Bart Yee on TSMC can't make enough 3nm wafers to meet AI demand - 'Gregg and everyone, yes, there’s no doubt Apple will get first priority on 2nm production since they’ve essentially prepaid for probably 250 million A-20 chips, meaning ordering 310-330+M chips, and say 20M M6 chips, meaning ordering up to 27M at supposed 75-80% yields. But that doesn’t guarantee anything in actual production yield terms for the future because 2nm has yet to enter full production, but initial 2nm yields look promising. Here’s a look at current 3nm TSMC production from Gemini: “TSMC’s 3nm yields vary significantly by die size, ranging from 85–92% for small dies, 60–75% for medium dies, and 45–65% for large GPU-class dies. These highly advanced processes have achieved a strong, mature, and stable mass-production rate. TSMC’s 3nm family has several iterations, with specific parameters and yields differing by node: Yields by Die Size Small Dies ( 600 mm²): 45–65% yield. 3nm Node Variants N3 (Original): The first generation used for Apple’s initial 3nm silicon. Early yields required special billing agreements for clients, but maturity has since stabilized. N3E (Enhanced): Widely adopted, offering a wider manufacturing window, better overall yields, and cost-effectiveness over the N3 node. N3P & N3X: Optimized for higher performance and compute capabilities, seeing widespread usage across major tech companies like Qualcomm, AMD, and NVIDIA. Cost and ProductionWafer Pricing: An average 300mm 3nm wafer costs roughly $19,500, though pricing ranges from $17,000 to $22,000 based on client volume commitments. Competitive Edge: Compared to competitors like Samsung—whose 3nm node yields have reportedly struggled to cross the 50% threshold—TSMC’s mature output rate of up to 90% for smaller logic chips has made it the primary destination for the industry’s highest-profile chip designs.” Regarding TSMC 2nm production progress: “TSMC successfully commenced volume production of its 2nm (N2) chips, transitioning from the FinFET architecture to Gate-All-Around (GAA) nanosheet transistors. Production is currently active at Fab 22 in Kaohsiung, with further capacity rolling out at the Fab 20 plant in Hsinchu. The ramp-up has progressed rapidly across several key areas: High Initial Yields: Early prototype and pilot yields have been exceptionally strong, reportedly surging to 70%—significantly accelerating the path to full commercial availability. Enhanced Performance: Compared to the 3nm (N3E) process, N2 technology delivers a 10%-15% speed increase at the same power level, or a 25%-30% reduction in power at the same speed, along with up to a 20% increase in transistor density. Capacity Scaling: The company is currently building or ramping up 9 fab phases simultaneously across Taiwan, the US, Japan, and Germany. TSMC expects N2 capacity to reach up to 90,000 wafer starts per month (WSPM) and is pushing for a 70% yearly increase in this advanced capacity through 2028 to keep pace with demand. Customer Adoption: Major industry leaders are already committed. Companies like Apple (for A-series mobile processors) and NVIDIA (for HPC GPUs) are primary early adopters, while AMD’s sixth-generation Epyc (Venice) server processors are slated to be built on this technology. Premium Pricing: Wafer costs for the 2nm process are estimated to exceed $30,000, nearly doubling the cost of 4nm wafers, which will likely constrain initial adoption primarily to flagship and AI applications.”'
on Premarket: Apple is green - 'It’s a (re)start, I guess. I just hope Muskrat hasn’t pulled all the cash away from other stocks. Curious about what Space X valuation was based on.'
on An influencer from Nottingham switches from Apple to Android - 'Kirk: Actually, no. If I did start getting green bubble texts from a friend I’d consider it a distress call and I’d contact them to make sure almost everything was OK and then I’d ask about the phone.'
on Mark Gurman on Siri AI - 'David, Agree, and in same boat (most of my Apple devices are spec’d out to their max when I bought them for photography editing (for example MacBook Pro M3 Max). I will be upgrading that and my iPad Pro 3rd gen. Very much an amateur at all other computing beyond photography editing.'
on Mark Gurman on Siri AI - 'That sounds pretty good to me. In fact, if my M2 Max MacBook Pro with 32 gigs of RAM can’t do it, I might have to upgrade. I have that machine for heavy lifting photography, and it’s still great for that, but I also want these capabilities that you’ve described. And they will only get better. If I’m upgrading a machine that is already strong enough for pro-level high resolution digital photography, how many others are also going to want to upgrade just for these capabilities? Or am I just a nerd? Is this going to kick off an AI upgrade cycle?'
on Jeff Bezos' WashPo takes Apple's side against the EU - 'Here’s another logic anomaly. Why is iOS a gatekeeper with only 27% of shipments. The iPhone holds between 27% and 42% of the European smartphone market, depending on whether you measure unit shipments or active operating system usage.” Apple is penalized because they don’t ship sh!t.'
on Jeff Bezos' WashPo takes Apple's side against the EU - 'The EU’s DMA is just a tax on American Big Tech.'
on This week's Apple trading strategies (6/15-6/19/26) - '”In my view, Apple is ahead in the game.” I agree, and I can’t help but think of MSFT at the turn of the Century. First there were desktop computers, a market that was handed to MSFT by way of IBM’s anti-trust problems over its OS (mainframe) dominance. MSFT won the OS war with a piece of crap that was cheaper than anything else. Then to protect its dominance MSFT didn’t innovate, it did some nasty things that got them in big trouble, and created a crack for Apple to exploit. And exploit Apple did. But now faced with the same scenario as MSFT 30 years ago, Apple is taking a different path to exploit the new paradigm, instead of strong arming its competition. Except for the copiers and luddites like Gurman, Apple has quietly positioned itself to own the consumer (70% vs 30% enterprise) market for AI (with no competition), no matter what the Gurman’s think.'
on This week's Apple trading strategies (6/15-6/19/26) - 'Wait a second Greg – are you talking about AI or Elon?! 🙂'
on An influencer from Nottingham switches from Apple to Android - 'Gregg, that probably depends on the quality of the lens(es) you have with that camera. Lenses have improved dramatically in the last 50 years (a friend said “lens design is sufficient justification for very expensive supercomputers”)'
on TSMC can't make enough 3nm wafers to meet AI demand - 'Hi, Bart, Richard, and others. Yep. Apple has been “bankrolling” chip manufacturers for years. You don’t shaft the (very powerful) company that’s literally put you where you are; rather the opposite. BTW, that also goes for China in general. They owe Apple a lot….'
on TSMC can't make enough 3nm wafers to meet AI demand - 'TSMC is expanding its wafer fab plant in Sherman Texas, which just started producing wafers 6 months ago. At full buildout, it could produce over 1 million wafers a month. And Apple is funding some of it. From Apple: “GlobalWafers has begun production at its new $4 billion bare silicon wafer facility in Sherman, Texas. At Apple’s direction, wafers produced in Sherman will be used by Apple’s chip manufacturing partners in the U.S., including TSMC and Texas Instruments.”'
on An influencer from Nottingham switches from Apple to Android - 'Question: Does anyone here in this forum know of anyone who has switched and stayed? Still to this day, so very, very few. Cook-era 98% sat rate stands firm.'
on Mark Gurman on Siri AI - 'If Gurman is mostly correct, then it’s as it should be. AI Siri for most of us most of the time. And for the other 5% or so with rarefied needs, there is the App Store. I suspect Apple will not mind this arrangement one bit.'
on This week's Apple trading strategies (6/15-6/19/26) - 'Problem is super intelligence is going hand and hand with hallucinations and bad machine reasoning with zero compassion. Bad combo.'
