🕶️ Disney got a unicorn's scraps for $20M

Good morning, founders 🕶️

Hope your coffee is still at the introductory price.

Every day-one price comes with a due date. Google's new flagship model costs half until November 30, then doubles. Adobe's $60 a month felt normal until a small team used Claude to build seven free look-alikes. And Character.AI, a unicorn in 2022, just filed for Chapter 11, with Disney paying $20 million for part of what's left. The launch price is a loan. Somebody repays it later: the customer, the vendor or the investors. So which of your prices, paid or charged, is running up interest right now?

In today's newsletter, we'll get into:

Connect your customer data, teams, and tools without the hassle of complex integrations or lengthy setups. One easy platform gives marketing, sales, and service teams a unified customer view and the tools to turn it into growth.

Why HubSpot and what's new
• Generate leads and automate marketing with Marketing Hub
• Build your pipeline and close more deals with Sales Hub
• Scale customer support and drive retention with Service Hub
• Keep customer data clean, connected, and actionable with one, unified platform

Join 306,000+ in over 135 countries using HubSpot to grow their businesses.

See what a more connected approach to growth can do for you and your team. Get setup quickly and start checking off your hardest tasks.

Get started free

Most consumer apps die because nobody uses them. Character.AI is going under even though people used it for hours a day.

The startup filed for Chapter 11 on Tuesday, listing assets and liabilities of $50–100 million. Disney agreed to pay $20 million for some of its assets plus a non-exclusive, perpetual license to its tech. There's no bidder yet for the rest. That check is roughly a tenth of the $193 million+ the company raised, per PitchBook, and it comes four years after the startup hit unicorn status.

The engagement trap. In 2022, users averaged more than 2 hours a day. In most apps that's the dream. In AI, every hour of chat burns compute, so the most loyal users were also the most expensive ones.

The brain drain. In August 2024, Google paid $2.7 billion to license the models, and founders Noam Shazeer and Daniel De Freitas went back to Google. Character.AI stopped training its own models and chased growth on a product it had stopped pushing forward.

The lawsuits. After a suit tied to the suicide of a 14-year-old user, plus a Texas investigation, the company settled some cases, added safety features and banned minors.

The pivots. Video generation, social feeds, microdramas. Then last month, former CEO Karandeep Singh Anand and staff left for Disney, the eventual buyer.

Run the stress test on yours. Remove the model, which anyone can now rent. Remove the founders, who can leave as part of a licensing deal. What's left is users and a brand. Disney priced that at less than 1% of what Google paid. If "the users" is your answer, make sure each one pays more than they cost to serve.

The deadline to find a buyer for the rest is late November or early December. No bid, and the platform gets wound down.

How much should you actually own of your startup?

Most founders shed more equity than they realize by the time they hit Seed or Series A and it could cost them millions down the line. In this video, we break down founder equity benchmarks, cap table traps, and why owning 5% of a unicorn might still beat 50% of a “meh” business.

Google just put its new flagship model on sale, and the sale has an end date.

Gemini 4 Argon hit the paid Gemini API on Tuesday at $2 per million input tokens and $10 per million output, the same as OpenAI's GPT-6.1 Sol. That price runs through November 30. On December 1 it doubles to $4/$20, matching Anthropic's Claude Opus 5.5. Cached input is 95% off, and Google has only promised that discount for the intro window.

Think of it as an adjustable-rate mortgage that resets after seven weeks instead of five years. Google isn't after seven weeks of revenue. It wants what you build while it's cheap. Your team wires up the gemini-4-argon endpoint, tunes prompts around its quirks and gets your evals passing on it. None of that work moves to another model for free.

Argon's quirks are real. Its 1 million-token budget is shared between input and output, so an 800,000-token prompt leaves only 200,000 for the answer. Claude Opus 5.5 caps output separately, at 128,000. A prompt built for one design doesn't drop cleanly into the other.

Say your product runs 500 million input and 100 million output tokens a month. That's $2,000 in November and $4,000 in December, before cache pricing changes anything. If you priced your plans off the promo bill, your gross margin shrinks on a date Google picked.

The takeaway for you: model your unit economics at $4/$20 today. Keep a second model passing your evals, and route traffic through a layer you control, so switching is a config change, not a sprint.

Watch December 1. Whether Google extends the promo or a rival undercuts it first will show who needed your tokens more.

Scale your growth: Join names like HubSpot, Ahrefs, and Deel. Put your product in front of 85,000+ startup founders. Let's talk.