Intel's Expensive Disco Ball

Re-Pawn writes "From the NY Times: The Disco Ball of Failed Hopes and Other Tales From Inside Intel (Registration Required.) Seems like Intel is losing market share to other chip makers - this article highlights a few problems that Intel has had including one very expensive disco ball made from a failed attempt to produce projection televisions."

Comments

Just Desserts for Intel

[Anonymous Coward]

Intel has 2 shocking policies: bell-curve grading system and preferential hiring of H-1B workers from China (which includes Taiwan province and Hong Kong) and India.

More than 50% of Intel's workforce in the USA (not China) is current or former H-1Bs. Intel claimed that it absolutely needs Chinese workers in order to build a competitive product: e.g. Itanium. Then, IBM proved Intel wrong by producing the Power5, which is mostly built by American engineers.

Further, Intel has a brutal job evaluation policy: strict bell curve. If an employee falls in the bottom 25% more than once, then the manager shows her the door. Exceptions are made when there is a labor shortage, but officially, the 25% rule is strictly enforced.

I, for one, am glad that Intel is losing. I hope that IBM beats the pants off of it.

Re:Just Desserts for Intel

[Anonymous Coward]

Whatever! The FOCAL process (as it is called) may seem brutal to outsiders, but you have be pretty lazy and completely incompetent to lose your job...and it is more like the bottom 10% that fall into the category that get put on corrective action plans.

Re:Just Desserts for Intel

[Colonel Panic]

As someone who has contracted at Intel, believe me I'm not a fan of the company.

However, many companies now practice the rating and ranking system you describe. And it's not the bottom 25%, it's the bottom 10% from what I've heard.

As for H-1B workers: When I was last contracting at Intel (June 2004) the policy was that all permanent hiring was to be done outside of the US. In the US they could only hire contractors unless there was some very special skills needed. I suspect that this policy is still in place. This is of course worse than your claim that they only hire H1B workers - at least an H1B worker would be paying taxes in the US and contributing to the economy here. Many of Intel's former permanent employees in the US have now become contractors (via layoffs) which means that they can only work for the company for 12 months out of every 18 months (but look on the bright side, you get a six month vacation after contracting there for a year!) and no health benefits.-

Re:Article text

[PhotoGuy]

"What was wrong was that I made the decision to go public on it at the Consumer Electronics Show," he said in a recent interview in Intel's Santa Clara headquarters. "Error of judgment. Mea culpa. I learned a lesson."

I like this statement. And I think it's consistent with what I've known of Intel first hand. A corporation this large and leading-edge, needs to dabble and branch off in "researchy" ways to test out areas of new market potential.

I was involved in a company whose seed money came from a sizeable (to us) contract from Intel, to license our technology in the digital imaging space. They were a great company to work with, talented people, good executives, and they got their demonstration technology, based upon our code, up and life in a respectable time.

The site was never marketed and never took off, but I believe it served their purposes in exploring this potential area of technology. It's a good thing to see a company like Intel taking part in this type of thing.

The only story here, as in the quote above, is that they made a bit of a visible statement about where they were headed, before knowing for sure. Minor mis-step, if mis-step at all.

-d

Intel has done a lot of stupid things

[jd]

Their hiring practices leave a lot to be desired. They prefer contractors to employees, but won't hire a contractor for longer than a year, and there has to be something like a 6 month break before re-hiring. This means there is zero incentive for contractors to do anything worthwhile (they're not going to be around for long, no matter what) and there's no continuity when something does go right.

Then, there's their design strategy - lock everyone else out. By making it damn-near impossible to use a standardized processor socket, anyone who currently uses some other chip-maker is essentially locked out of buying anything Intel. In other words, about now, they're locking out nearly half their potential customers. I'm sorry, but that's just plain dumb.

Their near-violent reactions against people making support chips for the Intel processors means that competitors are going to have to be based on AMD or some other x86 clone, for the most part. A few (eg: Via) will work with Intel, but I've also seen plenty of Intel docs on what breaks when you use Via with some Intel processors. Compatibility sells more products than coercian.

True, most of Intel's competitors aren't too smart on these points, either. But that simply means that the first seriously open competitor is likely to wipe the floor with the lot of them. Transmeta could have. In fact, they could have crushed most of the 32-bit market, if they'd provided people with the means to upload different instruction sets. That capability becomes a liability (it impacts performance and reliability) if nobody can actually make any use of it.

None of the current chip manufacturers has opted to move the southbridge or northbridge into the CPU, despite the fact that this would improve performance, without having to speed the chip up.

Intel moved to copper from aluminium for chip interconnects, because it reduced power consumption. If they moved to silver, they could reduce it further, so the chips could run cooler and/or faster, with no additional work. There's no evidence they're even looking at that.

Instead, Intel are working on projects such as TV decoder boxes running on low-end hardware. This isn't their field. They can't seriously compete in that market, because it's too crowded as it is. There's no money in it.

And now we're told they're going to do MORE of this generalization into markets about which they know nothing, have no solid expertise, no history and no track-record of getting projects complete. They're killing themselves.

What would I do, if I were in their shoes? Easy. I'd shore up the core products, by putting R&D cash into better product differentiation. In other words, cloning AMD's 64/32 is not good enough. That makes them equal to their competitors. Those who need that tech will already be with AMD, so why would they switch?

Intel needs to play the one-upmanship, if they want to survive. The Itanium has been a disaster, so they would be far better off dumping it than continuing to invest in a sure-fire loser.

Right about now, I'd be pushing for a 128/64/32-bit system, that can do everything AMD's chips can do AND support some limited 128-bit operations. Solaris 10 supports a 128-bit filing system, so a 128-bit processor isn't entirely stupid. If they added 128-bit support to controllers, they'd be able to get much smoother dataflow and a much higher throughput. Nice selling points, for servers.

Multi-cores are good, if you have enough processing elements, sharable, and distributed right to maximise what you can push through. Intel are looking at 2. Why, when most multi-processor needs are alread met with 2-way through to 8-way SMP? To compete with Intel's own products, you need to start at 8-core CPUs, or there just isn't any point.

Intel's operations are sluggish, compared to AMD. In fact, they're sluggish compared to anyone. Always have been. Anyone who