bring Mark Cerny along. We can take their experience with Battle.net and MMORPGs on PC and apply it directly to the new console's underlying network architecture. This will save us years of trial and error."
Takuya Nakayama pulled open a drawer, grabbed the faxes he had just set on the desk, and tossed them directly in front of Yuji Naka.
"Look at this."
Naka picked up the faxes. Hideki Sato leaned in to see.
The NVIDIA logo was printed on the papers.
"This is a project progress report sent last night by the financial representative stationed at NVIDIA by our North American branch," Takuya said, leaning back in his chair. "Flip straight to the last page."
Naka quickly flipped to the back.
His gaze landed on the final line.
Codename: NV10.Projected start date: First half of 1998.
"Hardware T&L?" Sato's voice rose as he read the technical notes. "Transform and Lighting will be handled entirely by the graphics chip? Is the era of CPUs brute-forcing all polygon modeling and lighting calculations finally over? So the graphics card won't just be responsible for texturing and rasterization anymore?"
Sato knew exactly what this technology meant.
With the graphics chip independently handling geometric transformations and lighting, the CPU's processing power would be completely freed up to handle more complex physics collisions and AI logic.
The number of polygons on screen would explode exponentially.
"This isn't just a graphics card anymore," Yuji Naka said, his eyes fixed on the target specifications. "This is a dedicated processor designed specifically for graphics data!"
Takuya Nakayama took a sip of water from the glass on the table.
"Jensen Huang calls it a GPU," Nakayama said, setting the glass back down. "A Graphics Processing Unit. He's planning to use this chip to redefine the entire 3D hardware industry."
Naka began to crunch the numbers.
"Project initiation in the first half of 1998," Naka said, tracing a line on the paper with his finger. "Based on NVIDIA's current iteration speed and tape-out cycles, they could have engineering samples ready as early as late 1999. If we get involved now and request a custom console version developed in parallel..."
Naka looked up.
"First quarter of 2000. That's when we could get our hands on a mass-production custom version of this chip."
Nakayama nodded.
"That aligns perfectly with my own timeline," he said, clasping his hands on the table. "Nobuyuki Idei gave Ken Kutaragi two years to produce a prototype. The PlayStation 2 won't be released until 2000 at the earliest. We're not falling behind at all."
Hideki Sato took the faxed documents and studied the architectural design closely.
"A perfect match," he muttered.
Yuji Naka turned to look at him.
"The UE Engine," Sato said, pointing to the documents. "The UE Engine led by Carmack still relies somewhat on hardware acceleration at its core. If the new console is equipped with this customized NV10 GPU, combined with the UE Engine's vast versatility, Sega's development ecosystem will become completely invincible."
Sato's words quickened.
"Third-party developers using the UE Engine on our machine won't even need to grind through low-level assembly code. They can just call the engine's API and render cinematic-quality 3D graphics! Development costs and cycles will be drastically reduced!"
Yuji Naka immediately chimed in.
"And with the porting barrier created by the SH-4," Naka said, slapping his thigh. "Easy to develop, hard to port. Third parties, in their quest for profit, will have no choice but to shift all their production capacity to Sega. If Sony wants to snatch games, they'll have to pay several times more to beg people to rewrite the code from scratch!"
Takuya Nakayama looked at his two key lieutenants.
Technology, ecosystem, business logic.
The loop was complete.
"Since you both understand the situation," Takuya said, standing up and taking his suit jacket from the hanger to put it on, "let's stop this armchair strategizing in the office."
He walked to his desk, picked up the faxed document, and stuffed it into a manila envelope.
"Book the flights," Takuya said, handing the envelope to Yuji Naka. "We're heading to California tomorrow."
Yuji Naka took the envelope. "What's the itinerary?"
"Fly to San Jose first. We'll meet Mark Cerny at NVIDIA Headquarters, grab coffee with Jensen Huang, and sign the custom host agreement for the NV10," Takuya said, striding toward the door. "Then we'll head to Silicon Valley Online. You two can discuss the preliminary research for the new Sega console's network architecture. Go pack your bags—I'll have my assistant book the flights."
Hideki Sato and Yuji Naka immediately rose to follow him.
San Jose, California.
The February sun shone brightly over Silicon Valley, the air carrying a hint of dryness.
