Transcending Space and Time
unlocking the next step function in web3 innovation
(M31 Capital recently invested in Space and Time’s Series A)
Space and Time (SxT) is a ZK coprocessor specifically designed to enable onchain smart contracts and AI agents to trustlessly query and process large offchain datasets, providing data-rich context that significantly enhances the developer design space and Web3 application functionality.
Investment Thesis
SxT significantly enhances the functionality of onchain smart contracts and AI agents by trustlessly integrating them with offchain data warehousing and compute capabilities.
SxT’s proprietary Proof of SQL technology powers the industry’s fastest and most cost-effective verification solution.
The founding team is composed of senior leaders from enterprise-grade Web2 data management giants, such as Teradata and AWS, as well as Web3 stalwarts, such as Chainlink and zkSync.
Chainlink and Microsoft’s early investments in the protocol, as well as their continued product partnerships, suggest strong product-market-fits in both Web3 and Web2.
The platform’s general-purpose design opens up an enormous $83B TAM by 2030.
Smart Contracts & AI Agents Lack Context
Smart contracts enable trustless automation, but they’re limited in the type of calculations they can perform. Even simple queries used to execute basic business logic can’t be done with a smart contract, let alone queries that generate larger analytic insights. You can run these queries in a data warehouse, but today’s data warehouses are centralized, so the query results can’t be trusted by (integrated with) smart contracts.
Enter Space and Time
Space and Time (SxT) is a hybrid transactional database and analytic data warehouse that can handle any workload in a single cluster. The platform can run queries against data indexed from all major blockchains as well as data loaded from any offchain source. The core innovation? Proof of SQL, a proprietary and patent protected zero-knowledge (ZK) technology that cryptographically proves the accuracy of each query result, ensuring the data's integrity and tamper-proof nature. This capability, trustless data warehouse querying, allows for new, expansive use cases across all Web3 use cases and Web2 enterprise-scale business processes.
SxT is also developing a novel Proof of Vector Search verification mechanism, enabling LLMs to trustlessly leverage tamperproof RAG (Retrieval-Augmented Generation) data, allowing onchain AI agents to access the relevant context needed to maximize user utility.
Target Markets & Use Cases
SxT is blockchain and use-case agnostic, allowing the platform to serve a wide range of Web3 applications. The platform’s verifiable data warehousing capabilities are also highly relevant for legacy Web2 enterprises and financial institutions.
DeFi Protocols: SxT empowers DeFi applications to securely offload complex logic, enhancing their compute power while verifying results onchain. This is a game-changer for onchain order book DEXs, lending protocols, and perps.
Example: MYSO Finance uses SxT to perform intricate financial calculations offchain, such as Black-Scholes pricing models and token price volatility surfaces, leveraging verified onchain token price data.
AI Agents: AI agents gain trustless access to both and offchain onchain data, such as user token balances and the state of DeFi applications, through SxT’s tamper-proof vector search capabilities.
Blockchain Games: Web3 games can merge game-generated data with real-time indexed blockchain data, facilitating more complex earning schemes and better understanding in-game events that lead to onchain transactions.
Example: Abyss World logs in-game events in SxT, publishing onchain attestations to ensure fair and neutral gameplay.
Web3 Data Companies: For data platforms, SxT offers a comprehensive solution, indexing multiple chains and transforming data into required formats without the need for separate tools.
Example: Chainlink uses SxT to index real-time interest rates from major lending protocols, verifying proofs on Chainlink Decentralized Oracle Networks (DONs).
Verifiable LLMs for Enterprises: As enterprises increasingly rely on large language models (LLMs), the need for verifiable data and auditable processes becomes paramount (evident in the many lawsuits that have recently been brought against major AI companies). SxT, in partnership with early investor Microsoft, uniquely enables enterprises leveraging LLMs to answer serious legal & security questions such as:
What datasets were used for training/fine-tuning? Did they contain copyrighted content or protected IP?
Was sensitive data removed before training or populating vector search databases?
Are user requests processed with the correct LLM binaries and weights? Can you trust third-party LLM services?
Has sensitive IP been accidentally sent to third-party LLM services due to RAG processes?
How do we govern and approve prompts that use enterprise datasets?
How do we certify content authenticity and provenance?
Microsoft Partnership
This project explores tamper-proof methodologies for verifying the training and inference of LLMs using ZKPs and other cryptographic methods.
