IBM Demonstrates a Nanosheet Transistor that Loves 77 Kelvin—Boiling Point of Nitrogen

[Operandi]

They sold the fabs, because they didn't have enough volume to justify spending on fabs. In fact, they had to pay Global Foundries $1.5 billion in 2014 to take them off their books. IBM didn't really play as a foundry even though there had been times in the past when it produced other companies' chips like Nvidia's GeForce FX in 2003.

Kinda figured they weren't able to keep them fully utilized on their own but you think a company like IBM with as much capital they have and being so close to research side of things they could have been a player in the foundry game had they played their cards differently. In hindsight seems like poor decisions given how reliant everyone is on TSMC and the nation security interest implications.

However IBM has remained very active in basic research. They demonstrated a 2nm nanosheeted prototype chip in May 2021, for example. I guess they (and Imec of Belgium) make a lot of money selling distant future tech to other parties.

Thats the confusing part to me, how exactly all that works since the research and development is anything but cheap and you would think the long term play would be to leverage the R&D advantage yourself (aka stay in the fab game).

 

[Panther_Seraphin]

Thats the confusing part to me, how exactly all that works since the research and development is anything but cheap and you would think the long term play would be to leverage the R&D advantage yourself (aka stay in the fab game).

Problem was that even though you could develop this, the capital outlay to then take advantage of that was becoming excessive. You only have to look at the costs of some of the new plants trying to get off the ground now to realise that you are talking billions if not tens of billions over the lifespan of a plant to keep it "up to date" on top of the now 10/20/30 billion to just get it built initially and without getting a sizeable contract from say Apple/Nvidia at the time there was very little else out there to warrant that scale of investment.

When people were really started to look around for different foundaries was around 2014/15.......Which meant you needed to actually break ground and commit to the build of a new foundary in 08/09 and we all know what happened in 08/09 with the economy. So you could imagine going up and proposing a multi billion dollar outlay at a time when the economy was in a complete meltdown with a prospective start of ROI being 5+ years down the line IF and only IF you managed to snag a good deal off Nvidia/Apple at that time.

However TSMC were completing a brand new "Gigafab" completing in 2009 which would stand them in great place for the recovery of the economy and the then stagnation of AMD/Global Foundaries fabs. AMD however was spinning off their Fabs at the same time and offloading them with over 1.5 billion dollars worth of debt while also agreeing to deal with those fabs for quite a few years after the spin off while having a product that wasnt anywhere near the equal of their Rival Intel.

EDIT/Contiunation:

So at this exact time you had Intel which still had a lead on foundry tech with 14nm being publically released around this time with the lower power Core M-5Y10 in late 2014.

Samsung had small chip in 14nm capability in early 2015 with larger chips coming in 2016 with Low end 1xxx series nVidia.

TSMC was primarily doing 28/22nm production with 16nm coming in 2015 with the Apple A9 in late 2015

Ryzen had barely hit tapeout stages and was still 2+ years away from public release.
IBM was on 22nm with their shift to 14nm coinciding with AMD at around 2017 with POWER9. This goes to show how behind Global Foundries had become after the 45nm parity with TSMC.