Vitals: Founded 2019, Peak valuation ≈ $1 B (Oct 2022)

Funding ≈ $125 M, Burn ≈ $8 M / month vs ≈ $5 M revenue (Feb 2024)

Mar 2024 CEO exit → Apr 2024 30% layoffs, “strategic options” in progress

What Went Right

  1. Stable Diffusion open-sourced → put Gen-AI in everyone’s browser.
  2. $101 M seed from Coatue & Lightspeed = instant unicorn status.

Where the Blood Flow Stopped

  1. GPU bill ≫ business model – $99 M annual compute vs low-single-digit revenue.
  2. Copyright crossfire – Getty lawsuit seeks up to $1.7 B.
  3. Leadership churn – founder resignation dented investor confidence.
  4. Enterprise trust gap – legal fog + hallucinations stalled big-ticket deals.

Prescription

✅ Stage burn to revenue curve—rent GPUs only as ARR grows.

✅ Bake licensing & compliance in early to avoid lawsuits.

✅ Pair long-cycle R&D with quick-cash products (fine-tunes, SaaS wrappers).

One Take-away

“Build a bigger engine only after you have the fuel money.”

Do it this week:

  1. List your costliest input (cloud, talent, ads).
  2. Link any cost jump to a revenue milestone.
  3. If revenue isn’t there yet, pause upgrade—or co-fund with a partner.

Future Scenarios – “Company is restructuring, not dead”:

Acquire-and-integrate: Cloud giant buys weights + talent, sells via API. Example: MosaicML → Databricks

Open-core pivot Free community model plus paid “Enterprise Diffusion.” Example: Red Hat, HashiCorp

GPU co-op Consortium splits compute costs; rev-shares inference. Example: Hugging Face cluster

Which unicorn should we slice open next Sunday? Drop your pick below 👇

Disclaimer: This analysis is based on public info and my own interpretation; facts may evolve and any errors are mine alone.