True Ventures @madelineminshew says the era of tokenmaxxing is over: spending $10K a day used to be cool. Now it's "what are you doing?" "Cost is a really big thing. For a long time, token maxing was super hot, and it was cool to say, I'm spending 10K a day. And now it's like, what are you doing?" "Sometimes it's definitely worthwhile to have that spend. If you are a small engineering team and you're thinking about your token cost as an employee cost, or if you are tackling a project that needs that, great." "But oftentimes, the hardest work that needs to be done once to prove something or to build something, you can then switch to an open source model after that." "Uber's CTO has spent a lot of time talking about how they're handling this: you do some of the hardest work on the frontier models, and after that, there's a real big push across their entire portfolio right now to find good open source alternatives." @trueventures @tonysphere

Breaking: Michael Burry says AI's biggest customers are "tokenmaxxing" The pyramid of customers who become competitors: Tier 1: frontier model builders: OpenAI, Anthropic, Google, Microsoft, Meta, xAI. 7 to 10 companies Tier 2: vertical LLM builders: Salesforce, ServiceNow, SAP, Bloomberg, IBM. 30 to 50 companies Tier 3: Fortune 500 building proprietary AI: JPMorgan, UnitedHealth, Walmart, CVS. 200 to 500 companies Tier 4: true pure consumers: mid-market companies and traditional manufacturers. Fewer companies remain as they climb the pyramid, per Burry's SubStack

[DEMO] Shift your organization from the old world of “tokenmaxxing” to the new era of “valuemaxxing” with Unity AI Gateway. At #DataAISummit#, Databricks Software Engineer @ankit_math showed how real-time budget policies can guide developers toward more token-efficient models, Omnigent can route work based on task complexity and quality needs, and centralized agent tracing in Unity Catalog can help teams govern usage and stay in control. Watch the full demo: