AI infrastructure provider Crusoe has secured a multi-year cloud deal with AI search startup Perplexity. The partnership arrives amid severe shortages of cutting-edge compute resources and growing demands to reduce inference costs as generative AI models scale up.

Partnership Overview and Contract Details Between Crusoe and Perplexity

On September 15, 2026, Crusoe, a provider of vertically integrated AI infrastructure, announced a comprehensive, multi-year AI infrastructure partnership with Perplexity, an AI search startup. Under this agreement, Perplexity will consolidate its entire model lifecycle—from frontier model development to production inference—on the single platform of Crusoe Cloud. Specifically, Perplexity will train and fine-tune large-scale AI models using dedicated compute clusters powered by Nvidia’s state-of-the-art GB300 NVL72 and Nvidia InfiniBand networking technology. The agreement also includes mutual platform adoption: Crusoe will deploy Perplexity Enterprise Pro and Max across its entire workforce of over 1,800 employees to streamline internal knowledge search, multi-step research, and data analysis. While financial details of the deal were not disclosed, it follows Crusoe’s approximately \$13 billion cloud contract signed with financial giant Jane Street in early September 2026, significantly bolstering Crusoe’s presence in the AI infrastructure market.

The Technical Impact of Managed Inference: Balancing Speed and Cost

For Perplexity, which delivers real-time AI search services, reducing response latency down to the millisecond is critical to user experience. The company currently processes over 1.5 billion search queries monthly, facing the dual challenge of expanding compute capacity while curbing inference costs during rapid growth. Crusoe’s Managed Inference service is a production-grade inference platform engineered to boost performance and cost-efficiency across major AI models. It enables application developers to focus on expanding agent capabilities and improving accuracy without having to manage complex hardware setups or infrastructure overhead. The diagram below illustrates an architecture overview that unifies everything from model training to inference operations. By adopting this integrated platform, Perplexity can maintain high-speed answer generation for users while keeping operational costs at a reasonable level as its models scale.

Escalating Competition in Neoclouds and Crusoe’s Market Position

Evolution from Crypto Mining to AI Infrastructure Powerhouse

Founded in 2018, Crusoe originally began as an energy-driven startup capturing surplus natural gas at oil and gas extraction sites to power bitcoin mining. However, driven by the explosive demand for generative AI in recent years, the company underwent a dramatic pivot to become a “neocloud” provider, building and operating hyperscale data centers and dedicated AI cloud services. Crusoe’s defining edge lies in its vertically integrated model, controlling everything from energy sourcing to data center design and cloud platform engineering. A \$1.375 billion Series E funding round in October 2025 pushed its valuation past \$10 billion. In 2026, Crusoe entered talks for an estimated \$3 billion pre-IPO funding round that could triple its valuation to \$30 billion or \$35 billion, while also preparing for an initial public offering (IPO) in coordination with major investment banks like JPMorgan. Currently, the company holds approximately 4.9 GW of contracted AI infrastructure capacity, with a development pipeline exceeding 40 GW.

Differentiation from Rivals and a Multi-Cloud Strategy

In the AI-focused cloud market, Crusoe faces fierce competition from rival neocloud providers like CoreWeave and Nebius, as well as hyperscalers such as AWS, Microsoft Azure, and Google Cloud. Although Perplexity previously relied on multiple major cloud vendors including AWS, it chose Crusoe for its next-generation frontier model development. The primary driver was Crusoe’s ability to reliably deliver Nvidia’s highly sought-after GB300 chips alongside low-latency inference environments. Beyond Nvidia GPUs, Crusoe offers AMD processors such as the MI300X and MI355X, providing a flexible multi-vendor environment. This multi-vendor approach appeals to AI companies seeking to mitigate single-vendor lock-in risks, cementing Crusoe’s unique competitive advantage against established cloud giants.

Future Outlook and Broader Implications for Generative AI

Impact of Vertical Integration Across the Model Lifecycle

This partnership underscores a broader shift in generative AI infrastructure trends—from merely renting raw compute capacity to adopting holistic optimization platforms across the entire model lifecycle. For companies building large-scale AI models, splitting R&D, fine-tuning, and production inference across fragmented environments increases data transfer costs and operational complexity. The diagram below illustrates how the infrastructure and application layers interface within modern AI ecosystems.

By tightly integrating Perplexity’s model-agnostic orchestrator, Perplexity Computer, with Crusoe’s vertically integrated cloud, the combined platform delivers faster response speeds alongside reliable execution for complex, multi-step AI agents.

Sustainability of AI Infrastructure Investment and Key Market Drivers

As generative AI models grow increasingly complex and massive, the underlying compute expenses incurred by AI startups are skyrocketing, making profitability a central industry challenge. With global AI expenditure climbing, infrastructure providers capable of supplying cutting-edge hardware efficiently and cost-effectively are becoming essential. Crusoe is making aggressive infrastructure investments globally, anchored by flagship developments such as its 1.2 GW data center campus in Abilene, Texas. Backed by massive capital raises, Crusoe’s strategy of making upfront investments and securing long-term contracts with major clients like Perplexity and Jane Street serves as a landmark model for success in the capital-intensive neocloud industry. Moving forward, the expansion of compute resources and progress toward a public listing will be closely watched for their impact on the pace of growth across the broader generative AI landscape.

References

• 【Crusoe and Perplexity Announce Multi-Year Partnership】https://www.crusoe.ai/resources/newsroom/crusoe-perplexity-partnership
• 【Perplexity AI】https://perplexity.com

【#Crusoe #Perplexity #AIInfrastructure #NVIDIA #Cloud #GenerativeAI #Neocloud】