Posts by Platinum blond(e) 25 publicly visible posts • joined 10 Dec 2020
$500M for an exaFLOPS+ super in the present AI-driven sales environment?
Two of the GPU vendors are selling everything they can make for high margins (some higher than others admittedly). The third has offerings that don't seem as competitive.
Assuming GPUs can be sourced for this project, what margin will be left for the system builder? Will that margin be enough to attract bids?
The thing to remember about AVGO is, their playbook has been the same through each and every acquisition. Target an entity who's dominant in a slowing market. Keep the groups with biggest profits and discard the rest. Jack up the prices while continually shaving heads to meet the AVGO margin requirements. In this way, squeeze every drop out of the acquisition as it slowly dies on the vine (looses customers and talent/employees). Line up the next target as soon as the last acquisition closes. Lather, rinse, repeat to generate buzz and keep fueling the margins and prop up the stock price.
For those employees retained, AVGO dangles lots of RSUs. But the remaining heads will have to work harder and harder to cash those in. That works for many but not for all. For the other employees, AVGO provides reasons to consider moving on. Like currently for the VMW folks, get them upset about RTO.
AVGO have done this for every acquisition. Why fix what ain't broke? And WallSt likes good margins, they dont look too closely how it gets done as long as it gets done.
Source: I worked for an outfit before it was pulled into this meat grinder. Heard from many what it was like to be cast off. And retained.
All cash, 100% up front deal for her, is the only thing that makes sense to me. $100M for up to 1 year of service. Nonrefundable. Maybe with a free ride to the ISS thrown in.
If she's as good as her reputation, that's the sort of deal she got.
Because she will be made a scapegoat and she knows it.
Yes as noted Stora Enso has oodles of lignin from their pulp and paper making processes. So this is finding a use for a byproduct. BTW they have many mills and pulp plantations around the world. If this works out there could be many others to source the lignin from, too.
Source: bro, relatives and many friends work(ed) at Stora Enso plants in central Wisconsin.
Fair point! May we explore this a moment more?
If TSMC chose to build fabs in Europe, the Community would be even more interested in Taiwan's security at least until those fabs and required supply chains are producing products there. This could be 3 to 5 years, more if the supply chains depend on TSMC's expertise. So they could have increased protection (if you can call it that) and some subsidies to help.
If Taiwan were seriously concerned about an imminent threat, it seems to me they would work a deal. They aren't making the short term play.
This could be an early bluff while all await the river card. But I still think they're playing a longer game.
TSMC's comments suggest they have the long view firmly in their sights. No supply chains in Europe which meet their needs *, and no big consumers (e.g., contract manufacturing houses) of the parts they ship. So even if they set up shop it would take many billions of euros more to keep from shipping the products halfway around the word to make finished goods. And then ship them back to end consumers. So with higher European labor costs to make things, and shipping stuff to and fro, this model is unsustainable once the subsidies run out.
Intel OTOH is selling the Field of Dreams story, both to Europe and the USA. And quite likely to themselves.
*: with the exception of ASML of course.
Reposted from yesterday's note on the topic:
Avago's playbook:grab and squeeze
Broadcom has now a pretty long track record in both SW and HW acquisitions. Always plays the same: reduce staff, pressure clients with exclusive contracts, raise prices; all to hit their target margins. Repeat with next acquisition as the prior carcass is squeezed dry.
This is good for shareholders and senior execs obviously. Bad for everyone else. Even the remaining staff: I cant imagine those RSUs can somehow pay for a decent work/life balance.
VMWare's many customers should/will raise strong concerns and the ruling bodies worldwide would likely give such an acquisition the electron microscope treatment as a result.
This requires a new popcorn machine.
ETA: the squeezing doesn't stop when the deal is done. It is constant and continuous. Staff can expect multiple waves of reductions. Best of luck to all.
Broadcom has now a pretty long track record in both SW and HW acquisitions. Always plays the same: reduce staff, pressure clients with exclusive contracts, raise prices; all to hit their target margins. Repeat with next acquisition as the prior carcass is squeezed dry.
This is good for shareholders and senior execs obviously. Bad for everyone else. Even the remaining staff: I cant imagine those RSUs can somehow pay for a decent work/life balance.
VMWare's many customers should/will raise strong concerns and the ruling bodies worldwide would likely give such an acquisition the electron microscope treatment as a result.
This requires a new popcorn machine.
Certainly, Morris Chang may be viewed as having vested interests to protect with these remarks. Even I believe that.
But that doesnt mean his assertions are wrong. If anyone in the world knows how difficult it is to staff, fit out, and run leading edge fabs, it's him. I'm pretty sure he has hundreds if not thousands of tales he could tell of hard won learning at the school of hard knocks. Stuff rarely mentioned in academia (though perhaps hinted at for awhile in Harvard Business' Ad Prac back in the day). Stuff that is considered 'secret sauce' and will never be described in patents or papers, for example.
Similarly, much can be learned in schools to start a young Engineer on the desired path. And there will be some brilliant minds devising new methods. But the physical details need a lot of hands-on work to get the part right every time and thus robust. This type of expertise is expensive to gain, along any metric one wishes to measure.
From the 60s thru the 00s, American firms have shown they couldn't find the value of these practical lessons, and divested themselves of nearly all the stateside fabs they had. Since then American firms have not stood up new fabs unless there are government funds and government goading to do so. The firms still do not see this value if it's standing by itself.
I predict (again) that American fabs will wither away once the government funds dry up. And the funds will dry up because Congress critters dont have much longer vision than the quarterly-profit driven champions of industry.
When the Feds bailed out the Big Three automakers, they took stock IIRC. And also IIRC the stock was sold shortly after and the funds were repaid in this manner. It worked then, why not now?
And as far as long term, I predict that US businesses will once again allow the manufacturing to move to wherever is cheapest just like they did since the 50s. Yes kids, going back to the middle of last century. Same as it ever has been in industry, of course.
Lather, rinse, repeat.
My take is that selling big iron to classic enterprise customers is usually done in person, with plenty of wining and dining to help grease the skids/seal the deal. This might help explain downtrend in that class of machine. Meanwhile, cloud builders are already marching to their own Cadence, and enterprise-class schmoozing has much less effect. Will the gallery please correct me if I'm wrong?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
