Scale AI
Scale AI builds data annotation pipelines, model-evaluation platforms, and government AI decision-support tools for defense and enterprise customers.
Background
Scale AI was founded in San Francisco in 2016 by Alexandr Wang and Lucy Guo to solve the data-labeling bottleneck blocking production AI deployments.
Wang, who became sole CEO after Guo's departure, built the company into the dominant third-party data annotation and model-evaluation platform; Jason Droege is reported as current CEO following organizational changes tied to Meta's 2025 investment.
Scale's core product stack combines automated labeling software with managed human annotators to produce training datasets for image, text, voice, video, and document AI.
Its government-facing layer includes Scale Donovan, a decision-support platform for intelligence analysis and mission operations, and Defense Llama, a jointly developed Scale-Meta defense-oriented large language model.
Scale reaches DoD customers primarily through direct prime contracts and other transaction authority agreements, and through Thunderforge, a DoD AI-agent program for military planning in which Scale holds the prime contract with Anduril and Microsoft as partners. The company employed approximately 1,200 people as of 2025.
Linked records
Wired to the rest of BattlePolicy — the systems it builds in the Lexicon and the contracts it wins in the Procurement Tracker. Every link below is a record in our own datasets.
- none linked yet
- none linked yet
Financial footing
- ✗ "$15.9B total" WRONG → ~$1.6B raised in rounds
- Meta bought 49% for $14.3B at $29B val (Jun 2025)
- Backers: Founders Fund, Accel
Scale AI raised a $1B Series F in May 2024 at a $13.8B valuation, led by Accel, with participation from NVIDIA, Amazon, Founders Fund, Coatue, Tiger Global, Greenoaks, Thrive Capital, Spark Capital, and Wellington Management. Total venture funding across all rounds reached approximately $1.6B prior to the Meta transaction. In June 2025, Meta acquired a 49% non-voting stake for approximately $14.3B, implying a post-deal valuation of roughly $29B.
The Meta transaction is structured as a stake purchase, not a conventional funding round, and Scale remains a private company. Earlier backers include Y Combinator, Dragoneer, Index Ventures, and D Capital. Revenue figures have not been publicly disclosed; the company's backlog is partially reflected in disclosed government contract ceilings noted below.
Major customers & contracts
The U.S. Department of Defense selected Scale AI in February 2024 under a one-year contract to test and evaluate large language models for military purposes. The U.S. Army awarded Scale a $99 million R&D contract. The Chief Digital and Artificial Intelligence Office awarded Scale a $100 million, five-year other transaction authority agreement to deliver AI tools on Top Secret and SCI networks.
Scale won the prime contract for Thunderforge in March 2025, a DoD AI-agent program for military planning, with Anduril and Microsoft serving as partners. Reuters confirmed the U.S. Air Force as a Scale customer. Scale Donovan is deployed for intelligence analysis and mission operations support across DoD agencies, per public reporting.
Leadership & contact surface
Leadership: Alexandr Wang outdated — → Meta ~Jun 2025 with its 49% stake; verify current CEO (Jason Droege)
Official site: https://scale.com
Competitive position
Against Palantir, Scale competes directly in government AI platforms and decision-support tooling; Palantir's AIP and Gotham are deeply embedded across combatant commands, giving it an entrenched data-network advantage, while Scale's edge is superior training-data infrastructure and the Meta-backed Defense Llama model. Against Labelbox and Appen, Scale holds a clear moat in enterprise and government scale, annotation quality, and its pivot from pure labeling into full AI data engines and model evaluation; Appen's financial difficulties have reduced its competitive pressure.
Against Microsoft Azure AI and Amazon SageMaker Ground Truth, Scale is a workflow layer that can run on either cloud rather than a direct infrastructure rival, but both hyperscalers can commoditize annotation tooling over time. Against Anduril, Scale is a partner on Thunderforge rather than a pure competitor, though both vie for prime DoD AI contracts. The market for government AI data infrastructure remains fragmented, with no single dominant vendor.
Battlefield relevance
No confirmed combat use to date.
Sources & verification
- Reuters — Scale AI clients include DoD and U.S. Air Force
- Fortune — Scale AI government and military expansion
- https://www.reuters.com/article/business/artificial-intelligence-firm-scale-raises-325-million-at-7-billion-valuation-idUSKBN2C02QM/
- https://economictimes.indiatimes.com/tech/technology/scale-ai-valued-at-14-billion-in-nvidia-amazon-backed-funding-round/articleshow/110308383.cms
- https://scale.com/blog/scale-ai-series-f
- https://finance.yahoo.com/news/scale-ai-hits-3-5b-213539913.html?guccounter=1
- https://www.theinformation.com/articles/ai-startup-scale-offered-funding-at-3-billion-valuation
- https://www.wsj.com/tech/ai/scale-ai-gets-meta-investment-that-values-it-at-more-than-29-billion-abeefc2f
- https://news.crunchbase.com/ai/scale-holistic-raise-big-accel-nvda-amzn/
- https://www.linkedin.com/posts/cviernickel_scales-series-f-expanding-the-data-foundry-activity-7198716425618767872-jLrV
- https://tsginvest.com/scale-ai/
- https://sacra.com/c/scale-ai/
- https://www.youtube.com/watch?v=Ag7a_SalbvQ
- https://fueler.io/blog/scale-ai-usage-revenue-valuation-growth-statistics
- https://fortune.com/2024/05/21/scale-ai-funding-valuation-ceo-alexandr-wang-profitability/
- https://en.wikipedia.org/wiki/Scale_AI
- https://www.intelcapital.com/scale-ai-raises-1-billion-series-f-to-push-the-frontier-of-ai-data/