Pinecone launches first serverless cloud region in Singapore

Pinecone has launched its first serverless cloud region in Asia, in Singapore, extending its footprint into the Asia-Pacific market.

Hosted in the AWS Asia Pacific (Singapore) Region, the deployment gives customers in Southeast Asia, Australia and other Asia-Pacific markets access to Pinecone's serverless vector database services with local data residency. The region is available now.

The expansion comes alongside a broader set of product releases across AI retrieval, search and pricing. They include Pinecone Nexus, a knowledge engine for AI agents; KnowQL, a query language for retrieval tasks; a Marketplace of pre-built applications; a Builder tier priced at USD $20 a month; native full-text search; and Dedicated Read Nodes for production workloads.

Asia has become an increasingly important market for cloud and AI infrastructure providers as companies seek to keep data within regional jurisdictions and reduce latency for services used in customer support, search, automation and software development. By adding a Singapore region, Pinecone is aiming to address both needs for users running AI systems closer to end markets in the region.

Ash Ashutosh, chief executive officer of Pinecone, linked the launch to broader access for developers and businesses.

"The best knowledge infrastructure should be accessible to every builder, in every region," said Ashutosh. "Launching our first serverless region in Asia marks a significant milestone for Pinecone. Organisations across the Asia-Pacific now have access to the same infrastructure that more than 9,000 customers worldwide rely on - with the data residency, low latency, and proximity that enterprises in the region require. Combined with today's launches of Nexus, KnowQL, Marketplace, and our new Builder tier, we are delivering the most comprehensive knowledge infrastructure for AI at a scale and price point that removes every barrier to building."

Agent tools

Pinecone Nexus is positioned as a system to improve how AI agents access and use stored knowledge. Pinecone says many agent-based systems currently spend most of their effort retrieving context rather than completing tasks, leading to slower responses and higher computing costs.

The company says Nexus shifts more of that work into what it calls knowledge compilation. The product transforms raw data into task-specific outputs that agents can use directly, while a retriever serves those outputs with citations and conflict-resolution rules.

Pinecone reported early results showing up to 90% lower token usage, task completion rates above 90%, and a 30-fold improvement in time to completion. Those figures were provided by the company and were not independently verified.

At the centre of Nexus is KnowQL, which Pinecone describes as a declarative query language for agent retrieval. It is designed to replace custom tool definitions and manual integration work with a single query that sets parameters such as output format, citation requirements and latency budgets.

Marketplace push

Pinecone also introduced a Marketplace, launching with more than 90 knowledge applications. The catalogue spans sales and revenue, insurance, property, legal and compliance, human resources, and customer support.

Built by Pinecone and its partners, the applications can be deployed and customised without assembling infrastructure from scratch. The Marketplace is free at launch, with partner-built commercial offerings to follow.

Pricing changes

Pinecone's new Builder tier will cost USD $20 per month and includes access to its infrastructure and support. The company is also introducing Dedicated Read Nodes, which provide provisioned read capacity and fixed hourly pricing for high-throughput use cases.

Pinecone says Dedicated Read Nodes can cut costs by 77% to 97% at scale for sustained workloads. It also offers a Bring Your Own Cloud model, under which Pinecone is managed within a customer's own cloud environment for organisations with regulatory or residency requirements.

Another addition is native full-text search in the core database, now in public preview. This allows users to combine semantic retrieval with exact-match search in a single system.

The Singapore launch marks Pinecone's first serverless presence in Asia, adding to existing serverless regions in the United States and Europe. Pinecone says it now serves more than 9,000 customers and 800,000 developers worldwide.