Why Apple Pulled the Plug

by Galen Gruman

from the November 1997 issue of MacWorld magazine

How quickly the dream died. In December 1994, Apple ended its Macintosh monopoly, giving Power Computing and Radius licenses to make their own Macs. With Wtndows 95 on the horizon and the success of an Intel/Microsoft duopoly clear, Apple’s leaders and the industry saw a chance for the Mac market to gain a new level of competitiveness, flexibility, and innovation.

Almost a year later, Apple and IBM put aside a yearlong rivalry and unveiled a partnership with Motorola to create the Common Hardware Reference Platform (CHRP), which would remove Apple’s burden of designing Mac hardware and let the entire industry fund the Mac platform’s development.

“We believe that an openly licensed Mac OS running on top of an open, industry-standard RISC hardware platform represents a broadside against the reigning Wintel platform, and will play a key role in our ongoing efforts to greatly increase the presence of Macintosh in all markets,” then Apple CEO Michael Spindler said during the November 1994 CHRP introduction. “We believe today’s announcement is good news for anyone who believes in innovation, competition, and responding to customer needs.”

Second Thoughts

Fast-forward to summer 1997: After months of complaints by Apple officials that licensing was hurting the company, Apple and the Mac OS licensees hammered out a new agreement that apparently satisfied Apple’s financial concerns and let the licensees stay in business. “My position on the clones was mat we had to support them,” former Apple CEO Gil Amelio told MacWeek of the tentative June agreement.

Then Steve Jobs took over Apple in July and sent strong signals that he wanted licensing dead. In a mid-August filing with the Securities and Exchange Commission, Apple said, “The benefits to the company from licensing the Mac OS to third parties may be more than offSet by the disadvantages of competing with them. The company is currently in discussions concerning the nature of such licensing arrangements going forward, including whether or not to extend such arrangements.”

On September 2, Apple essentially ended the licensing program, buying out its most successful competitor, Power Computing. At press time, it appeared that IBM Microelectronics and Motorola Computer Group were preparing to abandon their Mac businesses. Umax Computer was finalizing a deal with Apple that would largely restrict Umax to designing and selling Mac systems in the Far East, and selling or making for Apple sub-$1000 Mac systems, which Apple has been unable to profitably make itself.

Apple says it will honor current licensing deals, which limit licensees to shipping their current systems. Apple also says it will consider licensing companies that target markets Apple does not want. This seems to mean small or nascent overseas markets, such as China and Korea, not proven markets like the United States.

Essentially, the foray into an open Mac market is over.

Kill or Be Killed

How did the promise of fall 1994 devolve into the bitter frustrations that plagued Apple and the licensees in spring and summer 1997?

Simply put, Apple found itself with a do-or-die decision: unable to compete with the licensees’ faster, cheaper systems, and bleeding large sums of money due to lost sales (particularly in its high-margin systems), Apple decided it must end licensing or go out of business. “If we continued losing money, we would have disappeared,” says Apple marketing chief Guerrino De Luca. “Power [Computing] and the others were not in a position to carry this [market] forward if Apple died.”

Late Delivery Eroded Profits

In the second quarter of 1997, about a fourth of ail Mac systems were sold by companies other than Apple. A huge percentage of those were high-end systems from which Apple traditionally made its real profits. And because Apple usually lagged behind the licensees in introducing new technologies, by the time Apple had new systems out, its target customers had already bought elsewhere.

Although CHRP would allow Mac cloners to pay for their own research and development, Apple was unable to deliver the technology until spring 1997, a year later than originally planned. This kept the Mac makers tightly linked to Apple’s own designs and forced them to compete for the same customers, since their products could not be significantly different from Apple’s.

By this sununer, the situation had only gotten worse for Apple: several market researchers predicted tl1at by the end of 1997, Apple would have less tl1an half the market for new Mac sales.

Technology Missteps

Despite Apple’s crowing about its technology leadership, its actual technology delivery fell behind that of the licensees. Umax introduced in-line cache first, Power Computing made the first Mac to use the PowerPC 750 and 1:1 backside cache, and Motorola had the first Macs to use the CHRP architecture. Apple instead adopted the new Mach 5 version of the PowerPC 604e and slightly modified its existing systems.

The result was another round of Apple Macs that cost more and ran more slowly than licensees’ Macs. Macworld Lab tests this summer showed that Apple’s 350MHz Power Mac 9600 was no match for even a 266MHz Motorola StarMax 6000 XL. Apple likely won’t match the StarMax 6000’s speed until early 1998 when Apple integrates CHRP technology.

Given its tedmology disadvantage, Apple’s refusal to certify competing systems smacked of an effort to hold back competition it knew it could not beat. (Apple has denied such an intent.)

