Remembering Nuon, the gaming chip that nearly changed the world—but didn’t

In the Wild West of Silicon Valley startups of the late 1990s, one little company looked like it might accomplish something incredible. VM Labs had some of the best engineering talent in the world, an explosive mix of bright young minds with burning ambition and experienced old hands who once held key positions in companies such as Atari, Sony, and Sega. Their business revolved around a little chunk of silicon codenamed "Project X.” Later, they officially named their dream chip the Nuon. VM Labs believed it might change the world. (See their marketing specs [PDF] or OEM architecture guide [PDF] for proof.)

The Nuon was so much more than a chip—it was a complete multimedia platform with an operating system and a Web browser. It would turn any DVD player in the world into a game console. And at a time when DVD looked like it would soon be everywhere, the Nuon could be right there with it.

VM Labs' goal for the Nuon was huge but straightforward: total market penetration. The company wanted a Nuon chip inside every DVD player. For a time, it actually seemed attainable.

Yet Nuon failed spectacularly. After quietly gathering hype for years at consumer electronics shows and in magazines such as Wired and Next Generation, the Nuon launched more than a year behind schedule in the middle of 2000—the same year that Sony's much-hyped DVD-compatible PlayStation 2 game console debuted. The Nuon was included in only two of the dozens of DVD player models released that year.

Nuon reviews were positive, but a raft of factors—poor timing and diminishing finances chief among them—combined to sideline the platform, sending its creator into bankruptcy before it could gather much steam. The chip that nearly changed the world didn't, and the company that made it disappeared. But Nuon came so very close to transforming home entertainment.

Paradigm shift

Cast your mind back to November 1994. Games were mostly sold on cartridges, while movies came on VHS tapes. Somewhere in America, a Boyz II Men slow jam played on the radio as the Moving Picture Experts Group (MPEG) announced what would become the de facto standard in digital video compression: MPEG-2. MPEG-2 hardly sparked fireworks, but, in the years that followed, it quietly achieved what its predecessor MPEG-1 could not—making analog video obsolete and finally killing off VHS. MPEG-2 did this in large part because of a delivery technology under development at the same time: DVD.

VM Labs founder Richard Miller, a British ex-pat, was working at Atari Corporation when he heard the news about MPEG-2. Miller had been heavily involved in the creation of Atari's 64-bit Jaguar game console, which was already petering out after just a year on the market. "We were very proud of it," he said in a recent interview. "But it did not succeed because of the lack of content and a chicken and egg issue of getting mainstream developers to support it and having to discount the hardware and all these kinds of things."

"I was thinking there must be a better way," he continued. "I saw DVD happening and thought DVD could be a really good vehicle for launching a new game console."

Miller quit Atari and sold his house, and he then started VM Labs in a spare bedroom at the beginning of 1995. It was his second company. His first built the world's first parallel-processing graphics workstation, which was licensed to Atari but failed commercially. Miller had also, earlier in his career, been instrumental in developing an early, A4-sized laptop computer called the Z88.

Miller knew how to design systems and build platforms. He realized that it wouldn't cost much to install a smarter, more powerful processor in place of a DVD player's basic video decoder. The idea was a processor that could handle games, perhaps things beyond games. "I kind of hoped that it would be a real computing platform, much as the Atari ST had been a platform that was used for games as well," Miller said.

Five people invested $200,000 in the company (the first of whom was Miller's landlord), which was enough for Miller to rent office space and hire a few people in April 1995. He also brought in lawyer Nicholas Lefevre, whose biggest claim to fame is being the first to file an antitrust lawsuit against Microsoft (back in 1983). Lefevre had briefly worked with Miller at Atari before leaving to form his own private practice.

Lefevre helped round out the business plan. VM Labs would leverage the analog to digital shift by building a platform around its chip and establishing a two-tiered licensing model. The chip would be licensed to manufacturers, while the software technology would be licensed to content creators and publishers. First the team needed to build everything—chip and tools—from scratch.

Miller used his business plan to chase down more funding. "Atari had built relationships at Motorola," Miller said. "So I got those guys in very early on and they jumped in with both feet and said, 'Hey, we'll make your chips for you.'"

The next step was to find an electronics company that would put the chips in their products. Miller's first thought was of DVD's principal creators, Toshiba engineers Hisashi Yamada and Koji Hase. "I actually just picked up the phone, called Toshiba in Tokyo, and asked for those guys by name," Miller said. "I [eventually] got through to someone who worked directly for these two guys, and we had a couple more calls. He said, 'Why don't you come out to Japan?'"

Miller flew over to Tokyo and built a friendship with the engineers. Still, even with this promising start, he needed help. He had limited business experience and knew little of Japanese culture. Luckily, one of his engineers knew somebody who did—a guy called Bharath Ram, who had just moved on from a successful business development stint in the encryption market with RSA Security.

"Bharath came in and really cultivated that relationship [with Toshiba] because he speaks fluent Japanese," said Miller. "You wouldn't be able to tell he wasn't Japanese."

Ram bridged the divide between American and Japanese business cultures, which was important because Nuon's success depended on support from the "tier one" Japanese electronics companies—principally, Matsushita (now known globally as Panasonic), Toshiba, and Sony.

Early meetings with Toshiba went well. The multi-core Nuon processor that VM Labs was developing packed a big punch in terms of graphics and performance, and the Toshiba representatives were impressed. As early as 1996, the in-development chip could already demonstrate basic ray tracing, and it had a Mandelbrot set (a complex kind of fractal) browser that could produce several frames per second in real time. "Nothing else would do that at the time," former VM Labs Senior Software Engineer Ken Rose said.

Ahead of its time in architecture as well as concept, Nuon featured four fully programmable, very long instruction word (VLIW) processor cores—or "Media Processor Elements"—running at 108 MHz each. It was a parallel processor not unlike the PlayStation 3 that came several years later. Nuon's design suited its function. The MPEs were fast at exactly the kinds of calculations needed for interactive video and games, and they could also be generalized to other media-rich applications such as Web browsing, interactive audio, and the sorts of reference and educational multimedia that was all the rage in the 1990s. Being a programmable processor, Nuon could gain support for new media codecs from a simple firmware update. And being a specialized VLIW solution, Nuon hardware engineer John Mathieson explained, the chip could be dramatically more efficient than software solutions built with the general purpose processors found in personal computers.