Thursday, August 21, 2008

Sentimental stock picks from the dead

Palm (PALM [PALM] {PALM}]) -

They hired Apple's hardware division head, Jon Rubenstein, and gave him broad latitude. The changes he is making seem substantive and likely to yield interesting, compelling products. He is pure upside. No one who ran the division that produced the iMac and iPod could be worse than the person who greenlighted the Foleo.

Unfortunately, the software side is abysmal or Windows Mobile. Yeesh. Android or a Symbian license would solve this problem, and in light of adopting Windows Mobile the next step is clear. Palm has the expertise to build the whole cabanza, so skipping what Android covers both frees up resources, and piggybacks on the media frenzy surrounding Google's project. Symbian has been around and ignored so long I doubt it is in the running.

Palm is like Psion, not HTC. They are (duh) in the palm computing market, not the tele-phoney space. The iPhone and BlackBerry have made that market huge, but incompletely serviced. An iPhone cannot be had with a keyboard, SD slot, etc. etc., and the BlackBerry is uncomfortable in its consumer high heels. Palm has plenty of people to sell a range of Android-Palms to.

If the Treo 900 has the game-changing design and feel of, say, the Palm V, and is also the best Android phone, then Palm goes from Foleolithic to iRIMtastic and the stock zooms. The company is off 90% from its high, and the media and market surrounding the type of device they could create is frothing at the mouth to hype something else that's actually good.

Transmeta (TMTA) -

As far as I understand, the technology that Transmeta developed (look up Crusoe or Transmeta in your local library) would benefit massively from a multi-core re-implementation, something that would have not been possible when the Crusoe was kicking around. 

Indeed, the load on such a chip (taking in an instruction, transforming it, running it, transforming it, and sending it back) necessarily benefits from parallelism to a far greater extent than a design with fewer stages. Split 'em up yo!

Of course, if a stage is too costly (in the computational sense), or the technology somehow requires context that must come from a whole-state consideration, and not separated cores, my argument goes out the window. As I said, all of this is "as far as I understand." 

NVidia licensed Transmeta's tech a few weeks back, and is now rumored to be "entering the x86 space." NV has no ability to sell x86 chips without AMD and Intel saying it's OK (according to the Register), so something has to give. If Transmeta is getting another shot at the big time, than it is undervalued. 

If you look very long term (10 years plus), it is hard to imagine a company that employed Linus Torvalds among many luminaries, and actually executed on a rather audacious plan from a standing start, would not have an impressive patent portfolio. Their efforts are ahead of the trajectory of the industry, and they likely left land mines all over the place. When Intel and AMD march in, expect big-money licensing, settlements, or rulings.

No comments: