What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

The Plunging Price of Thought

Connections

Supports

Tokens Too Cheap to Meter Both examine the economic trajectory of AI computation costs reaching near-zero, with 'Tokens Too Cheap to Meter' being a direct expression of the pricing trend the new item explores
DeepSeek-V4.1-Flash: Smaller, Faster AI Model DeepSeek-V4.1-Flash's smaller, faster model represents the efficiency gains driving cost reduction in AI processing that the new item examines
SWE-2: Advanced Coding Model with Improved Cost-Capability Tradeoff SWE-2's improved cost-capability tradeoff is a concrete example of the declining price of AI cognitive work that the new item theorizes about
Kev: Tiny Decision Models on Qwen Tiny decision models running on minimal hardware like Kev exemplify the commoditization of AI cognition that the new item's price-decline thesis predicts

Related to

Inside the Inference Hardware Revolution Of 2026 The inference hardware revolution of 2026 is a primary driver of the plunging computation costs the new item analyzes, making them complementary perspectives on the same phenomenon

Supported by

Mistral Large 4 Mistral Large 4's public preview availability and competitive positioning contributes to the trend of plunging inference costs as multiple large models compete for market share
Asking for What You Want: A Key 21st Century Skill Plunging cost of AI 'thought' is the economic enabler of the skill described - when computation is cheap enough, expressing intent becomes viable as a primary interface, reinforcing why this skill matters now
The Micro-machine Universe The plunging price of thought mirrors the thesis of micro-machine economics - semiconductor fabrication's cost curves are the template; the new item's framework explains structurally why computation (and thus AI inference) gets cheaper as machine uniformity scales
Thinking Will Become a Hobby Both argue that AI cognitive costs are plummeting, making human thought economically uncompetitive — 'plunging price of thought' directly underpins the claim that knowledge work will diminish
Regenerative Software: The Phoenix Architecture Plunging cost of AI inference directly enables the economics of regenerative software — when 'thought' is cheap, regenerating code on-demand becomes economically viable over maintaining legacy codebases
OpenCode 2.0 The plunging price of thought underpins OpenCode 2.0's thesis: as AI reasoning becomes cheaper, those building products for consumers gain disproportionate leverage
Worth building The plunging price of AI inference directly enables the economics described — cheaper tokens make building personal-scale tools without monetization viable, as the cost to run LLM-powered apps drops toward negligible.
Claude Opus 5.5 Visualizes Invisible Cities The 6-hour single-session creation of a complex three.js visualization of 55 cities supports the thesis about plunging cost of thought — what once required large teams now happens in hours with one model
GPT-6 For Everyone GPT-6 broad accessibility directly embodies the 'plunging price of thought' thesis — making frontier AI cheap enough for mass consumer use
Code Execution with MCP: Building Efficient AI Agents Token consumption reduction via MCP code execution is a direct response to the economics of thought described in 'Plunging Price of Thought' — efficiency gains enable more agentic compute per dollar
Kevin Kelly's 2016 AI Predictions, 10 Years Later Kelly's prediction that AI becomes Google's main product aligns with the 'plunging price of thought' thesis—commoditization of AI was a precondition for search being displaced by AI-native interfaces
GPT-6 For Everyone Democratizing GPT-6 access directly aligns with the trend of plunging AI thought costs — broader access is both cause and effect of cheaper inference.
Introducing Ember-1 Ember-1's 40% token reduction directly exemplifies the 'plunging price of thought' thesis — specialized efficiency-focused models are a key mechanism driving down inference costs
Engineering a Calibrated Classifier Model Hobson's design - optimizing a 2B parameter model for low latency classification - is a practical engineering response to inference cost pressures described in 'The Plunging Price of Thought'

Regulating What We Do Not Understand The plunging price of thought creates the economic conditions that make AI regulation debates urgent - rapidly falling costs accelerate deployment and make the question of preemptive vs. reactive oversight practically consequential
Code Execution with MCP: Building Efficient AI Agents Reducing token consumption via MCP code execution is a direct response to the economics discussed in 'The Plunging Price of Thought' - optimizing agent efficiency as inference costs shift
SSRN Research Paper #7496559 SSRN hosts significant economics-of-AI research; papers on the declining cost of AI computation and cognitive labor are a prominent category on the platform
The Dawn of the Age of the Exoskeleton The exoskeleton piece mirrors the 'plunging price' narrative: cost reduction is the central driver moving exoskeletons from $150K military units to accessible worker tools, following the same democratization arc as cheap AI inference
Yes, and... Programming's Future with AI The plunging price of AI-generated thought is the economic backdrop for Gross's career concerns — cheap inference commoditizes routine coding, the exact dynamic he warns junior programmers about

Challenged by

Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning If meta-reasoning adds a structured overhead layer before task execution, it complicates the 'thought is nearly free' thesis — the framework implicitly argues that naive scaling of inference compute is insufficient without structured allocation decisions
Real-Time Video Generation on Trainium Kernel optimization for real-time video generation on specialized hardware demonstrates that inference costs require significant engineering effort to reduce, nuancing claims that AI thought is simply becoming cheap

Type:	link
First published:	Thursday, 24 September 2026 at 06:30 UTC
Location:	https://epoch.ai/publications/the-plunging-price-of-thought
URL:	https://mattwood.fyi/i/c14eca7c

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections