IBM will have - as the seller - a hard time arguing they weren't aware of the problems with delivering the 10 nm process chips on time or even at all on a schedule that was still commercially interesting to GF. For the 7 nm process the same holds to a large extent: if the company could not deliver on its 10 nm promises any bets involving the 7 nm process were instant long shots.
IBM knew very well what the state of the division was when it got rid of it, it will now have to prove in court that it could not have known that this was a real possibility.
Global Foundaries pulling the plug on 7nm was by most accounts a financial decision, not a technical one. Their 7nm process was essentially ready, but the owners of Global Foundary (Mubadala Investment Company) haven't seen a profit for the past decade, and decided to stick to older and less expensive process nodes rather than spending billions retooling their factories and trying to keep up with TSMC and Samsung.
> ... decided to stick to older and less expensive process nodes rather than spending billions retooling their factories and trying to keep up with TSMC and Samsung.
I would love to see their rationale for this.
> In March 2021, Mubadala Investment Company and its joint venture Abu Dhabi Catalyst Partners announced the purchase of Telegram convertible bonds worth $150 million.
> Back in 2013 and 2014, IBM was in one of the divesting moods it periodically gets into. It sold off its System x X86 server business to Lenovo and “sold” the Microelectronics division to GlobalFoundries, which itself was spun out of AMD in October 2008
Wow. Time for popcorn. It's itanic all over. On the other hand it is sad to see Power go.
I call it "we can't make money on chips, so will sue our foundry for that," and in part to save their filet parts from shareholders (who themselves were partially responsible for scrouging on capex.)
It's sort of wild that they just split out all their services and consulting as well, so I'm not sure what IBM does now, aside from Watson, an also-ran cloud hosting business, and the old big-iron.
IBM just split out “we will run your servers for you,” not consulting or other services. You’re forgetting about a large security services business unit, a large research org, and several other parts.
I think it's true that timing is related to potential IPO, but if the claim has merit, impossible to say.
IBM is suing for compensatory and consequential damages,restitution, equitable relief and punitive damages arising from the fraud and intentional breaches of contract committed by GF. This was no mere supply arrangement. GF did not commit to long-term strategic and financial commitment to the development of High Performance chips. IBM provided money and technology to develop those chips. ...
IBM reserved all its rights with respect to GlobalFoundries’ breaches with 10nm when GF moved to 7nm. It turns out that GlobalFoundries was never committed to developing High Performance semiconductors for the long term. GlobalFoundries admitted that it "never planned to be a leading producer of 7-nm chips in terms of volume".
GF says those are outlandish demands and claims have no merit under the plain language of the Parties’ Agreements. IBM tries to fish because there is potential IPO.
I was surprised this did not get more traction when was initially posted, but glad it taking some visibility.
Its likely to have a major impact for both companies.
IBM is basically accusing GlobalFoundries of taking its technology plus some billions and doing nothing on purpose.
So it looks like, knowing IBM Legal, they think they have a strong case. Or maybe IBM messed up this one so much this,
this is the real reason behind the change of CEO...
Well at least it can be said that, had IBM not gotten out of the foundry business, they would be minting money right now. If then-CEO of IBM had successfully pivoted the company to all things Watson, which seemed to be the plan, it would still have seemed like a good idea. It does not seem to have been a good idea.
I find it interesting the article seems surprised AMD said nothing. AMD for a long time was a major owner of shares of global foundries. GloFo Spent a while buying the stake that AMD held. AMD is still using GloFo for it's IO dies as they don't need the densest nodes. Heck AMD just this May announced it's gonna buy an other 1.6 billion dollars worth of wafers from GloFlo.
>I find it interesting the article seems surprised AMD said nothing.
AMD doesn't use 10nm or 7nm from GF. So it doesn't bother AMD at all. The continuation of IOD was mostly because of price. Basically GF has no leverage on their 14nm and AMD are their only customer. GF either find a customer to use those capacity or turn it to something else other customer would need like their FD-SOI node.
I do wish AMD could move IOD away from GF though. They need some node optimisation for power efficiency.
I would have to find it again but a while ago I was looking at thermal images of AMDs chipslets and the IO die is pretty cool compared to the chiplets for the cores. There probably is some power savings for using a newer node for the IO dies, but it would minor compared to the cores.
I am actually surprised it took this long for IBM to take action as it was the first thing when I saw GF announcing to stop developing 7nm. ( AMD was well prepared for Dual Fab at the time )
Their whole Z / POWER roadmap and release schedule was quite literally destroyed by GF. It now suffer a multi year delay, think Intel 14nm fiasco. Except IBM isn't Intel where they could squeeze and milk 14nm and their market. And funny enough IBM didn't went with TSMC and decided to use Samsung thinking they would have an easier time to port to their node due to similarities with GF. It turns out they might as well have gone with TSMC.
I think there are more to blame IBM for its management than GF pulling the plug.
19 comments
[ 4.7 ms ] story [ 49.1 ms ] threadIBM knew very well what the state of the division was when it got rid of it, it will now have to prove in court that it could not have known that this was a real possibility.
https://www.anandtech.com/show/13277/globalfoundries-stops-a...
I would love to see their rationale for this.
> In March 2021, Mubadala Investment Company and its joint venture Abu Dhabi Catalyst Partners announced the purchase of Telegram convertible bonds worth $150 million.
Wow looks like they like to waste a lot of money.
Wow. Time for popcorn. It's itanic all over. On the other hand it is sad to see Power go.
"AWS on track to be bigger than IBM by Christmas" https://www.theregister.com/2021/04/30/amazon_q1_2021/
A partnership with IBM was a major headline of the Intel foundry partnership program a few weeks ago.
I think it's true that timing is related to potential IPO, but if the claim has merit, impossible to say.
IBM is suing for compensatory and consequential damages,restitution, equitable relief and punitive damages arising from the fraud and intentional breaches of contract committed by GF. This was no mere supply arrangement. GF did not commit to long-term strategic and financial commitment to the development of High Performance chips. IBM provided money and technology to develop those chips. ... IBM reserved all its rights with respect to GlobalFoundries’ breaches with 10nm when GF moved to 7nm. It turns out that GlobalFoundries was never committed to developing High Performance semiconductors for the long term. GlobalFoundries admitted that it "never planned to be a leading producer of 7-nm chips in terms of volume".
GF says those are outlandish demands and claims have no merit under the plain language of the Parties’ Agreements. IBM tries to fish because there is potential IPO.
Its likely to have a major impact for both companies. IBM is basically accusing GlobalFoundries of taking its technology plus some billions and doing nothing on purpose.
So it looks like, knowing IBM Legal, they think they have a strong case. Or maybe IBM messed up this one so much this, this is the real reason behind the change of CEO...
The two complaints - IBM Case: https://3s81si1s5ygj3mzby34dq6qf-wpengine.netdna-ssl.com/wp-...
GlobalFoundries response: https://iapps.courts.state.ny.us/fbem/DocumentDisplayServlet...
AMD doesn't use 10nm or 7nm from GF. So it doesn't bother AMD at all. The continuation of IOD was mostly because of price. Basically GF has no leverage on their 14nm and AMD are their only customer. GF either find a customer to use those capacity or turn it to something else other customer would need like their FD-SOI node.
I do wish AMD could move IOD away from GF though. They need some node optimisation for power efficiency.
Their whole Z / POWER roadmap and release schedule was quite literally destroyed by GF. It now suffer a multi year delay, think Intel 14nm fiasco. Except IBM isn't Intel where they could squeeze and milk 14nm and their market. And funny enough IBM didn't went with TSMC and decided to use Samsung thinking they would have an easier time to port to their node due to similarities with GF. It turns out they might as well have gone with TSMC.
I think there are more to blame IBM for its management than GF pulling the plug.