In the early 1990s, Apple systems were routinely trounced in performance by PCs, and that trend began again in the mid-1990s, only to be reversed by Power Computing and the other licensees. Given Apple’s inability keep pace with a handful of licensees, it’s an open question whether the company will keep pace with the PC industry now that it has eliminated its competition.

A Shrinking Pie

Since 1995, Apple’s Mac sales had shrunk from about 4.5 million per year to about 1.8 million. The licensees accounted for about 600,000 w1its in 1997, up from a handful in 1995. Revenues per system also dropped, as Apple was forced to reduce prices to be more competitive.

Power Computing, which accounted for about 10 percent of all Mac sales in 1997’s second quarter, went for Apple’s jugular earlier this year. The company developed several fast, inexpensive Macs-based on Apple’s own technologies that outperformed anything from Apple. Power Computing also deemphasized its lower-performance systems, which tilted Apple’s sales even more to low-profit (or even money-losing) low-end Macs.

Motorola and Umax, which each held about 5 percent of sales in that period, targeted a broader mix of users.

When Apple inaugurated its licensing program, it did not anticipate a declining Mac market or such stiff competition for its core customers. It did not expect management chaos; expensive, spectacular product failures, such as the Copland OS and eWorld; or such disappointing investments as OpenDoc and the Newton, all of which caused Apple to lay off thousands of people and lose more than $ 1.6 billion over 18 months.

No Expansion?

Worse, from Apple’s viewpoint, the licensees did little or nothing to expand the Mac market. Instead, they focused on existing Apple customers, advertising in Mac publications almost exclusively. The licensees argued that they needed to get a large-enough share of the market to pay for growth. Ironically, it appears that Apple came to a similar conclusion about its own prospects when it decided to end licensing. The licensees also complained that Apple did not market the Mac to new customers and would not cooperate in joint marketing efforts to new customers.

It is clear that neither Apple nor its licensees were working together to significantly expand the overall Mac market. Compare the lack of cooperation tl1ere witl1 the PC market, where Intel subsidizes most PCs (someiliing Motorola and IBM do not do for PowerPC systems) and where Microsoft aggressively markets Windows (Apple rarely markets the Mac OS).

High Design Costs

Apple officials began complaining this spring that the licensees were not paying their “fair share” of Mac development costs. Apple had been charging a license fee of about $50 per Mac system for the Mac OS and another $50 to $75 for use of Apple’s hardware designs. (CHRP systems did not require a separate hardware fee.) Apple now believes these fees were too low: “The bet that low licensing fees were sustainable was made based on a growing Mac OS, which turned out not to be true,” De Luca says.

After months of negotiations, Apple and the licensees agreed in June to a new Mac OS fee of about $150 to $350 per system, depending on speed. (Compare that with about $50 to $65 for Wmdows 95 and $90 to $100 for Windows NT.)

The licensees also pay about $5000 for certification by Apple for every system variant they make, in addition to an annual certification fee, according to sources. Despite Apple’s protests to the contrary, the licensees were paying Apple real money.

While unhappy with the new fees, the clone makers accepted the June deal. But a few weeks later, Steve Jobs forced out Amelio and backed out of the agreements.

Even after the June agreements, Apple officials complained that the licensing fees were too low, and some hinted iliat Apple would need $500 or more per system to cover Mac OS and Mac hardware development costs.

Why so much? According to industry analysts, Apple’s cost structures are extremely high, given its need to pay for both Mac OS and hardware development. This means the overhead per Mac is more than any licensee would reasonably pay, and is one reason why Apple’s Macs are so much costlier than PCs.

The licensing program magnified Apple’s historically high costs because Apple could no longer recoup its software development costs through its own Mac sales. By contrast, Microsoft pays for Windows development while Intel pays for most PC motherboard development- and both companies sell about 16 times the volume of products that Apple could hope to. However, IBM and Motorola have paid most of the costs for CHRP, and practically all the costs of the PowerPC CPUs.

Microsoft also makes tremendous revenues from its Office programs, which helps subsidize the company’s R&D, much the way Apple’s hardware sales have subsidized OS development.

Crown Jewels

Although Apple has portrayed the licensing failure as a recent economic decision, the reality is that Apple, or at least parts of it, has never been comfortable with sharing the Mac crown jewels. That, plus the constant sniping by Apple officials on licensing for the last six months, has created a fundamentally negative relationship between Apple and the licensees.

Despite some thaws in the relationship and an occasional positive working arrangement, the basic Apple/licensee partnership was not based on trust or respect, merely on financial opportunity. As the market changed and the conditions for mutual success became more difficult, Apple and the licensees pulled apart rather than pulled together. The result was that when the Mac market’s economics went south, licensing became doomed to failure.