Transmeta Lays off 40% of its Workers

aftk2 writes "According to news.com, chip maker Transmeta - current home of Linux creator Linus Torvalds, has canned 40% (200 people) of its work force, and has shifted its goals toward obtaining profitability in 2003. No word on whether there were any penguins seen leaving the building."

Comments

There's a shock.

[Skyshadow]

Despite the "celebrity" factor Linus brought to the place, their product just never panned out. It was a good idea, and hopefully some larger company will buy up their proprietary technology, but I don't see how Transmeta on their own ever could have made a run at capitalizing the chips in an already severely swamped market -- the barriers to entry were just too high.

Still, having been laid off twice last year, I wish all the former Transmetites the best. I hope Linus is able to find an interesting job after Transmeta folds -- otherwise, my company could use a good code jockey...

Increased revenue by 82% from Q1

[mocm]

Why are all those articles so negative. Right now it looks like they overcame the production difficulties and are moving ahead with new costumers that are actually building notebooks with Crusoe CPUs. The U1 from Sony is the hihest selling notebook in Japan already and the Fujitsu P series is also selling well in the USA. With the upcoming HP tablet and the OQO, I would say they aren't doing so badly.

IBM

[cfadam]

I wonder how long it'll be before IBM snaps Linus up. What better way to get support from the Linux masses than to snap up its creator?
- A

How long have they got?

[Anne_Nonymous]

$180mm of cash and near-cash, and $25mm of burn this quarter (+/- depending how you count), gives them a life expectancy of two years. I suspect we'll see a catalyst one way or the other before then though.

Re:Overflow of workers go where?

[Anonymous Coward]

I'm a software guy and will look for work in the software industry and try to stay clear of hardware companies for the rest of my life. I'm sure most other software people will follow this same rule.

The hardware business is just plain harder than the software business because of the limited lifespan of practically all hardware designs. Software on the other hand can easily live forever.

Re:Shifted its goals?

[mooman]

A lot of smaller companies focus more on obtaining market share. Plus remember all the dot-com Superbowl commercials? I think they are pretty much intended for branding and establishing market identity, sometimes at great cost.

So technically these companies may have to make a conscious effort to focus on profits rather than marketing and growth...

Why to an outsider this seems obvious

[interstellar_donkey]

Let me predicate this by saying I have used or seen a Transmeta product. And in that, I think is the problem.

Going back a few years, I remember the buzz surrounding Transmeta. 'There is this company that's developing something ground-breaking... and Linus Torvolds is working there!' If memory serves me, investors--any investor would give their eye teeth to just be able to put money into something 'groundbreaking' being worked on by Torvolds.

Then we finally saw what it was. A chip. Oh. . . Well, what makes the chip so special? It uses less power. Oh. . . Does that change anything for us? Sure, your laptop batteries will last a little longer, and if you run a server your electric bills might be a little lower. Oh.

Of course, I'm not a programmer or do work on hardware, but for me this was a letdown after so much hype second only to learning 'It' was nothing more then scooter that was hard to tip over.

That was two years ago, and despite the fact that there is some benefit to the otherwise ho-hum technology, where is it? I buy a lot of computers, and I don't even know where to buy a Transmeta equipped machine (then again, I've never really looked, and have never been given a good reason to look).

So, again, this seems obvious. A company pours a big chunk of change into a product that never sees the light of day on a mainstream store shelf... a product that I quickly forget about and am only reminded from time to time on Slashdot stories.

I suppose, scanning the posts, that there are a handful of gee-wiz products out there (albeit not in the United States) with a Caruso chip, but I just don't see them, or see any reasoning to spend the extra money on them.

And so Transmeta starts laying off people. It just seems to be the next logical choice.

Re:F***

[rjamestaylor]

Happened to me in March - May. Small company, big project, bad management. Thing was we weren't alerted to the fact that payroll would be "late" (sometimes 2 weeks) until the very end of day payday. The last straw was finding out my family's health insurance, for which I paid $500/mo after taxes, had been canceled April 1. I found this out May 5, or so.

What a miserable experience. But we came out of it very well. My co-worker is working at a stable company for more pay (personal connection got the interview) and I'm working for my former client for much more pay, benefits, and equity (not options; equity in a profitable company). While I was sweating out the collapse of the old company I had very little hope going forward. Looking back, it was a great opportunity.

Re:Not surprising...

[Bastian]

I know nothing about the feasibility issues involved, but it seems like that kind of idea could be extremely useful. I'm thinking a machine with a CPU capable of running multiple instruction sets simultaneously coupled with a VM-type operating system that allows you to bootstrap virtual machines of various architectures.

