The First Xbox 2 Prototype
Game war. In Yoo Jae-won's eyes, sparkling like jewels, Dr. Lisa Su and her research team nodded in agreement. It was the same gaze he reserved only for those he loved—Tiffany and his family. Normally, Yoo Jae-won radiated the overwhelming aura of a young man who had built the colossal ID Group, but right now he looked every bit the computer-obsessed nerd. After all, only a tiny handful of people could look at semiconductor components with such affection. Yet neither Dr. Lisa Su nor the researchers found it strange. They, too, were people utterly captivated by semiconductors.
“NVIDIA’s chip is the G80, and ATI’s is the R520,” Dr. Lisa Su explained, using only model numbers like a true expert. The G80 was the chip released as the GeForce 8800GTX, while the R520 was the one that became the Radeon X1800XT. Yoo Jae-won’s eyes went first to the G80. It was the legendary model he had squeezed every last drop of performance from before his regression, starting with the 8800GTX, and it was also a leading candidate for the Xbox’s GPU chipset. The term “sagol” referred to a design so excellent that the company kept releasing minor process upgrades or small revisions as if they were entirely new models. The fact that the market never pushed back despite such repeated milking spoke volumes about how solid the original architecture had been.
“How’s the performance?”
“You’ve always said it yourself, Chairman—seeing it once is better than hearing about it a hundred times. I’ve prepared everything so you can check it directly.”
With an irresistibly charming smile, Dr. Lisa Su pulled away the veil covering the podium. Two brand-new Newegg units, already booted up, greeted Yoo Jae-won warmly. They were the latest 2003 models. Though the computer division had been overshadowed at IDDC 2003 by the Android S3 and AI Eyes, the iWorks and Newegg series continued to release new models every year without fail. Performance received major upgrades annually, while design saw significant refreshes every two years. Somehow, iWorks had settled into even-year updates and Newegg into odd-year ones.
“They’re configured to the tentative Xbox 2 specifications.”
Ever perceptive, Dr. Lisa Su had prepared exactly what Yoo Jae-won wanted. A hexa-core CPU paired with either an 8800GTX or X1800XT and 4GB of memory came close enough to the Xbox 2’s target performance.
“Where did you get the hexa-core CPU?”
“We received an octa-core sample intended for servers and limited it to six cores in the BIOS.”
Yoo Jae-won nodded at her explanation. ID Group was always the first company to receive new products from both AMD and Intel. Beyond Android OS optimization, the group purchased enormous quantities of chips every year to build data centers and cloud systems for machine learning. Because the two companies’ revenues rose and fell according to ID Group’s strategy, they had been treated as top-tier partners for years. This time was no different. The Xbox had already sold over fifty million units cumulatively. The Xbox 2 was projected to sell even more. If either chip were officially adopted, tens of millions of units were guaranteed. That was why they had sent even their secret high-end server models—so Yoo Jae-won could run the numbers.
“Excellent.”
Yoo Jae-won sat down in front of a computer immediately. The system administrator window was already open, displaying the specs clearly. Just as Dr. Lisa Su had said, the Newegg units were configured as close as possible to the Xbox 2 target. He launched Doom 3, a program widely used for benchmarking. Though the game had been out for some time, the addition of HD texture packs and high-resolution options had raised its visual quality considerably. In multiplayer, it still maintained over one hundred thousand concurrent users. Double-clicking the Doom 3 icon, the hard drive made its characteristic grinding sound as loading began.
“Hmm.”
Even on a system that represented the current pinnacle of performance, the hard disk remained the glaring weakness. It was the only computer component that still relied on analog technology. Spinning a motor to read data from the platters made it painfully slow. Yoo Jae-won already knew the solution. If storage were built from NAND flash memory, he would never again hear that grinding hard-drive noise. The problem was that, even for someone as wealthy as Yoo Jae-won, the economics still felt prohibitive. Since time would eventually solve the issue, he decided to wait patiently.
By the time loading finished, the benchmark scenario for Doom 3 began playing on the monitor.
“As expected.”
Running at a stable 60 frames per second in full HD resolution of 1920×1080, Doom 3 looked spectacular. The combination of hexa-core processing power and next-generation GPUs produced visuals that felt overwhelmingly powerful.
“It’s running even better than I hoped!”
Both the system with NVIDIA’s 8800GTX and the one with ATI’s X1800XT delivered stable 60 frames in full HD. Either company could be chosen without issue, as long as the price was right. The CPU performance was also satisfactory. Server chips carried high unit costs, but if unnecessary features were stripped out and custom optimizations were applied for gaming performance, they could deliver better results at a lower price. Memory chips posed no problem at all—ID Electronics was already mass-producing thousands of wafers every day. If everything aligned, the Xbox 2 would fully meet the performance Yoo Jae-won had promised.
“At this level, we should definitely be able to beat Sony’s PlayStation 3, right?”
“Of course.”
Dr. Lisa Su answered immediately and affirmatively. Sony was sharpening its knives for a decisive victory in the second console war against Xbox and had selected IBM as its partner. IBM was offering the CELL CPU, an expanded version of the Power CPU used in the PlayStation 2. While its theoretical performance looked impressive enough to make the Xbox team nervous, Yoo Jae-won, who knew exactly what kind of chip the CELL CPU truly was, found the situation merely amusing. Still, he could not afford to feel entirely at ease.
“The key will be improving yields on the 90nm process.”
Instead, Dr. Lisa Su pointed out ID Group’s sole point of vulnerability: yield. At the current 90nm process yields, it would be impossible to hit the target price point. In particular, the high-end GPU chips nicknamed “Big Beauty” had die areas exceeding 400 mm². Even on a 300 mm wafer, only about two hundred chips could be produced. Fortunately, because Yoo Jae-won had spared no expense expanding semiconductor facilities, the second factory was already on standby. Once yields were secured, the problem would be solved.
“That’s exactly why I came here in person. I have no intention of leaving until we reach C-step (70 percent). You can rest easy.”
At Yoo Jae-won’s words, Dr. Lisa Su and the entire team erupted in cheers. Normally, when a top executive announced they would personally join the front lines, it signaled the beginning of hardship. But Dr. Lisa Su, Professor Lee Jong-hyo, and the rest of the team were people who had already decided to dedicate their lives to semiconductors. They were anything but ordinary.
Yoo Jae-won changed clothes on the spot and threw himself into the research. From that day forward, the development capabilities of the ID Electronics Daejeon Research Institute began to explode in earnest.
While Yoo Jae-won was waging an all-out battle at the ID Electronics Daejeon Research Institute to complete the 90nm process, an equally fierce struggle was unfolding in the National Assembly. The opposition parties had banded together to occupy the main chamber and were blocking ruling-party lawmakers from entering. They had even gone so far as to confine the Speaker of the National Assembly to prevent him from exercising his authority to bring bills to a vote. Street protests had also begun, with demonstrators gathering in front of the National Assembly building and holding candles. To Yoo Jae-won, the scene felt utterly alien. Even after seeing the footage multiple times, the sight of people holding candles instead of the Taegeukgi still struck him as strange. It was a vivid reminder that the year was 2003.
Articles also appeared pointing out that while the rest of the world was nurturing new industries in artificial intelligence and autonomous driving, South Korea’s gridlocked National Assembly was struggling to pass related legislation. This was the handiwork of Vice Chairman Choi Kang-wook, faithfully carrying out Yoo Jae-won’s instructions. As expected, on the internet, articles titled “Opposition parties dragging their feet” received far more clicks than coverage of the street protests. It had only been recently that, when autonomous driving and AI technology appeared on television, proud Koreans had proudly claimed the achievements as Yoo Jae-won’s and drunk deeply of national pride. Yet at the legislative stage—the prerequisite for widespread adoption—the National Assembly had ground to a halt. Public anger naturally turned toward the opposition. With public opinion at this level, the bills might not pass during the regular session, but the likelihood of artificial-intelligence-related legislation passing in the next extraordinary session was very high.
“That aside, I wonder how next year’s general election will turn out.”
While browsing the news page on Nextcom from his private quarters near the research lab, Yoo Jae-won naturally began calculating the likely outcome of the following year’s general election. Many of South Korea’s pressing issues had diverged significantly from their pre-regression trajectory because of his own actions. In particular, President Roh, now in the Blue House, had not ignored Yoo Jae-won’s advice. The most concrete result was the disappearance of the “Conversation with Prosecutors” event. Although the appointment of Kang Kum-sil as the first female Minister of Justice had remained the same, the pointless “Conversation with Prosecutors” initiative had never been pursued. Moreover, the catalyst for that event had been the prosecution’s veto of a female Minister of Justice who had once been a judge. Now that the Corruption Investigation Office existed, even if the prosecution attempted overt resistance, any work slowdown would immediately trigger an audit. Receiving even a single point of criticism in a Gongsucheo audit would severely damage promotion prospects, so the kind of collective defiance seen before the regression was now unthinkable. There were murmurs that the Gongsucheo was overly harsh, but Kim Chang-wan, its director, remained as resolute as ever—whether on the day of his appointment or now. His stance was simple: there were plenty of people in Noryangjin who could become judges or prosecutors if these ones were removed. Those who had passed the notoriously difficult bar exam and become judges and prosecutors all believed that no one else could do the job, a belief that had twisted into a sense of entitlement. Director Kim Chang-wan viewed corruption as incompetence and believed that weeding out such individuals was the path to serving the nation; he investigated diligently. Thanks to him, the number of judges and prosecutors forcibly stripped of their robes each year reached double digits, and the number of successful bar exam candidates increased accordingly. Unlike those who had loudly claimed the courts and prosecution offices would collapse without them, the prosecution—now stripped of its high-ranking officials—simply did its job well. As a result, President Roh no longer felt any need to hold a “Conversation with Prosecutors.”
Furthermore, the relocation of the capital—one of the president’s campaign promises—had not yet become a contentious issue. Simply initiating judicial reform had already provoked massive resistance from vested interests. Dealing with that alone was enough to paralyze the National Assembly; there was no bandwidth left to push capital relocation as well.
“Hmm. Then there probably won’t be an impeachment either.”
The biggest reason the ruling party secured a majority in the 2004 general election had been the impeachment of the president. Internal factional fighting within the Democratic Party, sparked by the “Hudan-hyeop” group and backed by the opposition, had made the passage of the impeachment motion possible. However, the hastily attempted impeachment was rejected by the Constitutional Court and triggered an enormous backlash. The result was a Democratic Party majority achieved on its own. Yet the political landscape of Korea after the regression was remarkably stable. A paralyzed National Assembly hardly matched the word “stable,” but compared to the pre-regression era, the situation was solid enough to rebuild the ID Global Headquarters from the ground up. In that case, impeachment would not occur, and the 2004 general election would enter uncharted territory that no one could predict.
“That’s a relief.”
Though a trace of unease remained, judging by the opposition’s current behavior, their revival seemed unlikely. Moreover, Yoo Jae-won had the Unity National Party as his joker card, so he could afford to feel at ease. Refreshing the Nextcom news page to see if there was any other interesting news, his eyes caught the clock.
“Ah! Has it gotten this late already?”
It was time to return to work. Yoo Jae-won shut down the computer and stood up. The daily routine between his quarters and the research lab was like a squirrel running in a wheel—always the same—yet every day felt new. The more he worked, the more tangible the completion of the technology became, so both Yoo Jae-won and Dr. Lisa Su’s team poured themselves into the effort with equal intensity.
Yoo Jae-won finally concluded his time at the research institute in November, when the wind carried a noticeable chill. It was also the day the first prototype of the Xbox 2—crude in form because it had been hand-built inside the lab—was completed.