# Spice Firecache | Cloud-Scale DuckDB Luke Kim Founder and CEO of Spice AI December 5, 2023 Spice AI is not just another data indexer or data provider. [Spice.ai](http://spice.xyz/) is a set of AI/ML-infra building blocks for creating data and AI-driven applications - web3 data included. In October's [General Availability announcement](http://powering%20intelligent%20applications%20with%20composable%20data%20and%20time-series%20ai%20building%20blocks/) we also announced [**Spice Firecache**](https://docs.spice.ai/building-blocks/spice-firecache). [_Spice Firecache_](https://docs.spice.xyz/portal/apps/spice-firecache) _is a real-time, in-memory SQL service based on cloud-scale_ [_DuckDB_](https://duckdb.org/) _instances that enables blazing fast SQL query up to 10x the performance of general SQL query. For example,_ [_EigenLayer_](https://www.eigenlayer.xyz/) _uses Spice Firecache to enable scenarios not possible before, including serving dynamic data to their high traffic dashboards, real-time monitoring, and analytics. ML models hosted on the Spice.ai platform can be paired with Firecache to power super-fast, low-latency inferencing, as demonstrated by the_ [_AI predictions demo_](/content/?explore-ai=true#demo/index.html) _on the Spice.ai website._ If you like DuckDB, you'll love it at cloud-scale, automatically provisioned and updated with real-time data. Experience high concurrency, blazing fast SQL query up to 10x the performance of general SQL query in the Firecache Playground. #### 🔥 Spice Firecache in Preview [Spice.ai](/content/site-root.html) is preloaded with over [100TBs of Web3 and Asset Prices data](https://docs.spice.ai/building-blocks/datasets) which can be combined with your own [PostgreSQL, MySQL, and data lake](https://docs.spice.ai/building-blocks/external-data-sources) external data sources. While organizations need _access_ to terabytes or petabytes of data, the _working set_ of data that is used day-to-day is often orders of magnitude smaller, generally in the 10s of GBs. **Cloud-Scale DuckDB**. Spice Firecache enables you to configure specific datasets for inclusion into cloud-scale managed DuckDB instances for query performance 10x or sometimes even 100x faster than general SQL query. **Platform managed.** The entire Firecache data lifecycle, from ETL to query ensures datasets are updated in real-time and is completely platform managed. All the user has to do is enable Firecache on a dataset, via configuration as code through [GitHub-connected Datasets](https://docs.spice.ai/building-blocks/datasets-and-views), deploy it, and start querying. **Easy to query.** Once datasets have been deployed, they can be queried by the same interfaces as general SQL query, including with the Playground SQL Query Editor in the [Spice.ai](/content/site-root.html) portal, the HTTP API, the Apache Arrow Flight API, and SDKs. **Dedicated instances.** Dedicated Firecache instances are deployed per organization and are Spice app-specific, only available to the Spice app to which they were deployed. **Easy to integrate.** Simply swap to using the **Firequery()** method instead of the **Query()** method using one of the [Go](https://docs.spice.ai/sdks/go), [Python](https://docs.spice.ai/sdks/python-sdk), [Node.js](https://docs.spice.ai/sdks/node.js-sdk), or [Rust](https://docs.spice.ai/sdks/rust-sdk) SDKs to start using Firecache. #### What it enables Intelligent data and AI-driven software is often limited by the speed and cost of data retrieval from big data warehouses and systems. High-performance access to data is required to power real-time operations, observability, analytics, ML inferencing, and recently popular Retrieval-Augmented Generation (RAG) for large-language models (LLMs). By keeping the day-to-day working set of frequently queried data available in Firecache, data retrieval latency can be significantly reduced, resulting in higher-performance (and lower-cost) frontends, applications, and AI-powered insights. This is particularly beneficial for customers like [EigenLayer](/content/cases/eigenlayer/index.html) who quickly rose to popularity, needing to make 10s of millions of SQL queries a month and growing to power their dynamic community and user dashboards. Spice Firecache ensures that software can meet user expectations for speed and interactivity, supercharging the user experience, all while reducing cost. Spice Firecache is currently available in preview for Spice AI Design Partners and Enterprise customers. Get in touch at [hey@spice.ai](mailto:hey@spice.ai) if you'd like to trial Firecache for free. Thank you! _The Team at Spice AI_ **🌐 Community updates:** Spice AI recently partnered with [RISC Zero](https://www.risczero.com/), a leader in Zero-Knowledge (ZK) technology to provide ZK-provable ML inferencing for Ethereum gas-fees. Read the full case-study at [**spice.ai/cases/risc-zero**](/content/cases/risc-zero/index.html) **.** #### About Spice AI [**Spice.ai**](http://spice.xyz/) is a set of AI/ML-infra building blocks for creating data and AI-driven applications - web3 data included. Spice AI eliminates the complexity of building and operating costly data and AI infrastructure by composing real-time and historical time-series data, custom ETL, machine learning training and inferencing, in a high-performance, enterprise-grade platform. Have questions or feedback? [Get in touch on Discord](https://discord.gg/kZnTfneP5u).