Public link to this news item

Frontier Labs

Scope (new information surfaced): Dec 12–Dec 16, 2025 (inclusive)

Acquisitions (focus, approach, competitive advantages)

  • NVIDIA acquires SchedMD (maker/maintainer of Slurm) — announced Dec 15, 2025
    • What NVIDIA decided
      • NVIDIA said it acquired SchedMD and will keep distributing SchedMD’s software as open source (financial terms not disclosed). (reuters.com)
    • What the acquired company does (focus + approach)
      • SchedMD is the commercial entity most associated with Slurm, a widely used open-source workload manager / job scheduler for clusters and supercomputers (queueing jobs, allocating resources, monitoring execution). (reuters.com)
      • Their business model (as described by Reuters and SchedMD’s own materials) is effectively open-source core + paid engineering/support/services, positioned around running Slurm at scale in HPC / AI cluster environments. (reuters.com)
    • Competitive advantages / “why this asset matters”
      • Entrenched standard in high-end compute scheduling: Slurm is broadly used in HPC, and is described as the workload manager on a large share of top supercomputers (commonly cited around the TOP500). (en.wikipedia.org)
      • Direct leverage on frontier-model training operations: Reuters explicitly frames Slurm as “critical infrastructure” used by foundation model developers to manage training and inference workloads. (reuters.com)
      • Reinforces NVIDIA’s full-stack moat (hardware → software → operations): NVIDIA highlights that software (CUDA + ecosystem tooling) is central to sustaining its AI dominance; adding control over a core scheduler layer can tighten integration across NVIDIA-accelerated data centers. (reuters.com)
    • Unique details that help triangulate impact
      • SchedMD is relatively small (~40 employees, founded by Slurm developers Morris “Moe” Jette and Danny Auble). (reuters.com)
      • Reuters lists customers including CoreWeave and Barcelona Supercomputing Center, signaling relevance to both AI cloud infrastructure and elite HPC. (reuters.com)

Notable hires / leadership changes

  • OpenAI appoints Albert Lee as VP of Corporate Development — confirmed Dec 15, 2025
    • What OpenAI decided
      • OpenAI confirmed it appointed Albert Lee (from Google) as Vice President of Corporate Development, starting Tuesday (Dec 16, 2025), reporting to CFO Sarah Friar. (reuters.com)
    • Why it’s notable
      • This is a deal-making / strategy hire: Reuters notes Lee led corp dev for Google Cloud + DeepMind and worked on 60+ transactions totaling $50B+ (per his LinkedIn). (reuters.com)
      • Practically, this can be read as OpenAI strengthening M&A + partnerships capacity during a period of heavy infrastructure buildout and competitive pressure.
  • OpenAI Chief Communications Officer Hannah Wong to depart — reported Dec 16, 2025
    • What OpenAI decided
      • Axios reports OpenAI’s first CCO, Hannah Wong, will step down at the end of January 2026; Lindsey Held Bolton will serve as interim comms lead (reporting to CMO Kate Rouch). (axios.com)
    • Why it matters
      • While not a “frontier-model” technical hire, this is a high-signal executive change given OpenAI’s regulatory scrutiny, lawsuits, and major product/infrastructure narratives (where comms leadership materially affects risk posture). (axios.com)

Projects stopped / paused / materially delayed

  • Mattel–OpenAI collaboration: no product launch in 2025 (delay) — confirmed Dec 15, 2025
    • What changed
      • Axios reports Mattel’s collaboration with OpenAI “won’t emerge this year” (i.e., no 2025 release, despite earlier expectations). (axios.com)
    • Stated/reported drivers
      • Axios attributes the delay to heightened concerns/scrutiny around generative AI interacting with teens and broader concerns about AI-integrated toys after issues with similar products. (axios.com)
    • How to interpret against your “stopped projects” filter
      • This appears to be a timeline slip / postponement, not a confirmed cancellation. No clear public replacement roadmap or termination language was reported in the Axios item. (axios.com)
  • What I did not find in this 4-day window (Dec 12–16)
    • No well-sourced reports of Anthropic / Google DeepMind / Meta / xAI fully cancelling a named frontier-model training run or shutting down a flagship frontier-model program in this exact window, based on the news queries run.

(Optional) Other frontier-AI decisions surfaced in the same 4-day window (not in your 3 requested buckets)

  • NVIDIA launches “Nemotron 3” open-source model family — reported Dec 15, 2025
    • Reuters describes a new generation of NVIDIA open-source models, with staggered releases (small model now; larger versions early 2026), positioned against increased competition from Chinese open-source offerings. (reuters.com)
  • xAI partnership: El Salvador + Grok in public schools — reported Dec 12, 2025
    • AP reports El Salvador will partner with xAI to bring Grok-based tutoring into 5,000+ public schools (over 1 million students, per the report). (apnews.com)
Published on December 16, 2025