Sponsored by
TOGETHER WITH TODAY'S PARTNER
Smarter CRM. Less Busywork.
Disconnected data and tools make it harder to understand your customers. HubSpot's Agentic Customer Platform brings your data, teams, and tech stack together with AI built in to help your business work faster and create more personalized customer experiences.
Why HubSpot and what's new
Use AI powered tools to take action faster
Unify your data, teams, and tech stack in one place
Create one shared view of customer data
Connect teams around the same customer context
Bring your business tools into one place
Connect more of your business in one place and give every team a smarter way to work. Get set up quickly and start checking off your hardest tasks.
Good morning, {{first_name|there}}. The hub that 3 million open models live on just hired a bank to find a buyer.
Read time: 3 minutes. Same time, every weekday — rate today's issue at the bottom.
🚀 The Big Story: Hugging Face hired a bank to sell itself at $13B
Business Insider reported Sunday that Hugging Face is exploring a sale valuing it at $13B or more — nearly 3x its $4.5B mark from 2023 — and has brought in a bank to sound out bidders.
The asset is the hub, not a model: 3M+ public models and 1M+ datasets. Whoever buys it owns the default distribution layer for open-weight AI.
No bidder has been named: talks are early. The 2023 round put Salesforce, Nvidia, Google, Amazon, Intel, Qualcomm, and IBM on the cap table — every one of them is a plausible buyer.
It didn't have to sell: roughly $400M raised, about half of it still unspent as of November 2025. Revenue has never been disclosed.
The CEO called the bubble himself: Clément Delangue warned in November 2025 that the LLM bubble could burst in 2026. Selling into strength is consistent with that view.
Jason's take: Every open-source workflow you own has a single point of failure, and it's a URL. The weights are free; the hub is not — and a buyer with cloud capacity to sell has an obvious reason to make hosting there a little less neutral. Mirror the three or four models your product actually depends on this week. Not because a sale is bad, but because from_pretrained is a dependency on someone else's balance sheet.
⚡ Quick Hits
Alibaba raised $10.2B in a single day. Hong Kong's largest-ever follow-on offering, earmarked for AI capacity. Chinese compute is no longer capital-constrained.
OpenAI cut GPT-5.6 Sol pricing by more than 20%. Now $4 in / $20 out per million tokens through November 21. If you shelved a feature as too expensive this spring, re-run the math.
LinkedIn's AI-slop button passed 1M clicks. Views of flagged content fell 40%. Readers now have a working filter, and it isn't a vibe — it's a number.
Harvard is selling a $699 bootcamp taught by AI clones of its faculty. Founders pitch avatars of real professors and run mock board meetings. Credentialed expertise just became a licensable asset.
OpenAI reversed itself on California's SB 53. After a model-escape incident it's now urging stronger rules. The compliance work you deferred just got a shorter runway.
A humanoid robot beat Usain Bolt's 100m record. At Beijing's World Humanoid Robot Games on Saturday. Embodied AI is past the falling-over-on-camera stage.
📡 Trending on X
"Who should be allowed to own Hugging Face?" The timeline split within an hour between "any hyperscaler buyer kills neutrality" and "it's a hosting bill — somebody has to pay it."
Qwen 3.8 27B reverse-engineered a licensed app in 30 minutes. The XDA writeup hit the front page and reframed the argument: frontier models may already be overkill for most real work.
That 40% figure is doing the arguing. LinkedIn's slop-flag data gets posted as proof readers can tell — and as proof the flag is just taste policing.
Robot-games clips are everywhere, with an asterisk. Half the replies celebrate the record; the other half point out it's a purpose-built sprinter, not a general machine.
📺 Trending on YouTube
🛠 The Workflow: Publish so the slop filter skips you
Pew put 35% of post-ChatGPT pages as showing signs of AI authorship, and LinkedIn's flag now cuts views by 40%. Readers aren't guessing anymore — they're pattern-matching.
Put one unrepeatable thing in the first 100 words. A number you pulled, a call you sat on, a thing that broke. A model can't fake first-hand.
Kill the three tells. "In today's fast-paced world," adjectives in threes, and the hedge-both-sides closing line. Search your last five posts for all three.
Name the date and the source in-line. Specific attribution reads human because slop generalizes.
Take one position that costs you something. A recommendation you'd defend in the replies is the strongest anti-slop signal there is.
Read it out loud before you publish. Anything you'd never actually say is the exact part the filter catches.
Reply with the word "TELLS" and I'll send you the 12-phrase slop checklist I run every draft through.
🧰 Trending Tools
Cursor Origin — code hosting with pull requests and two-way GitHub sync. For teams whose agents now open more PRs than humans can review.
Google Antigravity 2.0 — drive your coding agents from any browser, on any device. For operators who want work running while they're away from the desk.
Runway Ruby — converts standard video to HDR, out to ProRes and EXR. For creators who can resell an archive as a premium deliverable.
FLUX Upscale — regenerates 480p footage to native 4K. For anyone sitting on old footage a client would pay to reuse.
Ox Alpha — stealth 1M-context multimodal model on OpenRouter with zero data retention. For teams that can't route client data to a named vendor.
PRDs by voice. Bug reports by voice. Ship faster.
Dictate acceptance criteria and reproductions inside Cursor or Warp. Wispr Flow auto-tags file names, preserves syntax, and gives you paste-ready text in seconds. 4x faster than typing.
📣 Put your brand here. The AI Innovator reaches AI-first operators, creators, and marketers every weekday. Primary sponsorships are now booking.
That's a wrap
Tomorrow: what Alibaba's $10.2B actually buys — and why the cheapest capable model in your stack may soon be Chinese.
Enjoying the new format? Share your link — every referral keeps this free:
How was today’s email?
You're reading the 3-minute AI Innovator — same time, every weekday. Hit reply and tell me what you want more of. I read every one.
