In NAND, you need scale to succeed, to amortize the tremendous cost of new node TD across. We tried so very hard, built out our factories, got a joint venture, even made our own Optane technology, and it was all to no avail, we could not scale up as big as our Asian competitor and had to bow out painfully.
Logic fabs are going to repeat history.
@b0 NAND is not a predictor for the fabs because they make high margin products and not the lower margin memory products.
To make money on memory (over more than 1 product life cycle) takes massive volume, and the R&D costs are substantial over multiple life cycles, which is why so few companies last in that business.
Staying leading edge (in compute) even while producing the highest gross margin products like CPU also takes a lot of volume, and x86 has lost enough market share that the ability to fund R&D is becoming questionable. AMD has take some meaningful share but x86 is being edged out in client and server by ARM, and with datacenters so focused on AI that has greatly slowed down the server refresh rate in recent years.
So memory foundry does not translate to compute foundry economics, but even in compute Intel is at a point where the scale needed (while no longer dominating the market) requires at least a few meaningful external customers.
The EUV-equipped fabs are large enough now that each one is capable of being profitable, so Intel no longer needs the older smaller fabs and should sell them. That will reduce the capital needs to keep them running and clean up the balance sheet as well.
Intel would be just fine with just the EUV fabs in OR, AZ and IR, and use the empty shells for expansion if external volumes someday become a thing.
@OP Every product has to scale to a degree, for the cost of developing that product to be paid and hopefully a profit to be made.
The scale needed is determined by the net margin and volumes. Some might say gross margin but the difference usually ends up in profits.
LBT called glass substrate an 'R&D Trap' and although Intel is at the leading edge on packaging, the issue is that without the kind of massive external customer volumes normally only seen in the packaging companies themselves, Intel can not make a profit on packaging.
Optane and ALL other memory products are very similar to the issue in packaging, and in fact have lower net margins.
As a tech company, Intel from time to time has developed products which are leading edge and able to change the market for that product. The issue is that the company usually is not able to generate enough sales of that product, because it is not even adjacent to the core markets that Intel serves.
Other tech companies have faced this same issue and the answer is usually to develop the product and try to sell it to another company where it is a better fit for their core portfolio, or where they are otherwise better able to scale that product.
This is the issue that LBT is working on when he says he wants to focus on core products. ELT members are exiting as they find that the funding for their tangential efforts is being cut. An analysis of these products likely finds that few of them have any chance of being profitable and paying back the development cost.
People can call it grifting or arrogance or whatever, but really it is the cleaning up of a lot of wasted efforts, that are not a good fit for Intel or its customers. Same could be said for Mobileye, Real Sense and other efforts that have been successful. They need to be run as independent companies, or owned by a company where that is a better fit.
ELT are compensated based on the size of their org, and Intel has allowed a lot of bad product fits to be built over the years as various ELT members tried to build their empire. CEOs like LBT then come in and cut that stuff off.
@OP on NAND Intel failed to fully commit to the buisiness and instead wanted to only get there feet wet. They really didn't try that hard and didn't have the stomach to dive head in to the entire NAND buisiness. That was the fatal flaw. I blame this failure on Intel executives. Don't you find it curious why Micron was able to make a ton of money on the same exact wafers? Should make you sick to your stomach.
Optane is a different story. They failed to listen to their partner. They should have exited Optane years before but arrogance and the thought they could force people to adopt this kept them going.
Just agree to work 12 hours a day 7 days a week for a nickel an hour so Intel can reduce costs and be competitive with foreign slave labor.
@a4: it’s actual history that happened
nonsense of course
China is the real superpower.