DELIGHT
SCALE
Dispelling DeepSeek Myths, Studying V3
Remember Jan 27th when Nvidia lost half a trillion dollars in market cap? An inconspicuous table buried in a paper played a key role: What did this paper actually say? Welcome to the third and final post in a series looking at DeepSeek. Let’s think things through from first principles. We can talk about misplaced investor panic…
Apple M4 MacBook Air – Apple’s Continued Aggressive SoC Power Management
Keeping up my theme of constantly analyzing the battery life of each new Apple M series silicon, I was excited to dive into the new M4 Macbook Air. Every year, gen on gen, Apple continues what their silicon design teams like to call the relentless pursuit of performance per watt. While the M4 is not…
A weekend with Apple’s Mac Studio with M3 Ultra: The only real AI workstation today
Executive Summary: The Apple Mac Studio featuring the M3 Ultra represents the most powerful AI workstation currently available, tailored specifically toward AI developers’ demanding workflows. With its unprecedented unified memory (up to 512GB) and robust GPU performance, the M3 Ultra Mac Studio excels at running large language models (LLMs) efficiently, surpassing even high-end PCs in…
Nvidia 4Q25 Earnings
Nvidia’s 2025 earnings call was more revealing than it seemed. Is cost-per-token now the key battleground for AI infrastructure? Why is Nvidia pushing Ethernet, and what does that mean for Infiniband? What happens to aging GPUs? We’ll also explore Nvidia’s positioning against AI ASICs, revenue concentration risks, and whether AMD’s modularity can still be a…
Alexa+: Amazon’s Home Assistant Grown Up to Be a Full-Fledged AI Agent
The recent unveiling of Amazon’s enhanced Alexa with generative AI capabilities marks a significant milestone in the evolution of home-based AI assistants. As tech companies increasingly focus on generative AI capabilities, Amazon has carefully reimagined its digital assistant while preserving what made Alexa special in the first place: a personal and context aware helpful companion…
DeepSeek MoE & V2
We previously looked at DeepSeek’s reasoning model R1 and distilled variants. I started with the reasoning model because it could upend OpenAI’s business model, which seems to be playing out: Source Why keep reasoning proprietary after DeepSeek’s open weights R1? Yet a lot of investors freaked out because of the V3 paper. So I set…