Article archive
Published news and blog articles, organized by category. Browse older coverage by month or search for a topic. Undated blog guides appear after dated news.
9,484 articles · Newest first
From Solitaire to ChatGPT: The Century-Old Math Behind Machine Prediction
The statistical machinery inside modern language models - Markov chains, the law of large numbers, Monte Carlo simulation, and next-token prediction - was assembled decades before anyone trained a neural network. N43 traces the lineage…
What's Actually Inside Your Smartphone: A Component-by-Component Tour
A flagship phone packs a computer, a radio stack, a camera system, and a sensor suite into a slab about seven millimeters thick. N43 tours the components one by one - the SoC, the memory stack, the RF chain, the display, the battery - and…
How Bluetooth Works: Frequency Hopping and the 2.4 GHz Shuffle
How Bluetooth actually works: the crowded 2.4 GHz ISM band, frequency hopping at up to 1600 hops per second, pairing and bonding, Classic vs Low Energy, and the LE Audio and LC3 shift.
How Under-Display Fingerprint Readers Actually Work
Optical and ultrasonic sensors turned phone unlocking into a trick the screen plays on itself: light bouncing off your ridges, or sound pulses mapping them in 3D through the glass.
How Wi-Fi Works: The Invisible Network Inside Your Walls
How Wi-Fi actually works: radio waves in the 2.4, 5, and 6 GHz bands, the CSMA/CA protocol that keeps transmitters from colliding, WPA2 and WPA3 security, and the OFDMA and multi-link features inside Wi-Fi 6 and Wi-Fi 7.
NFC and Tap-to-Pay: The Two-Inch Radio That Replaced Your Card
Near-field communication runs on induction at 13.56 MHz and works over about four centimeters. Here is what actually happens in the half second your card or phone spends on a payment terminal.
The Hidden Bill: How AI Data Centers Are Quietly Raising Prices on Everyone
The AI data center buildout is rewriting the economics of the American power grid, and the marginal cost of electricity is landing on ratepayers who never asked for it.
Open Weights, Closed Gaps: How GLM 5.2 Became the Open-Source Model to Beat
GLM 5.2 from Zhipu AI sits at the top of open-weight leaderboards. Its rise explains what open weights really mean for the LLM market — and where the strategy came from.
Twelve Leaks, One Phone: Reading the iPhone 18 Pro Max Rumor Tea Leaves
The iPhone 18 Pro Max leak cycle is in full swing: A20 silicon, battery chemistry, camera changes, and on-device AI. A guide to what leaks actually say — and how much to trust each.
Pixel 11 Pro XL Review Season: Why Google’s Flagship Is Dividing the Comment Section
Dave2D calls it worse. Matthew Moniz calls it not even close in Google's favor. The Pixel 11 Pro XL review split reveals what 2026 flagship phones are actually competing on.
The Cancellation Wave: Why Tech Giants Are Quietly Pulling Back on AI Data Centers
After two years of frantic AI data center construction, cancellations and delays are mounting. Rising capital costs, power grid constraints, and efficiency gains are forcing a correction, and the supply chain is already feeling it.
The iPhone 18 Pro Leak Cycle: What the 2026 Rumor Mill Says About Apple's AI Play
Everything leaked about the iPhone 18 Pro so far points one direction: thinner hardware, a 2nm A20 chip, and an Apple Intelligence stack built on on-device inference. We separate the reliable signals from the noise.
Colossus in Memphis: How xAI Built the World's Largest AI Supercluster at Record Speed
xAI's Colossus cluster put 100,000 H100-class GPUs online in Memphis in roughly 122 days. N43 examines the engineering, the power, the constraints, and what the fastest supercomputer build in history signals about the compute race.
The GPU Underground: Inside the Black Market for AI's Most Wanted Silicon
Export controls wall NVIDIA's AI GPUs off from China. A gray market grew anyway. N43 examines smuggled H100s, shell-company routes, price premiums, and what the underground means for the compute race.
The Quantum Computer Dream Is Getting a Reality Check
A decade of quantum hype is meeting engineering reality. We examine the qubit numbers, the error-correction overhead, and the one experiment that genuinely moved the field.
The AI Phone Grows Up: How Google Pixel 10 Puts Intelligence On Device
The Pixel 10 pairs the Tensor G5 chip with Gemini Nano to run language models directly on the handset. We examine what on-device AI actually delivers, and what it gives up.
What Makes the Fastest Phone in the World: Inside 2026 Flagship Chipsets
PhoneBuff-style speed tests crown a winner every year, but the gauntlet measures memory, storage and thermals as much as CPU muscle. We look inside the silicon that decides the result.
The 2026 AI Model Wars: GPT, Claude and Gemini Enter the Arena
Frontier models from OpenAI, Google and Anthropic now ship faster than reviewers can test them. We break down what the competition is actually about, and where the marketing outruns the machines.
From Words to Vectors: How Embeddings Give Language Models Meaning
How words became numbers: from word2vec to contextual embeddings inside LLMs, cosine similarity, semantic search, RAG, and why vector databases exist.
How MediaTek Won the Phone Chip War
From budget-chip underdog to the world's largest smartphone SoC vendor by volume: the strategy, silicon, and economics behind MediaTek's climb to the top.
The Siting Problem: Where AI Data Centers End Up, and Why
Cheap power, fiber, water, and interconnection queues decide where AI data centers are built — and why so many end up next to small towns that never expected them.
Why AI Models Hallucinate: The Mechanics Behind Confident Nonsense
Why large language models produce fluent, confident, false statements: next-token probability, training that rewards plausibility, and why hallucination resists elimination.
From Autocomplete to Assistant: How AI Chatbots Actually Work
Strip away the conversation and an AI chatbot is a probability engine over tokens. This is the machinery underneath the pleasantries — tokenization, training, hallucination, and the product layer that turns a model into an assistant.
The Water Bill of Intelligence: How AI Data Centers Tap Cities' Drinking Supplies
AI is usually described as compute-hungry, but in a growing number of communities the thirstiest part of the stack is water. Evaporative cooling turns heat into vapor, and the make-up water comes from somewhere real.
Also explore blog articles and guides or search all DutyStation pages.