Patel says chip orders take a minimum of three months from order to delivery from TSMC.
“the moment you even put an order in, even if TSMC started working on it immediately, you could not get a chip out for three more months.”
Patel says GPU orders take four to six months from placement to data center installation.
“call it four or five five, six months between, you know, when an order is placed and you can actually have it installed in your data center if it got worked on immediately.”
Patel says Google's TPU production numbers are closely guarded secrets that even most Google DeepMind employees don't know.
“Right? That's like a very closely guarded secret, most people that work at Google DeepMind don't even know number.”
Patel claims Google is massively ramping TPU production, a closely guarded secret most DeepMind employees don't know.
“Google's actually ramping up TPU production massively. And I think people in AI would be like, Well, duh. But like, okay, who has the capability of figuring out the number?”
Patel argues the semiconductor supply chain is highly fragmented with critical monopolies in countries like Austria.
“People don't understand how fragmented the semiconductor supply chain really is and how many monopolies there are. The US could absolutely shut shut down the Chinese semiconductor supply chain.”
Patel argues Western power grid supply chains have not expanded in decades unlike China, India, and Indonesia.
“The Japanese power grid has been like this. The Korean power grid has been like Only China and India and Indonesia, their power grids have been growing really fast.”
Patel argues TSMC's $100 billion US investment still relies on Taiwan R&D, not securing supply chain.
“Even if they do fully do the full $100,000,000,000 they commit to, completely still relying on R and D in Taiwan.”
Patel argues TSMC's $100 billion US investment won't secure R&D supply chain or new node development.
“The R and D fab they announced is not going to secure R and D supply chain, right? It's not going to secure new node development, right?”
Patel notes China was importing majority of robots as recently as 2022 from Japan and Germany.
“even as near as twenty twenty two, the majority of China's robots were being imported. Right? Japan and Germany being the largest market share players, but also USA, Korea is also in there.”
Patel notes that as recently as 2022, most of China's robots were imported from Japan and Germany.
“And so even as near as twenty twenty two, the majority of China's robots were being imported. Right? Japan and Germany being the largest market share players,”
Patel notes that as recently as 2022, China imported the majority of its robots from Japan and Germany.
“even as near as twenty twenty two, the majority of China's robots were being imported. Right? Japan and Germany being the largest market share players,”
Patel recounts that NVIDIA ordered Xbox production volume before receiving Microsoft's official order.
“No. No. No. Like, NVIDIA ordered the volume for the Xbox before Microsoft gave them the order.”
Patel reveals NVIDIA doubled system-level testing time for Blackwell compared to Hopper generation.
“NVIDIA learning from the issues on Hopper, with Blackwell, they actually doubled the time for system level test.”
Patel says 10-15% of NVIDIA GPUs fail and need RMA in the first two weeks after cluster deployment.
“When you first turn on the cluster, about ten to fifteen percent of them fail RMA in the first two weeks. Wow. And then that's fine.”
Patel predicts semiconductor shortages will return as the primary constraint in 2027 after power constraints ease.
“But it will fully beach semiconductors again in '27. Right? And so we see this across the entire space of the ecosystem. It's not just TSMC.”
Patel says memory makers have not built new fabs since 2022 due to cyclical nature.
“The memory makers, in fact, have just not expanded capacity. Basically, new they've not built new fabs since 2022. Yeah. Because their their cycle their cycle is so undulating.”
Patel predicts semiconductor shortages will return as primary constraint in 2027 affecting both TSMC and memory manufacturers.
“But it will fully beach semiconductors again in '27. Right? And so we see this across the entire space of the ecosystem. It's not just TSMC. It's also memory both.”
Patel claims China has the most vertical semiconductor stack and could run fabs independently unlike TSMC.
“China has the most vertical stack in semiconductors today and they're the best at semiconductors in the world because their fabs could still run somewhat on a lot of things because they have built some of these chemical supply chains. Right? Like TSMC for certain kinds of chemicals 100% share from Japan.”
Patel identifies semiconductors as the returning bottleneck, with chip fabs being most complicated buildings requiring multi-year timelines.
“Semiconductors are extremely cyclical. The the the buildings that chips are made in are the most complicated buildings people make. You know, they are they are know, they have multi year timelines.”
Patel explains TSMC must follow US export restrictions regardless of Taiwan's government due to banking and equipment dependencies.
“The way American export restrictions are upheld is that Taiwan utilizes American banking systems, American equipment industries. And so they'll still have to uphold any US export restrictions.”
Patel contrasts competitors' limited scale with NVIDIA's supply chain designed for tens of millions of chips.
“I can buy 10,000. I could buy a 100,000. I can't buy millions, tens of millions, which is what Jensen's, setting his supply chain up for.”
Patel says NVIDIA has locked up over 60% of market capacity in long-term contracts this year.
“NVIDIA has locked up over 60% of the capacity this year in long term contracts alone, and they're buying more on top of that.”
Patel reports NVIDIA will sign over $250 billion in supply contracts this year across components.
“as you look at what they're negotiating in the market today, there's over 250,000,000,000 of wafers, of memory, of substrates, of PCBs, of networking equipment that they're going to sign this year,”
Patel lists AMD MI350, Amazon Trainium 3, and Google TPU v7 as AI chips now on 3nm process node.
“MI three fifty from late last year from AMD or Tranium three and TPU v seven from late last year, or slash early this year from Amazon and Google”
Patel reveals NVIDIA acquired Grok partly because Grok chips are made on Samsung, accessing non-TSMC capacity.
“Part of it was because they want to have really fast inference, but like part of it is that Grok is manufactured on Samsung. Right?”
Patel says AI chips consuming all TSMC 3nm and 2nm capacity is forcing companies to alternative fabs.
“More importantly is that AI is buying all the capacity on three nanometer and two nanometer in a couple years that, people are having to turn to other directions.”
Patel says TSMC is forcing Apple, Qualcomm, and MediaTek off three-nanometer, pushing them toward Intel.
“They're telling Qualcomm and MediaTek to get off and all three of these companies are like, oh, maybe we should use Intel because they're not telling us to get off”
Patel explains AI chips are crowding onto three nanometer nodes, displacing mobile chip production.
“Right? Because all the AI chips are on three nanometer and it takes time, Small mobile chips are easier to make than big AI chips, so all the AI chips are moving to three nanometer now, right? MI three fifty from”
Patel says AI chips are consuming all three-nanometer capacity, forcing others to alternative fabs.
“AI is buying all the capacity on three nanometer and two nanometer in a couple years that, people are having to turn to other directions.”
Patel explains NVIDIA acquired Grok partly because it manufactures on Samsung due to no TSMC 3nm capacity available.
“Because there's no three nanometer capacity for them at TSMC, they need to chip somewhere else, and if you just if AI is as crazy as like we believe it is, and demand is as crazy as we believe it is, it's gonna be even crazier next year,”
Patel notes China has a fully self-sufficient vertical semiconductor supply chain, though behind by three to fifteen years.
“China has a vertical supply chain. Yes. In many cases, it's fifteen years behind, ten years behind, in some cases, only three years behind.”
“We try and track the entire supply chain from tools that manufacture chips, fabs, data centers, energy, industrials, and then AI models and who's using them, how much, and where.”
Patel estimates five to eight years of global tool manufacturing would be needed to replace Taiwan's chip capacity.
“It would take five to eight years of all of the world's tool manufacturing capability Yeah. To even replace the capacity in Taiwan.”
Patel reveals Google runs three separate TPU design programs with Broadcom, MediaTek, and undisclosed partner.
“They're making a TPU with Broadcom, that's a different architecture than a TPU with MediaTek, that's a different TPU than the architecture that is, I won't disclose, by research.”
Patel predicted memory prices would soar because capacity grows 30% annually while demand doubles.
“Memory capacity is only growing 30% a year for the next three years. And yet demand is doubling, is doubling. And so what's gonna end up happening is memory prices are gonna keep soaring.”
Patel predicts iPhone and MacBook prices must rise as AI demand drives memory shortages.
“And right now, you know, if MacBook prices or iPhone prices go up a $100, that market's not gonna adjust too much,”
Patel defends SemiAnalysis's 15-20 weekly reports as clear updates, saying industry clients make useful decisions with data
“@damnang2 What's the solution? Make every report a 2000 word salad that reexplains everything? My firm puts out 15-20 reports a week about supply chain and technology for our clients. We try to share clearly and consicely what's happening. The people who boil it down to some silly trade, I”
Patel says no ASML lithography tool can be made or operated without US supply chain
“@voidtarget @lithos_graphein Machine = tool And no, no ASML litho tool can be made and operated without US supply chain.”
Semiconductor production capacity exists but is constrained by a specific component shortage in lithography equipment.
“We could make a trillion dollars right now, but we’re just bottlenecked on the mirrors that go into the ASML machines.”