Long Bio (6/12)

The NeXT Cube

No detail was too small

There is probably no product in Steve’s career that was hurt more by his perfectionism than the NeXT Cube. No detail seemed too trivial to be overlooked; everything NeXT did had to be perfect.

First with software. When Steve started asking around to know what was the state of the art in computer operating system, he was told the most stable, modern software was called UNIX. It was a very complex but very powerful OS used in universities and by large companies in their mainframes. The most advanced UNIX technology was being developed at Carnegie-Mellon, where Steve hired some of his best programmers, such as Avie Tevanian. He was also told about object-oriented programming, a breakthrough from Xerox PARC which made software development very fast and efficient. This was the easiest and most efficient way to build software, another breakthrough invented at Xerox PARC. So Steve knew his priorities for the NeXT operating system: it would be a UNIX object-oriented system — on top of which would be added a graphical user interface, to make it user-friendly. These were the very ambitious foundations of NeXTSTEP, so ambitious that it would take several years before they would give birth to a stable operating system

Second, of course, was hardware. Steve had been thrilled by the factory that was used to produce Macintosh — he wanted to do even better this time with NeXT. He set up to build the most advanced automated factory in the world, in Fremont, not too far from the Mac factory itself. The NeXT computers would be built untouched by human hands, using robots operated by other NeXT computers. The factory was designed to mass produce NeXT Cubes and bring the costs down with volume... a disastrous choice for the future.

And finally, the design of the machine, of course, had to be a stunner as well. Steve hired frogdesign again, the same firm that had designed Macintosh, and they came up with a perfect black cube built out of magnesium. Although the Cube clearly deserved its place at the SF MOMA, many of its features made it a pain to build: from the perfect right angles to its materials to its color, it was extremely complicated — and expensive — to put together. In addition, Steve had made a point on also designing a “beautiful” board for the Cube. All the electronic components, which are usually on several different pieces of plastic, were melded on a single square board that the chairman considered as beautiful as the case itself. However it was a strenuous problem for engineers to solve.

There was also the problem of the computer storage system: Steve hated floppy disks, so when he learned about a new technology developed by Canon called magneto-optical drives, he jumped on the occasion. This was a brand new way to store data, but it had not been brought to market yet and its price was still astronomical. Steve bet on it and made it standard on the NeXT computer.

The introduction

Because of all its breakthroughs, in both hardware and software, the date of the NeXT computer’s introduction was constantly being put off. Originally, it was supposed to be out in spring 1987, since most universities shop for the next academic year during springtime. But the computer was nowhere near ready at that time! It was rescheduled for fall 1987, then spring 1988, and finally to fall 1988 — on October 12 to be precise.

I think I speak for everybody at NeXT, saying it’s great to be back. [...] I think together we’re going to experience one of those times that occur once or twice in a decade of computing, a time when a new architecture is rolled out that is really going to change the future of computing. And we’ve worked on this for three years. It’s turned out in-cre-di-bly great...

quoted in Randal E. Stross' Steve Jobs and the NeXT Big Thing

The hype was of tremendous proportions. Steve stayed on the stage of San Francisco’s Davies Symphony Hall for three hours, ending his presentation with a duet between a violinist and the NeXT Cube to demonstrate the computer’s breakthrough sound abilities.

The audience was so captivated that barely anyone realized that Steve had dropped two bombs in his speech. First, NeXTSTEP, the computer’s operating system, still wasn’t ready after more than two years of development. It was still in beta version and would not come out before another six months. Second, the Cube’s price was astonishingly high! It cost $6,500, more than twice the $3,000 universities had agreed to pay for a 3M machine. And that was without the $2,000 laser printer or the $3,000 for the hard disk that would prove necessary since the magneto-optical drive was so painfully slow... The Cube was certainly a bargain for everything it came with: the Canon drive alone cost $6,000 on the market, not to mention NeXT’s revolutionary software — but no one had the money to invest in such an advanced machine.

RenderMan & the Pixar Image Computer

The Pixar Image Computer II

In the meantime, Steve’s so-called “hobby”, Pixar, was also making progress. Remember it had been packaged by Lucas as a graphics computer maker. In 1986, shortly after being incorporated, the company had brought to market a graphics workstation: the Pixar Image Computer, or PIC. But even for the professional markets it was aiming at, its price tag of $125,000, the additional workstation it required to work, and its total lack of software made it a very exclusive solution. It essentially appealed to major universities and advanced labs in niche industries. Sales were so disappointing that two years later, in 1988, the company came up with a cheaper system, the PIC II, which sold for $29,500, while the price for PIC was lowered to $49,000.

They also released a breakthrough graphical computer language called RenderMan. The idea behind RenderMan was to do for 3D graphics what PostScript had done for fonts and 2D graphics in general: write a universal language that any machine could use to render 3D graphics (more on PostScript below). Yet the concept was too complicated and way ahead of its time. RenderMan was almost only used in Pixar’s own rendering software, PhotoRealistic RenderMan.

Getting real

Hopes at NeXT

There were a couple of events that led NeXT executives to falsely believe that they were on the right track, instead of realizing they were heading to the wall.

First was a major deal with IBM that was signed in September 1988, just one month before the Cube’s introduction. As we explained before, NeXT’s operating system, NeXTSTEP, was a revolution in software. It was the first UNIX ever to sport a graphical user interface, making an arcane piece of software accessible to mere mortals. We also said that UNIX was used on several computer mainframes, the vast majority of which were still IBM-made. That’s why Big Blue showed a substantial interest in the NeXT operating system: its intention was to buy a right to license NeXTSTEP on its mainframe systems, in order to add GUIs to its UNIX computers. Moreover, IBM was trying to find a way out of its morass with software developer Microsoft.

The personal computer industry in 1988

When Steve Jobs launched the NeXT Computer in 1988, the industry landscape was radically different from its situation in 1985, the year he left Apple.

The success of Macintosh

First, Macintosh, the sales of which were disastrous in its first two years, was now thriving on the marketplace. It all came down to one word: PostScript.

PostScript was a computer language developed by two researchers from Xerox PARC (again), John Warnock and Charles Geschke. Before it came out, printing text from a computer was a painful thing to accomplish. There was no standard, so every program had to know about every printer, and vice versa. PostScript became the standard: every text editor started speaking PostScript, and every printer as well, so they all started to understand each other seamlessly.

Warnock and Geschke started Adobe Systems in 1982 to build high-end laser printers (another one of their inventions from PARC) that were supposed to be the only machines to use PostScript. It was Steve Jobs who convinced them to make it a standard computer language: he cut a deal with Adobe to use their software and had Apple purchase 20% of the new company.

His vision proved right: the year after he left, Macintosh sales skyrocketed thanks to the desktop publishing revolution. Desktop publishing is the technical word for being able to designing beautiful texts and illustrations on a computer and then print them exactly as they appear on the screen (with so-called WYSIWYG software for What You See Is What You Get). This technology only worked on Macs for years, making Apple richer than ever...

The birth of a titan

Yet the biggest change on the personal computing scene was not from Apple but from one of its early partners, Redmond-based Microsoft.

In 1981, Microsoft signed a deal with IBM to be the exclusive provider of the operating system that would run on their PC, a piece of software called DOS. The deal also included the right for Microsoft to sell the system to other computer manufacturers. IBM agreed, as at the time, the prevalent belief in the computer business was that no money could be made from software, only hardware.

Big mistake from Big Blue! Shortly after the IBM PC came out, the market was flooded with so-called PC clones: cheaper computers from other companies that would run IBM PC-compatible software. One of the most popular was Compaq, which started in 1982 with the first “portable” PC.

So throughout the 1980s, IBM saw its share of the PC market shrink drastically, while Microsoft was becoming a software behemoth, selling DOS to basically every computer user that was not using Apple products. The firm from Redmond was turning increasingly independent from IBM and impossible to deal with. The solution? Big Blue intended to launch a new standard for its future products, that would have nothing to do with Microsoft, thus putting them out of business.

NeXT had the opportunity to become the next Microsoft: if the deal with IBM succeeded, their breakthrough software would run on every IBM-compatible PC in the future.

In addition, in June 1989, NeXT signed a deal with Canon, the Japanese maker of the Cube’s magneto-optical drive. In exchange for the exclusive right to distribute NeXT computers in Asia, they agreed to invest $100 million for 15.67% of the company — valuing it at a whopping $600 million!

The NeXT Station

Despite those signs of optimism, the NeXT Cube was a blatant failure on the marketplace. It simply did not sell: universities and students found it way too expensive. Firstly, by 1988, it was common for students to have a Macintosh in their dorm rooms. The days when you had to go to the computer lab to use a workstation like the Cube were long gone. There was also the problem of donations — universities were used to be given, not sold, computers, in the hope that students would use the same computers in their future corporate careers. Finally, the Cube was not as modern as it would have been had it come out the year before: it was monochrome at a time where color started to appear, its magneto-optical drive was a pain to use, and above all, it had very limited software.

The NeXT factory, which had the ability to turn out 10,000 Cubes a month, manufactured an average of 400 a month in 1989! At such rates, the competitive advantage of having an automated factory was turning into a disadvantage — it was a lot more expensive to operate and, lacking the necessary volume, it did not drive the costs down at all.

Steve Jobs tried to adjust by substantially changing NeXT’s strategy:

Our smallest competitor is $1.75 billion these days. [...] We have to get up to a certain level if we want to play in the sandbox.

The Entrepreneur of the Decade, Inc., April 1989

The company stopped selling exclusively to the education market, and tried to make its way into corporate America. NeXTSTEP’s object-oriented development environment was a key advantage in that market: it would allow for companies to write custom software in record time. In order to do so, in March 1989, NeXT partnered with Businessland, the nation’s largest computer retailer at the time, to distribute Cubes in all of its stores — with a $9,995 price tag. Of course this hardly worked.

In parallel, work started on a new, cheaper computer. The NeXT Station or “slab” (because of its pizza-box shape) was eventually introduced in September 1990, along with cheaper Cubes with color and breakthrough video capabilities. The cheapest Station sold for $5,000, still a high price but half of that of the Cube. On stage also came Lotus software, which announced that the revolutionary next release of their spreadsheet called Lotus Improv would run only on the NeXT platform, and would be bundled with every NeXT computer. The goal was to make it NeXT’s killer app (an app that alone justifies buying the machine).

Tough times for Pixar

As for Pixar, it was in a really painful situation by the early 1990s.

First of all, the computer animation department, headed by John Lasseter, had to fight regularly for its survival. Steve Jobs almost shut it down several times throughout 1987 and 1988, until the team had the idea of making animation for TV commercials. That way they could survive and keep all the talents they had spent years to gather, while making some money. For all that, work on “artistic” movies did not stop: the team’s Tin Toy got an Academy Award for Best Animated Short Film in 1988, and the following year, Lasseter earned critical acclaim for his short Luxo Jr. at the SIGGRAPH convention. Steve allowed for the animation department to continue such work because the prestige could be used for selling more PICs — although, ironically, Pixar only made one short movie on their computer in their entire history: Red’s Dream (in 1987).

However, sales of Pixar Image Computers were still extremely disappointing. On April 30 1990, Steve Jobs announced he was shutting down all of the company’s hardware operations, while the staff moved away from Lucasfilm’s premises to new offices in Point Richmond — not far from a Chevron oil refinery. From then on, they would have to focus only on their boss’s new vision: Steve thought that RenderMan was going to become the next PostScript, an open standard adopted by the masses to make 3D renderings at home, just like PostScript had made desktop publishing possible. He was denying the reality of how hard it was to master three-dimensional animation.

If costs were indeed cut a little by this move, it didn’t make Pixar more profitable. The startup was still relying on Steve Jobs’ line of credit, and in 1990 alone, its net operating loss was over $8 million.

In March 1991, Steve went further in his drastic moves to make Pixar survive. He declared he would continue to keep funding it only if he were given back all of the employees’ stock shares. The scheme involved shutting the company down on paper, and creating a “new Pixar” where he was the sole owner. He also fired almost half the staff, keeping only the software programmers as well as Lasseter’s animation department — which was, by then, the only part of the company to bring cash in, thanks to its works in TV advertisement.

Pixar failed nine times over by normal standards, but Steve didn’t want to fail so he kept writing the checks. He would have sold us to anybody in a moment, and he tried really hard, but he wanted to cover his loss of $50 million.

Pixar co-founder Alvy Ray Smith in The Second Coming of Steve Jobs

Many people at NeXT shared that same view...