Top all-time · #171
SemiAnalysis
Company blog · semianalysis.comhardwarebusinessai
SemiAnalysis is the company blog at semianalysis.com, writing about hardware, business and ai. Since 2020, 268 different Hacker News readers have linked it 479 times, most often “Google "We Have No Moat, and Neither Does OpenAI" (2023)”. It is still writing: the last post was on 2025-09-16, and you can follow it by feed.
Rarely recommended · 2 of 40 sampled comments recommend it?
Why HN links it
HN readers cite SemiAnalysis as a source on chips, AI hardware and GPU costs, especially its "We Have No Moat" post and its DeepSeek cost analysis.
Summarised by Claude from a sample of 40 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Independent research on the AI and semiconductor supply chain. Read by 318,000+ subscribers. Fourteen institutional models, open benchmarks, and advisory.
Most-linked posts?
- Google "We Have No Moat, and Neither Does OpenAI" (2023)38 people · 40 mentions · first linked 2023-05-04
- DeepSeek Debates: Chinese Leadership on Cost, True Training Cost11 people · 13 mentions · first linked 2025-01-31
- Qualcomm says ARM to prohibit proximity of CPU and 3rd-party modules in one chip8 people · 8 mentions · first linked 2022-10-28
- Zen 4c: AMD’s Response to Hyperscale ARM and Intel Atom7 people · 8 mentions · first linked 2023-06-05
- Google "We have no moat, and neither does OpenAI" (2023)7 people · 7 mentions · first linked 2025-01-27
- Amazon's AI Self Sufficiency – Trainium2 Architecture and Networking6 people · 6 mentions · first linked 2024-12-03
- GPT-4 Architecture, Infrastructure, Training Dataset, Costs, Vision, Moe6 people · 6 mentions · first linked 2023-07-10
- TSMC’s 3nm Conundrum, Does It Even Make Sense?6 people · 6 mentions · first linked 2022-12-21
- CUDA Moat Still Alive5 people · 6 mentions · first linked 2024-12-22
- Intel on the Brink of Death5 people · 6 mentions · first linked 2024-12-09
What Hacker News says?
Again, that's not a moat. The original "We Have No Moat, And Neither Does OpenAI" leaked memo from Google that memefied the term focuses explicitly on the increasing ease of competitors (especially open-source) entering the ecosystem: https://www.semianalysis.com/p/google-we-have-no-moat-and-ne...
Comment linking /p/google-we-have-no-moat-and-neither
on “Sora: Creating video from text”Submitted Apple’s A14 Packs 134M Transistors/mm² as a story · 314 points
HN discussion> In order to break even they would have to minimize the operating costs (by throttling, maiming models etc.) and/or increase prices. This would be the reality check. Is that true? Are they operating inference at a loss or are they incurring losses entirely on R&D? I guess we'll probably never know, but I wouldn't take as a given that inference is operating at a loss. I found this: https://semianalysis.com/2023/02/0…
Comment linking /2023/02/09/the-inference-cost-of-search-disruption/
on “LLM Inevitabilism”Submitted America Is Missing The New Labor Economy – Robotics Part 1 as a story · 268 points
HN discussionThe fact that TSMC publishes their own metrics and target goals for each node makes it straightforward to compare the transistor density, power efficiency, etc. The most interesting aspect of the M4 is simply it's debuting on the iPad lineup, whereas historically it's always been on the iPhone (for A-series) and Macbook (for M-series). Makes sense given low expected yielded for the newest node for one of Apple's low…
Comment linking /p/tsmcs-3nm-conundrum-does-it-even
on “Apple introduces M4 chip”Submitted Qualcomm says ARM to prohibit proximity of CPU and 3rd-party modules in one chip as a story · 242 points
HN discussionA reference to the good doc: https://www.semianalysis.com/p/google-we-have-no-moat-and-ne... While I'm linking semianalysis, though, it's probably worth talking about how everyone except Google is GPU poor: https://www.semianalysis.com/p/google-gemini-eats-the-world-... (paid) > Whether Google has the stomach to put these models out publicly without neutering their creativity or their existing business model is a di…
Comment linking /p/google-gemini-eats-the-world-gemini
on “Our next-generation model: Gemini 1.5”Submitted CUDA Moat Still Alive as a story · 221 points
HN discussionIf anybody Google would know most about that after their LLM memo all that time ago (basically "we're losing because we're trying to fight/compete with OS models"): https://www.semianalysis.com/p/google-we-have-no-moat-and-ne...
Comment linking /p/google-we-have-no-moat-and-neither
on “Diffusion models are real-time game engines”Submitted The Semiconductor Heist of the Century as a story · 163 points
HN discussion> uses an _embedded_ silicon interposer It sounds like they're using TSMC's new LSI (Local Si Interconnect) technology, which is their version of Intel's EMIB. It's essentially small islands of silicon, just around the inter-chip connections, embedded within the organic substrate. This gives the advantages of silicon interconnect, without the cost and size restrictions of a silicon interposer. It would not be visibl…
Comment linking /2022/01/06/advanced-packaging-part-2-review/
on “Apple M3 Ultra”Submitted An inside look at the custom CPUs in Tesla's Dojo Supercomputer as a story · 155 points
HN discussionAs the Google memo ( https://www.semianalysis.com/p/google-we-have-no-moat-and-ne... ) pointed out, a lot of OSS stuff/improvements are being built on top of Meta's work which somewhat benefits them as well. But still, Kudos to Zuck/Meta for doing it anyway.
