You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Bias can be in the programming and can be in the training data set. The problem is exacerbated in Gen AI as it uses vastly larger and diffentiated datasets than those used in machine learning, and so are even more difficult to sceen and control. That is why they put some hard rules on the...
There was actually only one 157nm tool I remember delivered: the ASML/SVG/PE Micrascan to imec. In the meantime ASML was working on the immersion 193 prototype and, as soon as the first images printed with it came out, 157nm was dead. Intel invested a lot in programs in 157nm over the years and...
157nm, good all times. Intel declaring at SPIE the needs of tons of CaF2 crystals for the hundreds of steppers they were going to buy just 3 weeks before ditching everything. Schott ever so glad of keeping with the plan of waiting until the first full demonstration of the stepper before boosting...
Competing agains NVidia on AI now is like competing with Xilinx in the past for FPGA. So much is tied in the software that either you are fully compatible or you do not get market share for your hardware. So you are always at a disadvantage.
A patent is worth only how much you are ready to spend to enforce it.
Is AMAT going to sue any of their customers using it without paying licenses ? I doubt it.
And ASML is not going to loose any business over this I think, as it applies only to a restricted set of patterns.
Seen from this side of the pond the TSMC fabs in US and Japan are quite different and not comparable proposition. In US the claim is for an advanced fab fully TSMC for 5nm and below with a tinge of "national interest". In Japan it is a joint venture with local co-investors (mostly customers) for...
When you compare you have to take into account that for IDMs you have both process and product development spending, fabless only product and foundries only process.
Gallium is present in bauxite at 10g to 150g per metric ton. By comparison Al is about 250kg per metric ton of bauxite. The real problem is then the extraction process from the byproducts of Al mining. Very dirty process with a relatively low yield. Some are looking into it as it can be a way...
A well run fab has no bottlenecks in front of litho, it is build to have uniform throughput and you do not buy one tool more or less than necessary to maximise overall use. Underinvesting and creating bottlnecks is very quickly anti-economic and you do not have more or less depo/etch/clean...
STM is very much Franco-Italian. Minority blocking stakes are still in equal measure with the two Governments. And, depending on the market segment one focus in, it can be perceived more French or more Italian.
A lot of "lost in translation" issues with the piece that makes a big confusion of...
300mm wafers move from tool to tool in standardised FOUPs (Front Opening Unified Pod) usually of 25 wafers (may be 13).
Production is really a batch process even if some tools are single wafer processing. So indeed you have time spent to serialise the process and then parallelise the transport...
AMD to GF all over again, but without Mubadala money ?
Who is going to pay for IFS while it gets customers, Intel? If so why spin off and why customers should believe they will be treated on an equal footing.
If not, where is the money to keep going and stay in the development race until it...
Details are not out yet. But there is a major difference between the two. In the US you have a central government taking the decisions and handing out the money. In EU no money from the Commission for the production initiatives (only R&D) and each member state will be able to support the...
Seen only now. There are multiple considerations for choosing thickness of water in an immersion stepper and not all of them purely due to the lenses. Just to name two of them, you need the flow of the water to be laminar otherwise you have distortions due to turbulence and you need the water...
It started as AMD and become GF later, but all the bulk of the investments (AMD and GF eras) were during the period of relaxation of EU State Aid limitations for former East Germany. Now Germany cannot give out the same conditions as that transition period is over. Any subsidy now need...
Scotten, one lesson I learned with monolithic 3D is that a good technological solution doesn't get picked up if it doesn't cover all the spectrum of products. If can only do SRAM and not general purpose logic is a no go. One of the option we looked into with monolithic 3D was to put all the SRAM...
Daniel, you and other reporters should spend a bit of time really looking into the announced EU Chip Act, start separating facts from fiction and do not take press releases as a reliable source of information. Everybody talks about 45B€ without reading the relevant, public, official papers and...
In Europe only Intel, ST and GF have fabs for nodes below 90nm. All other 300mm fabs (IFX and Bosch) are for equal or larger geometries devices. It is the reason why IFX, Bosch and NXP are rumored to be those trying to get TSMC here (no A-grade source there either ....).
Would like to be the fly...
Agreed that is doable, my point is that requires a lot of engineering resources to do it, as there are a lot of challenges that make it a risky development. And this for an even smaller market that existed before EUV.
I have not really looked into it since the demise of Mapper but I am not aware...
One of the issue with multibeams technology is that wafer throughput is also linked to the bandwidth of the transmission of data to the writing head. When you workout the numbers you see why is used for masks but not for wafers. Also, as one of my more experienced colleague told me many years...