Key Highlights:
Watermarking/data provenance: Utilizes signed data from the source to verify the integrity and provenance of the training dataset. This ensures that the training data has not been tampered with and its origin can be traced.
Verifying RAG proving: Verifies the soundness of the underlying mathematical proofs used in the LLM model. This ensures that the model is functioning correctly and has not been manipulated.
Sanitizing vector search: Consensus-based approaches are used to sanitize the vector search process, ensuring privacy and data protection while maintaining the accuracy of the LLM.
Overall, this collaboration aims to address the challenges associated with verifying the trustworthiness and integrity of LLMs, enabling customers to trust the training and inference processes with greater confidence.
SxT Leadership
The founding team is a mix of senior leaders from enterprise-grade Web2 data management giants such as Teradata and AWS, as well as Web3 stalwarts such as Sergey Nazarov, Founder of Chainlink (pre-seed investor), and Alex Gluchowski, CEO Matter Labs (zkSync).
Nate Holiday, Co-Founder, CEO | LinkedIn
Former SVP of GTM, Teradata, Bain & Co, and Chainlink Advisor
Scott Dykstra, Co-Founder, CTO | LinkedIn
Former VP of Cloud, Teradata, and Sotero Advisor
Jay White, PhD, Co-Founder, Head of Research | LinkedIn
Former professor at Dordt University
Johnny Debrodt, Co-Founder, Head of Database Engineering | LinkedIn
Former Head of Engineering, AWS Athena
Business Model
SxT employs a subscription-based pricing model, with plans ranging from $699 to $8,299 per month. It also offers a pay-per-compute model, allowing users to pay for individual queries. In contrast to other data warehousing solutions, all services are included, with users only paying for computing. This includes data storage, indexed blockchain data, APIs, streaming, Proof of SQL, dashboards, and the SxT Studio UI.
Competitive Landscape
ZK Coprocessors
While several projects address similar problems, SxT offers developers an interface more akin to a data warehouse, integrating with Web3, Web2, and LLMs, whereas others focus solely on Web3. Additionally, SxT users can integrate historical onchain data with their datasets, a feature not provided by other providers. Most importantly, SxT’s proprietary Proof of SQL verification technology is much faster and more cost-effective than other ZK Coprocessors.
Blockchain Indexing
Unlike most indexing protocols that are primarily used for analytics and front-end applications, SxT can be utilized by smart contracts because of its verifiability. Additionally, apart from SxT and Subsquid, other platforms do not allow developers to access offchain data.
Network Design
Node Architecture
Validators: The central nervous system of SxT, validators provide a suite of microservices to facilitate platform functionality. The diagram below illustrates a high-level overview of the validator architecture, which provides a means for data to enter the system (e.g. blockchain indexing) and data to exit the system (e.g. smart contracts).
Routing: Supports transactional and query interaction with the data warehouse network.
Streaming: Acts as a sink for high-volume customer streaming (event-driven) workloads.
Consensus: Provides performant BFT (Byzantine Fault Tolerant) consensus on data entering and exiting the platform.
Query Proof: Provides Proof of SQL to the platform.
Table Anchor: Provides proof of storage to the platform by anchoring tables onchain.
Oracle: Supports Web3 interactions including smart contract event listening and cross-chain messaging/relaying.
Security: Prevents unauthenticated and unauthorized access to the platform.
Data Warehouse: The Data Warehouse is the backbone of the SxT network. It’s a decentralized, Web3-native hybrid transactional/analytical processing (HTAP) engine that enables trustlessness, scalability, and lightning-fast performance for any data workload.
Data Ingestion: Saving data from external sources.
Data Transport: Warehouse-to-warehouse data transfer.
Data Storage: Persistent saving of data with a view to any point in time.
Data Transformation: Data cleaning, aggregations, multi-source data joins.
Data Serving: Easy and performant access to data, intelligent caching, and creation of data APIs.
Data Ingestion
SxT has built a powerful Rust-based indexer that retrieves every smart contract event associated with transactions from each block on a blockchain in a verifiable way. Here's how the process works:
Data Generation from the Blockchain: SxT queries individual blockchain archival nodes using RPC calls to retrieve both real-time and historical blockchain data.
Extracting Data from Blockchain Networks: The indexer automatically extracts finalized block data from the blockchain networks in three ways:
Real-time: polls the real-time blocks from blockchain nodes.
Historical: polls previous blocks as needed to create a complete historical record of blockchain data.
Decoding: decodes event data from smart contracts and stores it in new tables.
Consensus: The network leverages BFT consensus, which is executed on each Validator node. These nodes validate and agree on the loaded blockchain data, eliminating the need for any trusted intermediary.
Ingesting Data into the Data Warehouse: Once consensus is achieved, the processed data is inserted into the data warehouse, where it can be easily queried.
Proof of SQL Verifiability
Proof of SQL allows the data warehouse to generate a SNARK (Succinct Non-Interactive Argument of Knowledge) cryptographic proof of SQL query execution, proving that the query computation was done accurately and that both the query and the data are verifiably tamper-proof. Proof of SQL enables queries against both on and offchain data, and the results can be trustlessly connected back to smart contracts. This essentially allows smart contracts to ask complex questions about data on their chain, data on other chains, and offchain data from any source.
Proof of SQL employs zero-knowledge proof protocols, where an untrusted prover convinces a verifier of a statement's truth without revealing extra information. The prover, typically the data warehouse, generates proof demonstrating the accuracy of query results, leveraging complex cryptographic algorithms on NVIDIA GPUs for efficiency. Verification, carried out by redundant validators, ensures trustlessness. This approach significantly reduces computational burden, enabling efficient and secure data querying processes.
Let’s walk through the process of a tamperproof query step by step:
Data ingestion and commitment creation: Data entering SxT is routed through a validator layer, which creates a small fingerprint of the data for later use, while the raw data is stored in the data warehouse.
Query request: When a query is sent, the validator directs it to the appropriate engine in the data warehouse.
Proof generation and query result: The data warehouse processes the query, generates a cryptographic proof ensuring data integrity, and sends both the result and proof back to the verifier.
Proof verification: The verifier then cross-checks the proof, query result, and stored fingerprint to ensure data consistency, ultimately yielding a tamperproof query result.
Technical Differentiation
SxT stands out from other coprocessors due to its unique approach to ZK circuit design and optimization, enabling it to process billion-row datasets efficiently. Unlike conventional solutions, SxT leverages advanced techniques such as Sumcheck and Sona commitments, resulting in blazingly fast ZK proving times.
One of the key factors contributing to SxT's speed and cost-effectiveness is its custom zero-knowledge acceleration framework, known as Blitzar. This framework is specifically designed to efficiently scale from a single GPU to a cluster, ensuring optimal performance regardless of the scale of the computation.
Moreover, SxT employs intelligent blockchain indexing techniques to create Hypertables, which enhance data organization and retrieval efficiency. Because of this SxT can significantly reduce circuit sizes, further optimizing computational resources and accelerating ZK proof generation.
SxT's innovative approach to ZK circuit design, coupled with its custom acceleration framework and blockchain indexing capabilities, allows it to function much faster and cost-effectively compared to other solutions.
2030 Total Addressable Market (TAM)
Given the platform’s relevance across both the Web3 and Web2 data landscapes, we believe SxT’s TAM can reach $83B by 2030.
The SxT team estimates the market for ZK verifiable compute capabilities will reach $1B by 2030.
According to Markets and Markets, the global LLM market is expected to reach $36 billion by 2029. If 15% of this infrastructure is utilized for the verifiability of model training, inference, and datasets, it presents a $5 billion opportunity.
Expert Market Research estimates the Web2 data warehousing market will be worth $77 billion by 2030.
Roadmap
H1 2024
General Availability of the SxT data warehouse
Proof of SQL “bring your own database” connectivity
Proof of SQL natively verified on Chainlink
Begin onboarding node operators
Proof of SQL testnet
H2 2024
Testnet for the SxT network with governance
Proof of SQL verified onchain in a verifier contract
SxT open-source framework for building ZK rollups on SxT
End of 2024 / H1 2025
General Availability launch of the SxT network
The Next Generation of Web3
Space and Time’s coprocessing technology opens the door for the “Next Generation” of Web3 functionality and represents the most important step function advancement since data oracles ignited DeFi Summer in 2020. Similar to then, it’ll be hard to predict the most successful use cases and applications of this period, but I’m excited to watch the reinvigoration of Web3 experimentation and innovation over the next few years.
About M31 Capital
M31 Capital is a global investment firm dedicated to crypto assets and blockchain technologies that support individual sovereignty.
Website: https://www.m31.capital/
Twitter: https://twitter.com/M31Capital