← Explore

Posts tagged with inference-economics

GPU Economics · ·5 min read

Forty-Six Billion Through 2031, and Google Still Called AMD

SemiAnalysis dropped a client note on August 16 that read like a plot twist in a supplier drama: Google is working with AMD on a variant of its 10th-generation...

custom-asicgoogle-tpuamd
Neural Dispatch · ·4 min read

Per-Token Costs Dropped 95%. Your AI Bill Went Up Anyway.

Gartner published a prediction on Sunday that deserves more attention than it's getting: inference costs per agentic workflow will increase more than...

inference-economicsagentic-aigartner
GPU Economics · ·5 min read

AMD Paid Ten Percent of Itself to Land OpenAI

The number that matters in AMD's deal with OpenAI isn't six gigawatts. It's 160 million shares.

amdopenaiinference-economics
Neural Dispatch · ·5 min read

Eight Trillion Tokens Broke DeepSeek's Pricing Model

On August 1, DeepSeek's API processed 8 trillion tokens in a single day — 5 trillion from free-tier users alone.

deepseekapi-pricingself-hosting
GPU Economics · ·5 min read

Intel Bet on Cheap Memory. Then Memory Stopped Being Cheap.

Intel walked onto the Computex stage in June with a pitch that made a certain kind of sense: skip the HBM shortage entirely.

intelcrescent-islandlpddr5x
GPU Economics · ·5 min read

Eight Gigawatts and a Penny Per Share

AMD handed OpenAI warrants for 160 million shares at a penny each. Nine months later, it wrote Anthropic a check for $5 billion.

amdnvidiainference-economics
Neural Dispatch · ·5 min read

Microsoft Just Put Its Engineers on a Token Budget. Yours Is Next.

Jay Parikh doesn't do subtle.

microsofttokenmaxxingai-costs
GPU Economics · ·5 min read

A Dollar Twenty an Hour, Fifty-Two Weeks to Ship

Spot-market H100 instances dropped below $1.20 an hour in June.

hbmsupply-chaincowos-packaging
GPU Economics · ·5 min read

Forty-Four Percent Growth, Zero Percent Portability

Midjourney slashed its inference bill from 2.1 million a month to 700,000 by moving from NVIDIA A100/H100 clusters to Google TPU v6e pods.

custom-asicnvidiainference-economics
GPU Economics · ·5 min read

Who's Buying a $6,000 H100?

Somebody is. In volume.

h100secondary-marketdepreciation
Neural Dispatch · ·4 min read

Kimi K3 Shipped 1.4TB of Open Weights. Most Devs Will Never Load Them.

Moonshot AI's Kimi K3 weights went live at midnight UTC — 594 gigabytes of MXFP4 safetensors, the largest open-weight release in AI history.

kimi-k3open-weightsinference-economics
GPU Economics · ·5 min read

The First Hyperscaler to Admit It Built Too Much

On July 1, Meta announced Meta Compute — a cloud business unit that will sell surplus AI training and inference capacity, plus hosted access to its Llama and...

meta-computeovercapacityinference-economics
GPU Economics · ·5 min read

AWS Hiked GPU Prices Twice This Year. The Market Didn't Blink.

On July 1, AWS raised its EC2 Capacity Block prices for GPU instances by 20%. This was the second hike in 2026 — a 15% bump landed earlier in the year.

gpu-pricingawsinference-economics
GPU Economics · ·5 min read

Twenty Percent for Three Million Dollars

Three million dollars. That's the total exercise cost if both OpenAI and Meta decide to cash in every last warrant AMD handed them.

amdnvidiafinancial-engineering
GPU Economics · ·5 min read

Fifty-Two Weeks to Break Even

The math on GPU ownership used to be simple.

gpu-shortagesupply-chaincowos
GPU Economics · ·5 min read

$200 Billion Starts With a Laptop

On Monday morning, three chipmakers woke up to a problem they didn't have on Friday. Intel dropped 6%.

nvidiartx-sparkpc-market
GPU Economics · ·5 min read

Two Chips Are Cheaper Than One

Google just did something no major chip designer has tried at this scale: they designed two separate processors for a job everyone else handles with one.

google-tpuinference-economicsspecialized-silicon
Synthetic Media · ·4 min read

Fifteen Million Dollars a Day

OpenAI spent roughly thirty months building the most impressive video generation demo the industry had ever seen, then watched it bleed $15 million per day in...

video-generationsoraopenai
GPU Economics · ·5 min read

NVIDIA Paid $25 Billion to Compete With Itself

Between December 2025 and March 2026, NVIDIA wrote two checks that contradict everything the company built over the last decade.

nvidiagroqinference-economics
GPU Economics · ·5 min read

Your GPU's Most Expensive Part Isn't the GPU

Pick up an NVIDIA B200 and trace where the roughly 6,400 manufacturing cost actually goes.

hbmmemory-economicsinference-economics
1 / 2 Next →