WillItSurvive
AI companies and labs · subsidiary of CoreWeave, Inc. (Nasdaq: CRWV) · United States

Weights & Biases

ML experiment tracking and management platform, now a CoreWeave subsidiary.

Defensible for now
Assessed 20 July 2026 · Evidence as of 20 July 2026
Real protection, with named pressure against it.
On the recordSince 20 July 2026 · Checkable from Q3 2026 · Every call on the record Checkpoint: CoreWeave earnings call where W&B revenue/retention may first be broken out
The call

W&B's tracking layer is wired into how ML teams work, but its future now rides on CoreWeave's stock and GPU business, not its own numbers.

Weights & Biases is software that records what happens when engineers train artificial intelligence models, so they can compare results and learn what works. Rivals like Databricks offer free experiment tracking, and AI agents could eventually do some of this work automatically, but W&B has built a big library of real experiment data and just shipped an AI agent of its own to stay ahead. The catch: W&B was bought by CoreWeave in May 2025, so your investment is really in CoreWeave stock, not W&B directly. That means the company is now tied to GPU sales and the big cloud customers CoreWeave depends on, which creates new risks if those customers shrink or switch to competitors like Amazon or Google.

Why
01
AI pressure: AI coding agents (Cursor, GitHub Copilot) that auto-build in-house experiment logging. Agents that write and manage training code could let teams skip W&B's dashboards and log metrics themselves. Could AI do the core job? Unlikely.
02
Bigger players: Databricks (MLflow, free and bundled into its platform). MLflow ships free with Databricks, undercutting W&B on price for teams already on that cloud. Could a giant copy it? Probably.
03
What protects it. The wandb SDK sits inside training loops at hundreds of thousands of companies, and switching means rewriting pipelines and losing years of run history.
What would change this call
CoreWeave earnings call discloses W&B-specific revenue or retention figures. Shows whether the acquired unit is growing or stagnating post-deal.
A named enterprise customer publicly switches from W&B to MLflow or a rival tracker. Would confirm real customer loss rather than just competitive noise.
Reports of AWS or GCP-native customers reducing W&B usage citing CoreWeave ties. Would show neutrality loss is actually costing W&B non-CoreWeave customers.
Key numbers
  • ~1B experiment runs tracked on W&B (2026-06-29)
  • CoreWeave paid ~$1.4-1.7B for W&B (2025-05-05)
  • CoreWeave (CRWV) stock ~$73/share (2026-07-20)
Named risks
  • Employee and investor equity now tied to CoreWeave (CRWV) stock, exposed to GPU capex and debt strain. Severity: high
  • Loss of cloud-neutral positioning post-acquisition could push AWS/GCP-native customers to churn. Severity: medium
  • Core experiment-tracking tier commoditized by free MLflow, pressuring pricing and retention. Severity: medium
  • No standalone W&B revenue reporting since May 2025, making financial health hard to verify externally. Severity: medium
Rivals, as read in July 2026
RivalDirectionEvidence
MLflow (Databricks)gaining onFree open-source tracking pressuring W&B's low-end tier as of 2026
DatabrickscontestingBundles MLflow as loss-leader within broader cloud ML platform, 2026
Comet ML / Neptune.aicontestingListed as direct experiment-tracking alternatives in 2026 comparison pieces
Langfuse / Braintrust / FutureAGIgaining onNewer LLM-observability entrants cited as positioning aggressively against W&B, 2026
Questions people ask
Is Weights & Biases still an independent company?
No. CoreWeave (Nasdaq: CRWV) completed its acquisition of Weights & Biases on May 5, 2025, making it a wholly-owned subsidiary now operating as Weights & Biases, a CoreWeave company.
Is Weights & Biases profitable?
Unclear and unverifiable externally: W&B's revenue is folded into CoreWeave's consolidated results and no longer reported standalone. CoreWeave guided $4.9-5.1B total 2025 revenue as of June 12, 2025, but that figure covers all of CoreWeave, not W&B specifically.
Could AI replace Weights & Biases?
Partly the opposite: CoreWeave's ARIA agent, in public preview since June 29, 2026, automates the manual chart-comparison and analysis work W&B users used to do by hand, but ARIA itself runs on W&B's own telemetry, turning the automation threat into a W&B product feature.
Who is Weights & Biases' biggest competitor?
MLflow, the free open-source tracking tool backed by Databricks, is the main pressure point at the low end of the market as of 2026, since it undercuts W&B's basic experiment-tracking tier on price.
Who owns Weights & Biases now?
CoreWeave, Inc. (Nasdaq: CRWV) owns Weights & Biases outright, following a deal announced March 4, 2025 and closed May 5, 2025, reported at roughly $1.4 billion by SiliconAngle.
How big is Weights & Biases' user base?
Weights & Biases reported 700,000+ ML practitioner users as of August 9, 2023, up from 100,000 at its 2021 Series C, and by June 29, 2026 CoreWeave said the platform had tracked nearly one billion experiment runs and trillions of metrics.
Timeline
  • 2026-07-20CoreWeave (CRWV) stock trading near $73/share
  • 2026-06-29CoreWeave launches ARIA autonomous research agent in public preview, built on W&B Weave
  • 2025-06-18W&B Inference launches, serverless open-source LLM access via CoreWeave GPU infrastructure
  • 2025-05-05CoreWeave completes acquisition, W&B becomes wholly-owned subsidiary
  • 2025-03-04CoreWeave announces deal to acquire Weights & Biases
  • 2023-08-09W&B raises $50M at $1.25B valuation, launches W&B Prompts, reports 700,000+ users
  • 2021-10-13W&B raises Series C, $100M+ at $1B valuation
Revision history
20 July 2026
First assessed. Defensible for now. The text above is fixed as of this date and is not edited afterwards.
In the same band
Sources