Perplexity AI Launches Innovative Hybrid Inference System

WHO:

VentureBeat - Perplexity AI, a search startup valued at $20 billion, CEO Aravind Srinivas, Intel CEO Lip-Bu Tan.

VentureBeat - Nine organizations, including CNN, the New York Times, News Corp, and others, have filed lawsuits against Perplexity for copyright and trademark infringement.

WHAT:

VentureBeat - Perplexity AI's hybrid inference feature is set to launch in the coming weeks, allowing the system to autonomously decide which AI workloads stay on a user's device and which are routed to cloud models.

VentureBeat - The hybrid inference orchestrator is described as the world's first "hybrid local/server agentic inference orchestrator."

VentureBeat - The CNN lawsuit, filed on May 28, 2026, accuses Perplexity of scraping over 17,000 CNN stories and other content.

VentureBeat - Perplexity AI's hybrid inference feature is not yet available to users and is set to launch in the coming weeks.

VentureBeat - The hybrid inference orchestrator is described as the world's first "hybrid local/server agentic inference orchestrator."

VentureBeat - The system autonomously decides, in real time and mid-task, which AI workloads stay on a user's device and which get routed to cloud models.

VentureBeat - The demonstration at Computex showcased the system processing confidential deal materials using local models running on Intel Core Ultra Series 3.

VentureBeat - The hybrid inference orchestrator is described as the world's first "hybrid local/server agentic inference orchestrator."

VentureBeat - The system reportedly asks for user permission before sending sensitive tasks to the cloud.

VentureBeat - The hybrid inference system aims to address compliance requirements for regulated industries by allowing sensitive data to remain on local devices while still accessing cloud-based models for complex tasks.

VentureBeat - The hybrid inference orchestrator is described as the world's first "hybrid local/server agentic inference orchestrator."

Bloomberg - Perplexity CEO Aravind Srinivas stated that the system is "chip agnostic," indicating it can operate across various hardware platforms.

VentureBeat - Perplexity has raised $200 million in new capital at a $20 billion valuation, just two months after raising $100 million at an $18 billion valuation.

VentureBeat - The company has raised a total of $1.5 billion in funding since its founding three years ago.

VentureBeat - Nine organizations, including CNN, the New York Times, News Corp, and others, have filed lawsuits against Perplexity for copyright and trademark infringement as of May 31, 2026.

VentureBeat - Perplexity AI unveiled the first hybrid local-server inference orchestrator at Computex 2026.

VentureBeat - The hybrid inference feature will launch in the coming weeks.

VentureBeat - Perplexity has raised a total of $1.5 billion in funding since its founding three years ago.

VentureBeat - The CNN lawsuit, filed on May 28, 2026, accuses Perplexity of scraping over 17,000 CNN stories and other content.

WHERE:

VentureBeat - The demonstration took place at Computex 2026.

VentureBeat - Perplexity's headquarters is not specified, but it operates in the AI sector.

WHEN:

VentureBeat - The hybrid inference feature is set to launch in the coming weeks.

VentureBeat - The CNN lawsuit was filed on May 28, 2026.

WHY:

VentureBeat - The hybrid inference system aims to balance intelligence, accuracy, privacy, and cost by autonomously deciding where AI workloads should be processed.

VentureBeat - The system addresses concerns about data governance for enterprises by keeping sensitive data on local machines.

HOW:

VentureBeat - The system reportedly asks for user permission before sending sensitive tasks to the cloud, addressing data governance concerns for enterprises.

VentureBeat - The hybrid local-cloud inference system autonomously decides, in real time, which AI workloads stay on a user's device and which are routed to cloud models.

VentureBeat - The system reportedly asks for user permission before sending sensitive tasks to the cloud.

(UPDATED June 3, 2026 18:28 UTC)

Perplexity AI is preparing to launch its hybrid inference feature in the coming weeks, which will enable the system to autonomously determine whether AI workloads should remain on a user's device or be sent to cloud models. This innovative technology is being touted as the world's first "hybrid local/server agentic inference orchestrator," designed to help regulated industries comply with data governance by keeping sensitive information on local devices while still utilizing cloud capabilities for more complex tasks. Meanwhile, a lawsuit filed by CNN on May 28, 2026, accuses Perplexity of scraping over 17,000 stories and other content from the network.

(UPDATED June 3, 2026 14:28 UTC)

Perplexity AI is preparing to launch its hybrid inference feature in the coming weeks, which is touted as the world's first hybrid local/server agentic inference orchestrator. This innovative system autonomously determines, in real time, whether AI workloads should remain on a user's device or be sent to cloud models, all while prioritizing intelligence, accuracy, privacy, and cost. During a demonstration at Computex, the system showcased its capabilities by processing confidential deal materials using local models on Intel Core Ultra Series 3, and it also seeks user permission before sending sensitive tasks to the cloud to address data governance concerns for enterprises.

(UPDATED June 3, 2026 10:31 UTC)

Perplexity AI is set to launch its innovative hybrid inference feature in the coming weeks, which will autonomously determine whether AI workloads are processed on a user's device or routed to cloud models. This system is being touted as the world's first "hybrid local/server agentic inference orchestrator," and it prioritizes user privacy by requesting permission before sending sensitive tasks to the cloud. Designed to meet compliance requirements for regulated industries, the hybrid inference system allows sensitive data to remain on local devices while still leveraging cloud-based models for more complex tasks.

(UPDATED June 3, 2026 06:36 UTC)

Perplexity AI is set to launch its hybrid inference feature in the coming weeks, which will allow the system to autonomously determine whether AI workloads should remain on a user's device or be sent to cloud models. This innovative hybrid local/server agentic inference orchestrator is touted as the world's first of its kind, and according to CEO Aravind Srinivas, it is chip agnostic, meaning it can function across various hardware platforms. The system is designed to meet compliance requirements for regulated industries by keeping sensitive data on local devices while still utilizing cloud-based models for more complex tasks.

(UPDATED June 2, 2026 23:30 UTC)

Perplexity AI is preparing to launch a hybrid inference feature in the coming weeks, which will enable the system to autonomously determine whether AI workloads should remain on a user's device or be sent to cloud models. The company has recently raised $200 million at a valuation of $20 billion, just two months after securing $100 million at an $18 billion valuation, bringing its total funding to $1.5 billion since its inception three years ago. However, as of May 31, 2026, Perplexity is facing legal challenges from nine organizations, including CNN and the New York Times, over allegations of copyright and trademark infringement.

(UPDATED June 2, 2026 21:54 UTC)

Perplexity AI, a search startup valued at $20 billion and led by CEO Aravind Srinivas, recently unveiled its first hybrid local-server inference orchestrator at Computex 2026, with plans to launch this innovative feature in the coming weeks. However, the company is facing legal challenges, as nine organizations, including CNN and the New York Times, have filed lawsuits against them for copyright and trademark infringement, with CNN accusing Perplexity of scraping over 17,000 of its stories. The new hybrid inference system is designed to balance intelligence, accuracy, privacy, and cost by autonomously determining where AI workloads should be processed, while also ensuring user consent before sending sensitive tasks to the cloud.