Intelligent Delivery functions across heterogeneous hardware fleets, models, and ranking stages to optimize compute based on user intent, candidate quality, and system conditions. The architecture employs multi-layered mechanisms alongside agentic AI to continuously balance resources across boundaries. Transitioning from coarse manual segmentation rules to a shared machine learning intelligence layer improved the identification of highest-value compute opportunities by approximately 20%.
Key points
The system evaluates requests in milliseconds to distribute compute based on the marginal…: The system evaluates requests in milliseconds to distribute compute based on the marginal utility of each impression.
Infrastructure constraints are managed by shifting resources across heterogeneous hardwar…: Infrastructure constraints are managed by shifting resources across heterogeneous hardware fleets using multi-layered controls and agentic AI.
Automated request understanding successfully handles nonlinear signal interactions that p…: Automated request understanding successfully handles nonlinear signal interactions that previously limited static, manual compute allocation methods.
Sources checked by our analysis pipeline: 1
- Ads ranking models are growing toward LLM scale faster than the hardware to run them can keep up. An extra unit of compute has more impact on some ad requests, models, and chips than on others. The challenge is deciding where to allocate it to get the best outcome for the entire system.
- We built Intelligent Delivery, an ML-driven system that allocates resources across the serving stack toward the system-wide optimum. It has delivered ~10% incremental ranking performance on top of model and system advancements since 2017.
- We’re sharing a few of Intelligent Delivery’s capabilities: how we build ML models to predict the marginal utility of compute, multi-layered mechanisms to move resources across the system, and how we increasingly leverage agentic AI to continuously detect and respond to system shifts.
As frontier ads ranking systems advance toward larger models, broader candidate sets, and richer delivery architectures, they also place greater demands on serving efficiency. Compute resources do not scale at the rate that models and system complexity do. The challenge is not just how to build more…
Google is rolling out its Gemini artificial intelligence voice assistant to Apple CarPlay dashboards. The integration brings the AI-powered voice assistant directly to compatible in-car displays for CarPlay users.
The substantial capital infusion positions the British startup to tackle critical memory and capacity bottlenecks that have intensified with the rise of agentic AI. Backed by state and private capital, the initiative supports national ambitions to retain strategic ownership over foundational compute infrastructure. The move directly challenges high artificial intelligence inference expenses through a vertically integrated hardware and software approach.
Aviation carriers face emerging operational challenges as autonomous web systems increasingly interact with consumer infrastructure. The airline is actively implementing restrictions to prevent non-human tools from interfacing with its digital services to protect system integrity. This precautionary measure aims to address potential vulnerabilities that automated software might introduce.