What's really happening with Sun's Java-chip strategy? Is the company's plan to promulgate a series of Java-specific microprocessors at death's door? That's the way one prominent industry analyst has it pegged.

Or has Sun merely been painfully slow off the mark? Under this scenario, the company is poised to disclose at Comdex some real news about the chips its licensees will be making. The announcements will come nearly two years after Sun tipped its intention to design the devices.

It's obvious why some perceive the Java chips as non-starters. However, I think there's life in the technology yet. The challenge for Sun is to start issuing consistent messages.

Take the confusion caused by the recent revelation that Sun has tabled plans to sell any Java chips under its own brand name. That certainly makes it seem like the game is up. But it isn't: Silicon will be forthcoming come from many of Sun's licensees. There are currently five such companies: NEC, Fujitsu, LG Semicon, Rockwell Collins Inc. and IBM. (A sixth company, Siemens, has licensed Java specifically for use in smart cards.)

Even here, the positive spin to be garnered from having multiple sources of Java chips has turned into a negative. Most of the licensees have been unwilling to talk publicly about their plans, fostering the assumption that there are none.

Part of the problem is undoubtedly geographic. Asia-based NEC, Fujitsu and LG aren't used to playing American-style public relations games. However, a bigger stumbling block may be poor coordination between Sun and its licensees.

The key guideline should be balance. On the software side of Java, Sun has sometimes been too talkative. When it comes to silicon, however, the company has passed on many opportunities to present its case.

For the near term, Java silicon will probably be more workable as a core element of ASIC libraries, rather than as standalone chips. That is, engineers with specific applications requiring native Java execution will seek out the technology, specifying the combination of core and peripheral elements needed to handle a particular design. On the other hand, Java chips as an off-the-shelf item may remain a hard sell.

Maybe Sun should take a lesson from its archenemy Microsoft. Namely, take its lumps and keep on going. Sun shouldn't let competitors, critics-or even the apparent, initial indifference of potential customers-deter it from its goal. Java chips are a viable idea; just how viable remains to be seen.