Anthropic S-1, the enterprise math: $16.3B revenue in H1, up 1,300% ~80% of it metered by tokens Fortune 500 $100K+ customers: <0.1% of their expenses 10x from here doesn't come from price. It comes from CFOs agreeing on what the value is. That's what the Tokenomics Foundation is standardizing. tokeneconomics.com/projects/

Last Tuesday I brought about ~80 people into a working room at the Westin St. Francis for Tokenomics SF. Chatham House, no sales pitches. AI Practitioners, engineering leaders, finance leaders, researchers, and the people building the AI factory, from NYU, Pinterest, Adobe, AMD, NVIDIA, Nebius, AWS, and BNY, among dozens of others. What I took away: Cost per token is the wrong place to stop. The chain runs from token to task to outcome to something a CFO can actually book. Almost nobody connects all four links yet. The people making tokens and the people using them barely speak the same language. "More tokens, more revenue" is not a business case. Value is the thing everyone forgets. Most teams feel confident about visibility into spend. Very few can show value a CFO would defend. Task-level productivity gains don't automatically add up. If code still waits on review and deploy, the saved minutes disappear. Hard token caps per person can backfire. People wait for tomorrow's allowance instead of finishing the work. Inference software is moving so fast that skipping upgrades means leaving performance on the table. And no single data source tells the story. Cost, telemetry, outcomes, and value have to land in one place. Thank you to everyone who showed up and argued in good faith. Next up: Tokenomics NYC Executive on Dec 8, on the 102nd floor of One World Trade Center. Same format, for CFOs, CIOs, CTOs, and SVP/EVP leaders in the Fortune 500. Applications are open through Nov 30: tokeneconomics.com/tokenomic…

Scale doesn't buy clarity on AI value. State of #Tokenomics data: 34% of companies under $100M revenue can confidently tie AI spend to an outcome their CFO accepts. At $100M-$1B, 14%. At $10B+, 23%. More models, more fog. tokeneconomics.com/state-of-…

Every token begins as electricity with a hope of driving value. Tokenomics is what happens between the grid and the ledger.

Weekly tokens on @OpenRouter went from 0.5T to 128.4T in 20 months. Roughly 250x. At that growth, cost per token is not your budget problem. Volume is. Enterprises need unit economics before the bill finds them. openrouter.ai/data/newslette…

youtube.com/watch?v=gPWtDsMp… 17 min highlights reel from the State of Tokenomics with interview of Deutsche Bank's CTO. Key takeaways - Three in four enterprises cannot show their CFO value on their AI spending. Only 7% named price as their biggest challenge. - The mix is shifting: 83% expect to move more toward open weights models over the next 12 months, while still running frontier. - Model routing is being run or evaluated by 86% of organizations, and the biggest enterprises are building and buying routers at once. - Ownership matters. Organizations with a defined owner were 3.7x more likely to show value. - Mission critical AI cannot simply be switched off, which pushes AI OpEx back toward capacity planning.

Enterprises aren't asking AI providers for discounts. In the State of #Tokenomics data, 23% asked for more transparency and granular usage data. Only 4% asked for cheaper prices. They want a bill they can explain to the CFO. tokeneconomics.com/state-of-…

From the State of Tokenomics… Half of AI consumers spread across 4 of the 6 procurement/hosting channels. Nearly one in five already uses at least one of Deepseek, Qwen, Kimi or GLM. tokeneconomics.com/state-of-…

State of Tokenomics data: three in four enterprises cannot confidently prove AI business outcomes to the CFO. tokeneconomics.com/state-of-…

AI #tokenomics is at that moment where most of the market is staring incorrectly at token costs as the entire thing.... But the full bill includes energy, capital, silicon, data centers, routers and gateways and observability software and licensing, and critically, the labor to build and review and supervise all that AI. Tokens carry somewhere between a tenth and a quarter of total run cost. The total line from tokens is not the bill. Time to focus on the Total Cost of AI (TCA). tokeneconomics.com/insights/…

Episode 17 #Tokenomics Brief: Why Tokens Aren’t the AI Bill: Total Cost of AI (TCA) I took a Labor Day break to record learnings in Tokenomics Foundation's working groups: the proposed "TCA (Total Cost of AI)" metric over "Value Realized" unit economic equation... Here I recap a busy 10-day stretch for the Tokenomics Foundation’s four working groups and proposes a new organizing metric: Total Cost of AI (TCA), arguing that token/API costs are only 10–25% of total AI spend once energy, capital, infrastructure, software, labor, and process impacts are included. 00:00 Why AI Spend Is Misread 02:43 Introducing Total Cost of AI 04:18 Definitions Update and Personas 07:08 Production Meter Boundaries 10:31 Consumption Routing and Gateways 14:00 Value Architecture That Books 20:17 How the Groups Converge 21:39 Events Announcements 22:38 Three Practical Takeaways 23:47 Wrap Up and Call to Action #Tokenomics #AIEconomics

I started to cringe when he said "Token cost - is token cost the right thing to talk about? Absolutely." I disagree that token cost is the thing to focus on... wrong metric, not consistent between models, not tied to value, etc…. but he does recover it with "after Tokens you start to move more toward OUTCOMES." Yes outcome is a vague thing but moving toward accomplishing a unit of work is the way to go, regardless of how many tokens it takes to get there. A subtle but important nuance to moving away from counting tokens or even token budgets, which are only a part of the equation. Token costs are the tip of the AI Tokenomics iceberg, there many more costs to consider and cost per outcome is a start to the way it needs to be measured.

Don't count tokens... #tokenomics 101. Episode #15. Tokenomics Goes Official. Defining AI Economics, Value, Routing Layers & the Per-Watt Economy. 00:00 Foundation Goes Official 00:34 Governance and Definition 02:11 Not Counting Tokens 03:48 Summit and Survey 04:47 Working Groups Kickoff 05:07 Measuring AI Value 07:13 Deflection and Baselines 09:31 Consumption Stack and Big T 11:27 Defining Tokenomics Scope 13:45 Energy and Per Watt Metrics 18:34 Industry Headlines 22:16 Academics Join In 25:03 Key Takeaways and Next Steps

Many who know of my son's death have been afraid to ask what happened. Others who don't have wondered why I've disappeared. Here is the story: linkedin.com/pulse/its-later…

FinOps Foundation Soars to 300 Members and Introduces New Partner Tiers for Cloud Service Providers and Vendors - Welcome Ancestry, Intuit, Letgo, Neustar and Pearson Education businesswire.com/news/home/2…

Great to be quoted by @jpruth from @infoweekonline in his article on cloud spend: informationweek.com/cloud/ti… #aws #GCP #Azure