Re: more Intel bad news

On 29/06/2025 2:58 am, john larkin wrote:

On 28 Jun 2025 17:12:27 +0100 (BST), Theo
<theom+news@chiark.greenend.org.uk> wrote:
 

Bill Sloman <bill.sloman@ieee.org> wrote:

On 29/06/2025 1:01 am, john larkin wrote:
Bean-counting mostly works - more often in a relatively mature
technology. Your list of things that they "messed up" is a bit strange -
I was in England when "ARM" was being invented.
>
https://en.wikipedia.org/wiki/ARM_architecture_family
>
Intel never had a chance to mess it up.

>
They did:
https://en.wikipedia.org/wiki/XScale
>
It came from DEC, of course. So they both messed it up, in sequence.
>

Intel was always a customer for EUV lithography, and quite how Philips
ended up doing so well with it (ASML is a Philips spin-off) is a bit of
a mystery. Philips had an electron-beam microfabricator business, and
messed it up badly enough that they sold it off to Cambridge
Instruments. The machine was fine, but the customer service wasn't good
enough.

>
Philips was good at optics (lighting, medical imaging, LCDs) and the
semiconductor research divisions were really good. Another company to add to
the list of messes.
>

When your compensation is mostly the value of your stock options,
complicated things like transistors are an annoyance.

>
Not when they are your entire business.
>

Exactly.
>
Theo

 Intel kicked in a billion or so to help ASML buy Cymer for the
tin-droplet EUV source. Then they failed to invest in the machines for
their own fabs.
 Meanwhile Intel kept buying back their own shares. That's like trying
to grow by eating your own legs; it doesn't scale. [1]

But it's great for people who owned a lot of Intel shares (or stock options).

Intel lost a heap on Altera too, ballpark 8 billion. They lost our
business too; now we use mostly Efinix and some Xilinx SOCs.

I never did like the Altera programmable parts. I was much more taken with the Philips Cool-Runner parts that they sold to Xilinix, not that that I ever managed to design them into anything that sold.

The pattern seems to be that Intel chases fads and always messes them
up. That will become business school courses some day.

Everybody in the semi-conductor business chases fads. Somebody always ends up doing the fad better than everybody else, so most of the competition has always messed up. Intel is still in business - unlike DEC - so they haven't messed up completely.

[1] Hey, one could raise some venture capital, go public, and do
nothing but buy back shares. Sorta like bitcoin.

Three merchants marooned on a island who got immensely rich by trading hats is the classical example.
--
Bill Sloman, Sydney