Unfortunately, their push was toward the mobile market, so they appear to have put more effort into power consumption than they did performance, and I dont think they even tried to get a Crusoe processor running multiple instruction sets simultaneously yet, so anything along those lines that we would see anytime soon would probably not be better than just buying two different machines of different architectures, and I doubt many companies percieve much of a need to have a machine capable of handling 3, 4, or 5 instruction sets, which is probably where the cost of purchasing such a machine would start to be justified. . .

There's also the possibility of using it as testing machines for software being developed for CPU architectures that haven't had fully functional prototypes come off the line yet, but that's wouldn't provide nearly enough business to keep a company going. . .

Code Morphing

[PhotoGuy]

Code morphing with only one target implemented (Intel), offers nothing above buying an actual Intel chip. And as mentioned, the power savings advantage is something others have jumped on very quickly, so there's little to differentiate it. (Although laptops using TransMeta still seem to have battery life ratings beyond the competition.)

Have they ever stated any intention to implement another target for the code morphing? Being able to have the same computer be a Mac or a PC (or a Sparc) would be far more compelling, and is what I had hoped the original story was all about. Is that just not lucrative? Do they not have the resources to pull it off? Was the TransMeta designed too much with Intel in mind, so that a PPC or Sparc emulation isn't possible???

It's biggest advantage seems to have gone completely by the wayside.

Do not count these guys out

[John Murdoch]

Hi All!

I think it would be a big mistake to count Transmeta out any time soon. I say that not because I'm a penguin-loving Linus-worshipper. To the contrary, I primarily use Microsoft development tools, and when I'm feeling giddy about Unix I use FreeBSD. The only Linux boxes around here belong to paying customers.

So why not write off Transmeta?
Simply put, they're working their way into the product channel. Transmeta does have a very low-power chip--and that Transmeta technology is at the core of an emerging form of hardware: the smarter embedded system. Don't think "desktop replacement"--think "death to the PLC."

What's a PLC?
Programmable Logic Controllers are tiny CPUs that appear in all sorts of specialty uses: controllers, valves, automated-just-about-anything. They're cheap, they're generally very reliable--and they have zero memory, have very limited functionality, and require programmers who demand significant coin. When you try to add a feature to an embedded application you will typically a response on the order of "that will take--at least--200 bytes of memory. And we only have 68 bytes left. So what feature do you want to drop to do this?"

Coming soon, to a factory floor near you...
The Palm OS, WinCE, and the Transmeta chips are going to change all that. Handhelds and rugged semi-embedded handhelds are appearing in larger numbers--with gigabytes of flash storage, and 128 MB of RAM. Skip counting bytes--add all the features you want. Connectivity? They have 802.11 already embedded, along with USB, serial ports, etc., etc., etc. Some of the vendors I've browsed recently include InfoCater and SyntegraTech; they're both distributors for Tablets, WebPads, and handhelds that run with WinCE or Midori Linux. Very, very cool stuff.

Laying off 20% of your staff may be painful--but it is not the same thing as shutting the doors. For example, note that VA Software is still around....

Still a Perfect Match for Apple?

[EvilSuggestions]

Back when Apple switched the Mac from 68k to PPC, they did better than expected keeping the old 68k code going, but they did so by using slow, clunky software emulators. Imagine if they could have just had one chip than ran both instructions sets!
Since Jobs has finally reached the point where he doesn't immediately shoot down the idea of switching to x86 (In a recent interview: "We like having options..."), maybe they should check into this as a way of keeping PPC going.
It would solve many of their potential spin problems:

• Not emulating PPC on x86. After hyping the superiority of PPC over x86 for so long, they'ld be insane to use an x86 based architecture to do the emulation that would absolutely need to support.
• The "x86" Mac would not just be a pretty PC clone. Running MacOSx86 on Apple hardware would have a tangible advantage over running it on generic PC hardware: the ability to run all the current PPC based Mac software at reasonable speeds. Not a big deal for current x86-ers who just wanna dump Windows, but it would be crucial for their current customers.
• On x86 hardware, but not Intel hardware. Given some historic biases, this might be a bigger deal than it should be. Suggesting AMD instead doesn't seem to help.
• Lots of options for their appliance/"digital hub" ideas Imagine if they could use the same CPU in everything from a multi-cpu PowerMac Server down to a settop box or handheld?

Unfortunately, my curmudgeon side says this all makes too much sense to ever become reality.