The business suite on the top floor of the Marriott Hotel.
Takuya Nakayama leaned back on the sofa, a cup of black coffee in his hand.
The ten-plus-hour transpacific flight and the jet lag left his head feeling heavy.
Hideki Sato and Yuji Naka sat opposite him.
A pile of technical sketches lay spread out before them, covered in a dense web of motherboard traces and chipset architecture.
They had landed less than four hours ago, and sleep was the last thing on their minds.
The doorbell rang.
Yuji Naka stood up to answer it.
Mark Cerny walked in, carrying a slightly worn briefcase.
He had just driven over from San Francisco.
"Mark." Takuya Nakayama set down his coffee cup and pointed to the empty sofa beside him.
Mark Cerny sat down, placed his briefcase on the coffee table, unzipped it, and pulled out a stack of printed documents.
"Sony's making a move." Mark pushed the documents to the center of the coffee table. "I've checked with a few friends at Sony North America. Ken Kutaragi's team has been aggressively contacting major storage manufacturers, making extremely demanding requests for the DVD drive's read speed and caching mechanisms."
Hideki Sato picked up the documents and began to flip through them.
"Nobuyuki Idei really gave him an unlimited budget," Hideki Sato said, tapping the paper with his knuckles. "At this rate, their new machine will definitely ship within two years."
"That's exactly why we're here." Takuya Nakayama stood up and walked to the floor-to-ceiling window, looking down at the streets of San Jose. "Sega's new console can't just rely on the CPU to hold its ground; it must gain a decisive advantage over Sony in graphical processing power."
He turned to face Mark.
"After you've caught up on your sleep, we're heading to NVIDIA tomorrow. You understand PC architecture and console low-level systems. When we discuss the technical details with Jensen Huang, you and Mr. Sato will take the lead."
Mark nodded and put the documents away. "No problem."
The next day, at NVIDIA Headquarters.
The office area was a cacophony of clacking keyboards. Whiteboards covered in blueprints and empty coffee cups littered the space.
With 3dfx's Voodoo2 set to launch next month, all of NVIDIA was in a state of intense, feverish preparation.
In a modest-sized conference room, Jensen Huang pushed open the door.
He wore a black jacket and held several test reports he had just brought from the R&D department.
"President Nakayama," Jensen Huang said, pulling out a chair and sitting down. He tossed the reports onto the table. "Congratulations on the promotion. Wall Street was buzzing over Sega's financial report last year."
"Empty praise," Takuya Nakayama replied, taking his seat opposite him. Sato, Yuji Naka, and Mark followed suit.
There was no further small talk.
During their transoceanic phone call before this meeting, Takuya Nakayama had already laid out the general terms of their proposal.
Jensen Huang clasped his hands on the table.
"Sega wants to use the NV10 architecture for its next-generation console," Jensen said, getting straight to the point. "I've considered this carefully. Technically, it's completely feasible. The NV10 is a revolution in graphics processing. Stuffing it into a console would absolutely crush the current machines in terms of visual performance."
He paused, his tone shifting.
"But there's a fatal problem. Manpower."
Jensen grabbed a whiteboard marker from the table and drew two parallel lines on the whiteboard.
"The top line is the development progress of the NV10 for PC. The bottom line is the custom console version you're asking for."
He drew an X between the two lines.
"All of NVIDIA's core engineers are currently thrown into the PC development of the NV4 and NV10. 3dfx is breathing down our necks, and I can't spare even half a team to do the low-level porting and bus debugging for your console. The workload for a custom console chip is immense, and we're short on time. We simply can't run these two development tracks in parallel."
Jensen Huang tossed his pen into the holder.
"Without the manpower, I can't take this on."
Takuya Nakayama leaned back in his chair, tapping his fingers on the armrest.
"If you lack the manpower, Sega will provide it."
Jensen Huang froze.
"Mr. Sato." Takuya Nakayama turned his head toward Hideki Sato.
Hideki Sato immediately opened his briefcase and slid a list of names across the table.
"Sega's Hardware Department will second thirty senior architecture engineers," Sato said, pointing to the list. "Mark will lead the team, supported by twenty low-level programmers. They will fly directly to San Jose next month and be stationed at NVIDIA Headquarters. You provide the core architecture; our people will handle the joint debugging on the console side, bus interfacing, and video memory optimization."
