#AI chips
24 articles tagged AI chips. This tag now lifts the site in topic search.
Timeline
-
Malaysia seizes 30 AI chip servers — $12.1 million, the fifth case tied to one syndicate this year
-
What is transshipment? Why AI chips reach China via Malaysia, Singapore and Thailand
-
What is model quantization: how a 1.5TB AI model shrinks to 4 bits and fits on one server
-
Taiwan moves to criminalize AI chip smuggling to China — all of China, not just Huawei and SMIC
All articles
-
Tech · 5 min readMalaysia seizes 30 AI chip servers — $12.1 million, the fifth case tied to one syndicate this year
Seizure — 30 AI servers worth about RM49.5 million found in an airport free-zone warehouse on October 15
-
Tech · 7 min readWhat is transshipment? Why AI chips reach China via Malaysia, Singapore and Thailand
Definition — unloading in a middle country and reloading onto another carrier, a daily part of trade
-
Tech · 2 min readWhat is model quantization: how a 1.5TB AI model shrinks to 4 bits and fits on one server
Idea — store each number in 8 or 4 bits instead of 16
-
Tech · 2 min readTaiwan moves to criminalize AI chip smuggling to China — all of China, not just Huawei and SMIC
Plan — AI servers and advanced chips added to Foreign Trade Act controlled items
-
Tech · 2 min readTSMC Q3 2026 net income NT$763 billion — a sixth straight record, and 2nm is already 8% of sales
Results — net income NT$763.06B (+68.7%), revenue US$46.72B (+41.2%), above guidance
-
Tech · 2 min readMicrosoft Maia 300 unveiled — TSMC 2nm, eight HBM4 stacks, its main AI inference chip from 2027
Chip — Maia 300: TSMC N2P compute chiplets, eight HBM4 stacks at 18 TB/s, 950 W (Maia 200: 750 W)
-
Tech · 3 min readWhat a 2nm process is — nothing on the chip is 2 nanometers, so why AI chips line up for it
Name — '2nm' is a generation label, not a dimension. The real change is the transistor: FinFET to GAA nanosheets
-
Tech · 3 min readLisa Su meets 14 Korean AI chip firms — AMD bets on mixing GPUs with Korean NPUs
Meeting — Lisa Su met Korea's science ministry and 14 local firms, incl. Rebellions and FuriosaAI, Oct 7
-
Tech · 3 min readWhat is an NPU — how it differs from a GPU, and why it wins at inference
Definition — a chip for neural-network math only; it trades generality for efficiency
-
Tech · 2 min readEtched weighs $40B–$50B valuation offers — four times its July price in three months
Numbers — $10.3B in July → $21B in September → $40–50B offers in October
-
Economy · 3 min readMarvell falls 8.2% on August 28, 2026 — $120bn from Google, but not until FY2029
Results — quarterly revenue $2.739bn (+37%); data center $2.17bn (+46%), 79% of the total
-
Tech · 3 min readCustom AI chips (ASICs) — why Google builds its own instead of buying Nvidia
Definition — silicon designed for one company's one workload, not sold as a general part
-
Tech · 3 min readNvidia adds $442 billion in market value on August 27, 2026 — second biggest ever, behind Microsoft
Record — 442 billion dollars added in one session, second largest ever, 8 billion behind the record
-
Tech · 3 min readNvidia Q2 revenue $96.2 billion (August 26, 2026) — the outlook assumes zero from China
Results — Revenue $96.2bn (+106% YoY); Data Center $89.02bn (+117% YoY). Both above consensus
-
Tech · 3 min readOpenAI unveils Jalapeño, its first chip — 700W beat Nvidia parts rated at 1,200W
What it is — A general-purpose LLM inference accelerator. It runs models; it does not train them
-
Tech · 3 min readNvidia reports on August 26, 2026 — the bar is $92 billion
Timing — 4:20pm ET release and 5:00pm call on August 26, which is the morning of August 27 in Seoul
-
Tech · 4 min readWhat AI chip export controls are — three years of moving the line
Definition — US rules on which AI chips go where, enforced by the Bureau of Industry and Security
-
Tech · 3 min readWhat comes after HBM — Samsung stacks onto the GPU, SK bridges to NAND
Samsung unveiled zHBM and 400-plus-layer 10th-generation NAND at FMS 2026 in Santa Clara
-
Tech · 1 min readWhy great earnings can sink a stock — reading AI results
The same evening: AMD -9% on record revenue, SpaceX +9% on a $541M loss
-
Tech · 1 min readAMD's record quarter — and the 9% after-hours drop
Q2 revenue hit a record $11.5B (+50%) with data center doubling to $6.7B
-
Tech · 2 min readSK hynix lifts capex near 60% for the AI build-out
SK hynix is expanding this year's capex to the high 40 trillion won range, up about 60%
-
Tech · 3 min readHBM4 vs HBM3E: what doubles, and why it matters
HBM4 doubles I/O pins from 1,024 to 2,048 and channels from 16 to 32, taking per-stack bandwidth past 2TB/s
-
Tech · 2 min readHBM4 ramp: the race turns to supply
SK hynix started HBM4 mass-production shipments in Q2 and signed long-term deals
-
Tech · 2 min readHBM explained: why stacked memory is AI's bottleneck
HBM stacks 4 to 16 DRAM dies connected by through-silicon vias to maximise bandwidth