NEAR Protocol: Privacy Is Critical Infrastructure for the AI Agent Economy
NEAR Protocol co-founder Illia Polosukhin declared at BUIDL Asia 2026 that privacy is emerging as the decisive infrastructure layer for an AI-agent-driven economy, introducing a combined TEE and MPC stack to guarantee data sovereignty at the hardware level.
AI Analysis
Trend Correlation
This signal extends the infrastructure privacy discussion surfaced in the eu-dora-blockchain-node-classification-april-2026 signal, where regulators began scrutinizing on-chain data exposure, and the dtcc-blockchain-infrastructure-risk-framework-april-2026 signal, which flagged enterprise risk management requirements for blockchain deployments. NEAR's TEE-plus-MPC stack is a direct technical response to the class of risks both regulatory signals identified.
Autheo Relevance
Autheo's integrated Layer-0 architecture already places AutheoID at the base of every interaction, meaning identity assertions travel with compute and storage operations by design. NEAR's privacy argument reinforces why AutheoID's on-chain identity layer must support selective disclosure: agents acting on behalf of users shouldn't expose the full identity graph with each transaction. The AEE (Autheo Execution Environment) can incorporate TEE-based execution for sensitive AI workloads, directly mirroring what NEAR is building. THEO AI's agent orchestration layer would benefit from MPC-style coordination so multiple AI nodes can reach consensus on agent actions without any single node seeing complete user context. The QSDAG consensus layer adds a further privacy angle: its directed acyclic graph structure allows transaction graph obfuscation that complements hardware-level TEE protections.
Quantified Impact
If NEAR's 1M TPS capacity is mapped against the McKinsey $3 to $5 trillion agentic commerce projection for 2030, and assuming an average transaction value of $0.001 (micropayment range), sustaining that market would require roughly 95,000 to 158,000 transactions per second on average. NEAR's current stated throughput already covers that ceiling, suggesting throughput isn't the bottleneck; privacy and identity verification are. Protocols that solve the privacy problem first capture disproportionate enterprise market share.
Full Analysis
Privacy has moved from a nice-to-have feature to a hard technical requirement. That's the argument Illia Polosukhin made during his keynote at BUIDL Asia 2026 in Seoul on April 16, and it's hard to dismiss coming from someone who co-authored the Transformer architecture before co-founding a major L1 protocol.
The core problem Polosukhin identified is simple: blockchain's complete transparency, the property that makes it trustworthy, is now a liability in a world where AI agents handle micropayments, healthcare decisions, and financial management on behalf of millions of users. When an agent pays for coffee with a crypto wallet, location and activity data can be exposed immediately. Scale that to millions of concurrent agent transactions, and you have a surveillance infrastructure built right into the payment layer.
NEAR's answer is a hybrid privacy stack that combines trusted execution environments (TEEs) and multi-party computation (MPC). A TEE runs computation inside a secure hardware enclave isolated from external access, including from the hardware provider itself. MPC allows multiple parties to compute a joint result without any single party seeing the raw underlying data. Together, they let agents transact and compute without exposing user data to validators, hardware operators, or network observers.
The practical implication is significant. Many Fortune 500 companies currently hesitate to adopt AI because of data leak concerns. NEAR is betting that verifiable, hardware-backed privacy removes that hesitation and turns crypto into a practical payment layer for enterprise AI workloads. That's not a small market. McKinsey projects agentic commerce will reach $3 to $5 trillion globally by 2030.
NEAR's Nightshade 3.0 sharding already supports over one million transactions per second, providing the throughput needed for high-frequency agent interactions. The chain abstraction layer, which routes transactions across 35-plus networks via a single account, complements the privacy stack by letting agents operate cross-chain without exposing account associations at each hop.
The timing of this announcement aligns with a broader ecosystem shift. Coinbase launched Agentic Wallets in February 2026, giving AI agents dedicated crypto wallets with programmable spending limits. The x402 protocol, now backed by the x402 Foundation with members including Google, Visa, and AWS, is becoming the default HTTP payment rail for agent-to-agent commerce. What's been missing from that stack is a credible privacy layer that enterprises can trust with sensitive workloads. NEAR is positioning itself to fill that gap.
Polosukhin's framing of blockchain as a new operating system for the agent economy is notable. It echoes how earlier infrastructure layers were positioned at category-defining moments: TCP/IP was the communication OS, HTTP was the document OS. If he's right, the privacy properties of that operating system will determine which sensitive applications migrate onchain and which stay on centralized servers.
Key Facts
NEAR co-founder Illia Polosukhin announced at BUIDL Asia 2026 in Seoul on April 16 that NEAR combines TEEs and MPC so that even hardware providers and network operators cannot view underlying user data.
Bloomingbit→NEAR's Nightshade 3.0 sharding supports over one million transactions per second, and its chain abstraction layer manages assets across 35-plus chains via a single account using email or FaceID login.
KuCoin Blog→NEAR AI's private inference runs inside Intel TDX and NVIDIA Confidential Computing hardware-secured TEEs, keeping data and model weights encrypted throughout every request.
NEAR Protocol→Explore the Autheo Platform
See how Autheo's unified infrastructure addresses the challenges and opportunities in blockchain.