Cloudflare’s AI inference strategy differs from hyperscalers by focusing on maximizing system efficiency and utilization per capital expenditure dollar, optimizing the cost structure of AI inference. They use a custom large language model (LLM) inference engine called Infire, improving GPU utilization and minimizing overhead costs.

Cloudflare’s Infire system runs models closer to users, improves startup speed and efficiency, and allows model weights to be cached locally on edge nodes for faster loadups. This system utilizes off-the-shelf hardware in tier-1 cities, enabling quick setup and revenue generation while maintaining flexibility and fast response times for capacity needs.

In contrast, hyperscalers like Amazon and Microsoft use large-scale facilities for high-volume data processing, leading to higher power consumption and latency. Amazon introduced Lambda@Edge to run code closer to applications, while Microsoft employs a hybrid cloud strategy for AI workloads at the edge.

Cloudflare’s stock has risen 9.9% in the past six months, with a forward price-to-sales ratio of 26.19X, higher than the industry average of 4.86X. Analysts project a 21.3% year-over-year growth for Cloudflare’s 2025 earnings, with a Zacks Rank #2 (Buy) at present.

Zacks Investment Research highlights a little-known satellite-based communications firm as a top stock pick with the potential to double in value. Space industry growth and a rapidly expanding customer base could lead to significant revenue increases by 2025. This pick stands out among Zacks’ previous successful selections like Hims & Hers Health.

Read more at Nasdaq: Can Cloudflare’s Edge AI Inference Reshape Cost Economics?