Costs hit $34B, with R&D at $19.18B and an estimated $17B of it Azure compute, per the financials Ed Zitron published. A roughly 3-to-1 loss-to-revenue ratio weeks before the S-1 makes the IPO math brutal, and the Azure line means Microsoft holds the whip hand over OpenAI's cost base, not just its GPUs. Our read.

Filed June 15 in NDCA, the suit alleges the $200/mo Max 20x plan returns 6-8x the Pro baseline and the $100/mo Max 5x returns roughly 3.5x, per Engadget. It is the third subscriber-side AI suit in 18 months, and it puts every provider's "Nx" multiplier language on legal notice. Our take.

GA landed June 15 for Creator and Creator Pro plans, exposing 68 read/write tools across subscribers, tags, broadcasts, and sequences, with deletes gated behind explicit AI-client confirmation. Run Claude Code or Cursor and you can now query and edit your list in plain language, no dashboard hop. Details, or try Kit.

Defense Secretary Hegseth posted that the DoD "kicked Anthropic out" months ago, and CBS confirmed a Pentagon supply-chain-risk designation barring contractors from the company. Anthropic now sits between two contradictory federal positions: too capable to export, yet too untrusted for defense. Both cannot be true. Our take.

Andrew Bosworth circulated the memo through Meta's 6,500-person Applied AI unit after engineers reported reassignment to data-labeling work instead of product roles. Morale cracks at that headcount read as real organizational stress on Meta's flagship internal-AI bet, not PR noise. Read it.

The free plan covers 10k subscribers; Creator at $33/mo (per Kit) unlocks the Craft and Commerce features that matter once your list is actually growing. Our pick for who upgrades and when. Read it, or try Kit.

Scale is $43/mo (per beehiiv) and that is where the platform stops being a publishing tool and starts paying for itself. We worked the subscriber math so you know the break-even list size before you commit. Read it, or try beehiiv.

Firecrawl ships 15 installable skills; most teams need four. We name which four and what each costs in credits so you do not burn budget testing the rest. Read it, or try Firecrawl.

Claude and ChatGPT cite the same source only 8% of the time, per Search Engine Land, so chasing both at once is a trap. We show why Brave Search rank moves Claude visibility and how to work it. Read it.