Comment linking /p/google-we-have-no-moat-and-neither
on “Llama 3.2: Revolutionizing edge AI and vision with open, customizable models”Submitted Jensen Huang’s vision for data center dominance may destroy the Arm ecosystem as a story · 135 points
HN discussionI think you'll be surprised if you see the lift karpathy demonstrates from `fineweb.edu` vs `webtext` (he went back later and changed the `nanogpt` repository to use `openwebtext` because it was different enough that it wasn't a good replication of GPT-2). But from an architecture point of view, you might be surprised at how little has changed. Rotary and/or alibi embeddings are useful, and there's a ton on the infe…
Comment linking /2023/05/04/google-we-have-no-moat-and-neither/
on “The new skill in AI is not prompting, it's context engineering”Submitted Intel Betting the Farm as a story · 121 points
HN discussionI think they will be, but more for hand-off. Local will be great for starting timers, adding things to calendar, moving files around. Basic, local tasks. But it also needs to be intelligent enough to know when to hand off to server-side model. Android crowd has been able to run LLMs on-device since LlamaCPP first came out. But the magic is in the integration with OS. As usual there will be hype around Apple, idk, in…
Comment linking /2023/05/04/google-we-have-no-moat-and-neither/
on “iPhone Air”Submitted The Memory Wall: Past, Present, and Future of DRAM as a story · 106 points
HN discussion- nl · 2026-09-02
You are absolutely right, I apologize. I think I was talking about A100 prices (which are still only 6 years old) and conflated a few different things there. But A100 rental prices have climbed since 2024 (as far back as free account records show on https://semianalysis.com/gpu-pricing-index/ ). Coreweave has announced they will keep A100s in use until 2029 which will be 9 years old then. I think that is where I got…
Comment linking /gpu-pricing-index/
on “How accurate have Ed Zitron's AI skeptic predictions been?” Submitted Google "We have no moat, and neither does OpenAI" (2023) as a story · 95 points
HN discussion- nl · 2026-09-02
> These companies have been caught tweaking their numbers... they build data centers through holding companies, who have to absorb the costs and massive capex based financial liabilites, so that the brand-name big-tech companies get to keep their expenses off their books. This is about as far from "tweaking their numbers" as you can get. It's a standard way infrastructure-heavy industries structure their investments…
Comment linking /gpu-pricing-index/
on “How accurate have Ed Zitron's AI skeptic predictions been?” Submitted Arm Changes Business Model – OEM Partners Must Directly License from Arm as a story · 85 points
HN discussionWhen you start using Samsung 4 nm are you switching from SRAM to HDM? If yes, how's that going to affect all the metrics given that SRAM is so much faster? Someone said you'll eventually move to HDM because SRAM improvements is relatively stalled. I read an article that indicated your Bill Of Materials compared to NVidia's is 10x to get 1/10 the latency, and 8x BOM for throughput if Nvidia optimizes for throughput?…
Comment linking /p/groq-inference-tokenomics-speed-but
on “Groq runs Mixtral 8x7B-32k with 500 T/s”Submitted The Tesla Dojo Chip Is Impressive, but There Are Some Major Technical Issues as a story · 82 points
HN discussionThis article was a really interesting take on this too: https://semianalysis.com/2025/03/11/america-is-missing-the-n... The tldr of that post is: - To be really good at making robots, you need to iterate fast - To iterate fast, you need all component manufacturers nearby (or you’ll be wasting weeks shipping parts from somewhere else) - To be really competitive at manufacturing, you need to be good at robotics. - If…
Comment linking /2025/03/11/america-is-missing-the-new-labor-economy-robotics-part-1/
on “America underestimates the difficulty of bringing manufacturing back”Submitted Nvidia's Blackwell Reworked – Shipment Delays and GB200A Reworked Platforms as a story · 67 points
HN discussionThe ARM China venture has gone rogue anyway. It already has exclusive access to ARM IP within China. https://semianalysis.com/the-semiconductor-heist-of-the-cent...
Comment linking /the-semiconductor-heist-of-the-century-arm-china-has-gone-completely-rogue-operating-as-an-independent-company-with-their-own-ip/
on “Nvidia prepares to abandon $40B Arm bid”Submitted AMD Genoa Detailed – Makes Xeon Look Like a Dinosaur as a story · 54 points
HN discussion> 5. This means, the after ARM gets owned by a US company, the US can possibly just cut-off ARM supply to China China has already hijacked the ARM branch in China[1] and taken over ARM's IP. [1] https://semianalysis.com/the-semiconductor-heist-of-the-cent...
Comment linking /the-semiconductor-heist-of-the-century-arm-china-has-gone-completely-rogue-operating-as-an-independent-company-with-their-own-ip/
on “Nvidia prepares to abandon $40B Arm bid”Submitted The iPhone 13 Does Not Have Satellite Internet, Band N53 and Globalstar as a story · 54 points
HN discussion
