Snap says its $400M deal with Perplexity ‘amicably ended’

In a move that underscores the volatile nature of the generative AI landscape, Snap Inc. has confirmed the dissolution of its $400 million partnership with AI search startup Perplexity. The deal, which was brokered late last year, aimed to integrate Perplexity’s real-time, LLM-powered search capabilities directly into the Snapchat ecosystem. While both companies have characterized the split as "amicable," the termination of such a high-value agreement suggests a significant strategic pivot for Snap as it reevaluates how to best deliver information to its younger demographic.

The partnership was initially viewed as a masterstroke for both parties. For Snap, it offered a way to elevate its "My AI" chatbot from a simple conversationalist to a high-utility research tool, capable of answering complex queries with real-time web access. For Perplexity, the deal represented a massive distribution channel, placing its "answer engine" in the hands of millions of Gen Z and Millennial users. At the time, Perplexity was the darling of the venture capital world, positioned as a viable challenger to Google’s search dominance. However, the rapidly evolving AI market has since introduced new pressures, ranging from copyright litigations to the rising cost of compute, which may have complicated the long-term viability of the integration.

The mechanics of the deal were built on a revenue-sharing and licensing model that would have seen Perplexity’s API serving as the backbone for Snapchat’s search functions. Integrating a third-party LLM into a social platform is a complex engineering feat that requires balancing latency, safety filters, and cost-per-query. As Snap continues to build out its own internal AI capabilities and hardware ambitions, such as its AR Spectacles, the company may have realized that relying on an external provider for a core utility like search was a strategic vulnerability. By ending the deal, Snap regains control over its hardware-software synergy, potentially favoring more specialized or in-house models that are optimized for augmented reality and quick-fire messaging.

This split carries broader implications for the AI industry, particularly concerning the "platform-as-a-service" model. It highlights the growing tension between specialized AI startups and established social media giants. As tech titans like Meta and Google bake their own proprietary AI models directly into their apps, the "partnership route" for startups is becoming increasingly precarious. For Perplexity, the loss of this deal is a blow to its scaling strategy, though the company remains well-funded and continues to iterate on its Pro and Enterprise offerings. It serves as a reminder that in the AI era, massive distribution deals are never set in stone until the technology becomes inseparable from the user experience.

Regulatory and ethical considerations may have also played a silent role in the deal’s conclusion. Perplexity has recently faced intense scrutiny from publishers regarding its web-scraping practices and how it attributes content. For a publicly traded company like Snap, which must navigate a complex global regulatory environment, close association with a startup embroiled in copyright controversies can be a liability. If Snap intends to push further into news and information discovery, it may prefer to build a framework that is fully insulated from the legal challenges currently facing the broader LLM industry.

Moving forward, the industry should watch how Snap fills the void left by Perplexity. Whether the company doubles down on its partnership with OpenAI or tilts toward its own locally-run models on-device will signal its long-term AI philosophy. Simultaneously, Perplexity’s next moves will be a litmus test for the independence of the "search startup" category. If more large-scale distribution deals falter, these startups may be forced to pivot toward direct-to-consumer subscriptions or deeper enterprise integrations rather than hoping to be the "intel inside" for existing social giants. The era of the $400 million experiment is over; the era of consolidation and self-reliance has begun.