Scale AI navigates defense windfalls and autonomous vehicle shifts

Defense contracts replace lost tech giants

The Department of War’s Chief Digital and Artificial Intelligence Office increased the potential value of its agreement with Scale AI from $100 million to $500 million. This five-fold increase follows the rapid uptake of the Production Other Transaction Authority agreement since its original award in September 2025. Scale AI provides the Scale Data Engine, Scale GenAI Platform, and Scale Donovan to support the Department. The Department names the program Thunderforge, which integrates AI agents into workflows for troop movements, supply chains, and mission timing. Microsoft supplies the AI models for these systems, while Anduril provides war-gaming and model tools. This expansion follows a period of significant client defection triggered by Meta’s $14.3 billion acquisition of a 49% stake in Scale AI in June 2025. After that deal, Google severed its $200 million annual contract, and Microsoft, xAI, and OpenAI followed. These companies abandoned the platform because they could not ensure that proprietary training data stayed away from Meta’s competing AI teams. Scale AI employs approximately 240,000 people through its Remotasks and Outlier subsidiaries to handle these massive datasets. The Pentagon’s expansion of the agreement allows any Department of War component to route funding to Scale AI for computer vision, generative AI decision-support, and data operations without ever requiring a new competitive solicitation. The program rolls out at U.S. Indo-Pacific Command and U.S. European Command to track China and Russia. Scale’s annotators perform defense-domain tasks like labeling EO/IR and SAR imagery for autonomy programs. This work helps generate ground truth for aided target recognition and sensor fusion.

The vacuum in autonomous vehicle data

General Motors is cutting 50% of the remaining workforce at its Cruise autonomous vehicle unit. This decision follows a strategic shift from the ride-hail business toward personal autonomous vehicles. Since its 2016 acquisition, GM spent more than $10 billion on Cruise. The layoffs follow an investigation into the October 2023 accident in which a robotaxi ran over a pedestrian. California regulators suspended the license to operate in the state following this incident. In the first nine months of this year, Cruise posted pretax losses of $1.9 billion. As the robotaxi market shrinks, the demand for specialized data labeling in the autonomous vehicle sector faces new pressures. You should understand that the loss of commercial ride-hail operations directly impacts the volume of real-world machine hours available for training perception models. Scale AI originally founded itself to label images for autonomous vehicles, but it has since grown its annualized revenue run rate to $1.5 billion. How will the shift to personal vehicles affect the long-term training needs for sensor fusion and target recognition? Cruise leaders, including President and Chief Technical Officer Mo ElShenawy, directed the restructuring to focus on improving technology and vehicle performance to build trust. The job cuts involve 900 workers, which is about a quarter of the total workforce. Nine leaders also left the company following the investigation into the October accident.

Precision versus scale in the labeling market

Scale AI remains the premium choice for large enterprises that require managed, high-volume annotation across complex modalities like 3D LiDAR and multi-turn conversational data. While Scale AI targets large budgets, competitors capture the mid-market and specialized niches. Surge AI reported over $1 billion in revenue for 2024 and maintains a workforce of 50,000 specialists. Labelbox provides more direct control to engineering teams that want to manage annotation workflows and evaluation logic themselves. Snorkel AI uses programmatic labeling to reduce the need for human annotation by using labeling functions and weak supervision. Appen focuses on multilingual needs with a contributor pool of over 1 million people. Mercor is a contractor marketplace that connects AI labs with expert annotators, and Scale AI sued them for trade-secret misappropriation in September 2025. Turing also competes in this space with a $300 million annualized revenue run rate. iMerit provides high-quality data services for computer vision and NLP. Scale AI builds its capacity through a global workforce located in Kenya, the Philippines, and Venezuela. This scale of operations allows them to provide the high-volume throughput that larger enterprises demand.