Mark Cerny chimed in.
"We'll take full responsibility for the differences between the PC and console development environments. We won't take up any of NVIDIA's core team's time."
Jensen Huang picked up the list and scanned the names.
Every single one was a top-tier technical expert in the industry.
Sega was truly pulling out all the stops this time.
Jensen Huang set the list down.
"Even if we solve the staffing issue, there's still a major problem," Jensen said, leaning forward with his hands braced against the edge of the table. "Capacity."
He fixed his gaze on Takuya Nakayama.
"The NV10's transistor count is several times that of the previous generation. A chip of this complexity demands the highest precision in manufacturing. TSMC's advanced production lines are already completely booked. As a fabless design company with no leverage, NVIDIA is already fighting tooth and nail just to secure the tape-out and initial shipment of our PC chips."
Jensen's voice grew more emphatic.
"Once the console launches, the initial shipment will be in the millions. If we demand capacity for millions of custom chips, TSMC simply won't have enough lines to spare. If the chips aren't delivered, Sega's console will face supply shortages. That's a burden NVIDIA cannot afford to shoulder."
The conference room fell silent.
Capacity—this was indeed the Achilles' heel of fabless semiconductor companies.
Takuya Nakayama smiled.
He picked up the paper cup in front of him, took a sip of water, and cleared his throat.
"Mr. Huang, have you forgotten what Sega holds in its hands?"
Jensen's brow furrowed.
"Sega acquired a stake in TSMC during their very first public offering."
Jensen's head snapped up.
"Sega is now a strategic investor in TSMC," Takuya Nakayama said, leaning back in his chair with his hands clasped over his stomach. "Sega will negotiate the production capacity for the custom console chips directly with TSMC. We'll secure priority scheduling and a dedicated production line. This will not take up any of NVIDIA's PC-side capacity."
He looked at Jensen Huang.
"Furthermore, Sega can even help you push forward the tape-out schedule for the PC version of the NV10."
Jensen Huang leaned back in his chair and let out a long, heavy breath.
Funding, personnel, production capacity.
Sega had paved over every possible obstacle. With such a "nanny-level" partnership offer, there was simply no reason to refuse.
"Alright," Jensen Huang said, slapping the table and standing up. "NVIDIA will take this deal."
The tension in the meeting room instantly dissipated.
Hideki Sato unzipped his briefcase and pulled out a stack of A3-sized engineering blueprints.
The papers spread across the table with a crisp rustle.
"Now that the general direction is set, let's go over the underlying architecture," Sato said, pointing to the dense network of blocks and traces in the center of the blueprints. "This is the preliminary plan the Sega Hardware Department in Japan spent half a month developing. All our initial ideas on how to modify the NV10's foundation are here."
Jensen Huang pulled up a chair, leaning half his body over the table.
Mark Cerny and Yuji Naka also crowded around.
"Look here," Sato said, pointing his pen at a standalone module on the edge of the schematic. "The DVD decoding chip. We've removed it from the main SoC."
Jensen raised an eyebrow.
"Removed it? Why not integrate the functionality into the CPU or GPU?"
"Integrating it would be suicide," Sato scoffed. "Cramming all those features into a single chip would send the transistor count spiraling out of control. If we tried to tape out something that complex, we'd be lucky to get a 20% yield. The chips wouldn't even be produced, and the entire console launch schedule would be pushed back."
Takuya Nakayama leaned back in his chair and took a sip of the cold coffee on the table.
"Sega doesn't do flashy, superficial tech stacking," Takuya said, setting down the paper cup. "Controllable costs and an on-time launch—those are our iron rules. DVD decoding technology is already mass-produced; we just need to find a cheap, standalone chip to hang off the side. By designing a loosely coupled SoC, the GPU doesn't have to swallow all this miscellaneous work. The CPU, the standalone decoder, and the GPU each handle their own domain, leaving the rest of the chip area for raw graphics computing power."
Jensen nodded repeatedly.
These Japanese guys know their stuff.
Reducing the on-chip integration would cut NVIDIA's tape-out pressure by more than half.
Please Support me by becoming my Patreon member and get 30+ chapters.
[email protected]/Ajal69
change @ with a
Thank You to Those who joined my Patreon
