Research Partner at 1kx
Check size: No Wei-specific check range is published. 1kx raised a $75M fund in 2024 and says it writes first checks across cycles, but neither fact establishes an individual ticket; founders should use the firm's pitch channel for current sizing.
Wei invests in the financial and security infrastructure that lets blockchains and autonomous software reduce the cost of trust without sacrificing safety. His work joins deep cryptography with venture underwriting: trust-minimized horizontal scaling, programmable cryptography and purpose-built hardware, threat-resistant privacy, and external enforcement, provenance and governance for agentic AI.
Use 1kx's Pitch contact lane or team@1kx.capital, and reference Wei's exact @_weidai identity when the problem is in his domain. Define the trust or security bottleneck, affected production workflow and failure consequence. Show the cryptographic or systems architecture, threat model, benchmarks and what must be enforced outside an LLM or trusted operator. Then identify the initial customer outcome, adoption wedge, network-effect path and why the design requires blockchain, privacy or verifiability rather than merely adding it.
Technically exceptional founders working on hard, security-critical systems; a narrow workflow where missing trust infrastructure blocks real adoption; products that ship an end outcome rather than a human-facing intermediate tool; deterministic controls outside the model for irreversible actions; paths from a vertical wedge to identity, reputation or other network effects; and architectures that preserve interoperability, privacy and verifiability while scaling horizontally.
Likely weak fits include AI agents whose safety rests only on prompt-level refusals; I/O filters presented as complete protection against multi-step failures; security products that detect issues but cannot enforce or verify resolution; blockchain scaling that fragments state and network effects or imports opaque external trust; generic infrastructure without a concrete adoption-blocking wedge; and cryptography whose cost prevents deployment with no credible hardware/software path.
Evaluate observation and provenance, deterministic enforcement, trajectory-level evaluation and governance as one system; local per-step controls cannot secure a multi-step agent lifecycle alone.
Start with a workflow where security blocks an already valuable agent outcome, ship that outcome end to end, then compound identity, reputation or tool-network effects across organizations.
A scalable application should add independent capacity without a global bottleneck while preserving liveness, validity and data availability with as little trust external to settlement as possible.
Underwrite the theory, implementation, developer surface and purpose-built hardware together; programmable cryptography has little impact if performance overhead keeps it out of production.
The model may propose a sensitive action, but a non-LLM checkpoint using scope, authorization, provenance and effect policies must decide whether it executes.
Prefer configurations whose trust compression and compounding network effects require a blockchain rather than products using one as an ornamental database.
ZK and FHE adoption requires a programmable processor designed for advanced cryptography and deep co-design across research, software and hardware.
An edge blockchain can preserve privacy, auditability, interoperability and network effects while allowing individual applications to scale without a global execution bottleneck.
Networking and memory systems purpose-built for decentralized networks can remove the default peer-to-peer gossip bottleneck without compromising decentralization.
Agent adoption is now constrained by security, and the key startup whitespace lies in provenance-aware observation, deterministic enforcement, trajectory-level assurance and governance outside the LLM substrate.
Onchain finance cannot scale sustainably without privacy infrastructure that remains secure against realistic threats while satisfying institutional requirements.
+ 120 more investments. View fund →