After Being Laid Off, Tian Yuandong’s New Offer Lands in France — Turns Out He Was Dropped Right After Training Llama 4.5
Silicon Valley: Fast at Innovation — and Faster at Layoffs
Silicon Valley’s pace applies to everything — even job cuts.
Quantum Bit uncovered a vivid example at Meta: minutes after senior engineer Tian Yuandong tweeted he’d been laid off, all his internal accesses were revoked. The move, attributed to Alexander Wang, was swift and decisive.
---
The Controversial “One-Sweep” FAIR Layoff
The current layoff round has sparked heated debate.
Tian Yuandong, a Meta veteran of over ten years, along with his entire FAIR team, were cut in one swoop.
Speculated Reasons
- Factional alignment: FAIR (Tian’s team) reportedly not in Alexander Wang’s camp.
- Performance blame: Leaks claim 600 layoffs tied to the team’s inability to deliver a reasoning model based on Llama 3, especially after competitor DeepSeek’s success.
- Llama 4’s underperformance allegedly needed a scapegoat — Tian’s team was targeted.
---
Tian Yuandong’s Response
Tian refuted the rumours directly:
> The people you mentioned weren’t even laid off.
> Since January, our team dropped all ongoing research and was moved to GenAl for post-training bug work on Llama 4.
> After release, we hunted and fixed bugs — then got laid off.
AI luminary LeCun also clarified he had minimal involvement with Llama development:
---
Layoff Timing: “Donkey Slaughter After the Millstone”
Insiders say layoffs were deliberately timed:
- Llama 4.5 was trained before cuts.
- Meta did not announce affected teams internally.
- Employees discovered departures only through public “goodbye” posts or private comparisons.
---
Impacted Teams & Regions
- ~600 affected employees — exact breakdown undisclosed.
- FAIR London likely on the chopping block next.
- EU layoff details unclear.
TeamBlind charts show job role proportions:
---
Alexander Wang’s Internal Memo
> Smaller teams = fewer conversations, wider scope, bigger impact.
Voting among 47 former employees suggests product application research and cross-functional roles also hit hard.
---
Why the Delay?
- Layoff list finalized in June/July.
- Delay ensured teams completed Llama 4.5 training.
- Now results can be handed off, 600 with merit removed in one stroke.
---
Exit Details
- Official departure: November 21, 2025.
- Non-working notice period: salaries retained, systems access revoked.
- Severance: 16 weeks base + 2 weeks/year worked − notice period.
(PS: Severance is generally a corporate policy gesture in the U.S., not a statutory obligation.)
---
Lingering Risk for Remaining Employees
If you’re not in the TBD department, expect:
- Continued heavy workloads until value extracted.
- Possible February layoff wave.
Meta’s “super lab” AI initiative under Alexander Wang has produced chaos more than calm.
---
Compensation & Opportunities
Tian Yuandong confirmed 8 months’ salary in severance.
His “Reach out :)” post drew recruitment offers from:
- OpenAI
- xAI
- Google DeepMind
- NVIDIA
- Anthropic
…and numerous startups:
High-profile recruiters like NVIDIA’s Jim Fan and Ma Tengyu also joined in:
> Consensus: This is Meta’s loss.
---
Cultural Reactions
- Poetic analogies invoked Li Bai’s verse for demoted Wang Changling.
- Xie Saining remarked: “All good things must end — may better chapters follow.”
---
Meta’s Next Restructuring: Risk Control Department
One day later, Meta announced dismantling its Risk Control unit.
Reason: transition to AI-based compliance review.
> “Compliance results will be more accurate and reliable — AI can handle it.”
---
Open Platforms as a Safety Net
Events like these prompt tech professionals to diversify and build independent channels.
Platforms such as AiToEarn官网 offer:
- AI content creation & monetization
- Cross-platform publishing (Douyin, Kwai, Bilibili, Facebook, Instagram, LinkedIn, YouTube, Pinterest, X)
- Analytics & model rankings AI模型排名
---
Meta Prepares for AI-Driven Org Shifts
Insider & Social Media Reports
- Post 1: All risk roles cut; duties moved to ML models for fraud detection & regulation.
- Post 2: Notices issued, multi-phase transition planned.
Business Insider Coverage
Move aligns with cost-cutting and deeper AI integration — especially for back-end operations.
Financial Context
Heavy infrastructure investments, particularly Blue Owl data centers, may enable such AI replacements.
---
Broader Implications
- Job displacement — net loss or redeployment?
- Governance — Can AI match human judgment in nuanced compliance cases?
- Transparency — Stakeholder trust in AI-only risk decisions?
---
The Creator “Parallel”
Corporate AI expansion mirrors trends in the creative industry — where open-source AI tools like
AiToEarn (博客, 开源地址) enable creators to scale and monetize content across Douyin, Kwai, WeChat, Bilibili, Rednote, Facebook, Instagram, YouTube, Pinterest, LinkedIn, Threads, and X.
---
In Summary:
Meta’s restructuring reflects a wider shift — AI as the operational backbone in both corporate and independent fields. The key challenge: making these systems efficient, ethical, and trusted in complex human contexts.
