Microsoft–Perplexity: A $750 Million Cloud Power Play
The AI infrastructure race has entered a new phase as Perplexity, a fast-growing AI search startup, signs a three-year, $750 million agreement with Microsoft Azure. The contract goes beyond standard cloud procurement and reflects a deliberate strategy to diversify infrastructure, expand model access, and support large-scale AI deployment without relying on a single provider.
Invest in top private AI companies before IPO, via a Swiss platform:

Perplexity has positioned itself as a challenger in AI-powered search, aiming to rethink how users discover and verify information online. Until now, the company has been closely aligned with Amazon Web Services (AWS), using its Bedrock platform to access advanced models, particularly from Anthropic. The new Azure agreement adds a second major infrastructure pillar while maintaining its existing AWS footprint.
Why $750 Million Matters
A $750 million commitment over three years signals preparation for sustained growth. For an AI company, long-term cloud agreements effectively secure the computing capacity required to train, fine-tune, and run large models at scale. By locking in this capacity early, Perplexity reduces infrastructure uncertainty as demand increases. The deal strengthens its operational foundation without requiring full vertical integration.
Opening the Door to Multiple Frontier AI Models
A central component of the agreement is Microsoft’s Foundry service, which enables Perplexity to deploy and manage advanced AI models on Azure. Through Foundry, the company gains access to models from OpenAI, Anthropic, and xAI. This architecture allows Perplexity to select models based on task performance, cost efficiency, and latency requirements. Instead of committing to a single model provider, the company can dynamically integrate the most suitable system for reasoning, summarization, or conversational search.
Multi-Cloud: A Strategic Survival Move
The agreement reflects a broader industry shift toward multi-cloud infrastructure. Large AI customers increasingly distribute workloads across multiple providers to avoid vendor lock-in, optimize pricing, and maintain redundancy. Perplexity retains AWS as a core infrastructure partner while expanding into Azure in a significant way. For cloud providers, this underscores a competitive reality: leading AI startups are unlikely to rely exclusively on one platform.
Tension Behind the Scenes: From Partnership to Legal Battles
The strategic diversification also unfolds against a more complex backdrop. Perplexity’s CEO, Aravind Srinivas, previously described the company as going “all-in” on AWS, and Amazon had highlighted Perplexity as a flagship AI customer. However, a legal dispute has since emerged between the two companies related to Perplexity’s AI-enabled shopping features connected to Amazon’s marketplace. The conflict reinforces the importance of infrastructure flexibility, particularly when commercial partnerships and platform access become legally contested.
Competitive Positioning Against Industry Giants
Perplexity operates in a highly competitive environment dominated by Google, OpenAI, and Anthropic. Unlike Google, which tightly integrates its own models into its search ecosystem, Perplexity does not control proprietary foundational models. Instead, it positions itself as an orchestrator of leading AI systems across providers. This approach allows the company to focus on user experience, interface design, and answer accuracy while sourcing model capabilities externally.
The Future of AI Search Innovation
The Azure agreement strengthens Perplexity’s ability to scale rapidly while preserving strategic independence. By combining AWS infrastructure with Microsoft’s model ecosystem, the company gains both operational depth and model flexibility. In an environment where infrastructure access increasingly determines AI competitiveness, Perplexity’s multi-cloud strategy provides resilience, negotiating leverage, and access to frontier models without binding its future to a single technology partner.