libp2p

A modular and extensible networking stack which solves many challenges of peer-to-peer applications.

[libp2p @libp2p]

Introducing the libp2p Annual Report 2025 at the forefront of decentralized networking and AI.

From securing global networks to powering next-gen data availability and light clients, libp2p continues to be foundational infrastructure for Web3 and beyond.

Let’s dive in 👇

[Manu Sheel Gupta @manusheel]

@lla_dane Very nice contribution, Abhinav :)
Great to see experimentation around decentralized AI coordination using pubsub primitives. The generator/verifier model with peer-ranked consensus is neat.
Wish to recommend to explore potential intersections with @libp2p, @IPFS. Keep it up :)

[TET @stevenexusz]

@manusheel @libp2p Thank you @manusheel — libp2p foundation made this possible.

Spec doc here for review:
https://t.co/d0wtTedRrX

Particularly the /tet/v1/tmail topic isolation from consensus tx (B.1.2 in spec). Curious if that's the right approach for
production-grade ephemeral messaging.

[Manu Sheel Gupta @manusheel]

@sam_commonly @libp2p @IPFS Thanks for your feedback. Wish to share that there is an opportunity to build open Agent Membership Protocols with @libp2p.

Combining @libp2p identities with verifiable credentials decentralized coordination could enable agents to discover, authenticate, collaborate at scale.

[libp2p @libp2p]

5/5 A real treat was Soham Bhoir one of libp2p's core maintainers across GossipSub and NAT traversal modules in py-libp2p.

The use case was drug discovery.

The demo showed libp2p used for:
- peer discovery
- capability advertisement
- distributed coordination
- GossipSub communication between quantum nodes
- IPFS is used for experiment data pinning and provenance.

The system ran docking simulations for protein-ligand interactions across distributed quantum peers, reducing processing time by roughly half.

Why this matters: Quantum compute is fragmented across providers.

libp2p + IPFS can create a coordination layer where compute resources, experiment outputs and provenance are composable across networks.

That’s the bigger unlock.

[libp2p @libp2p]

The next bottleneck in AI is no longer models.

It’s coordination.

We kicked off Ground Truth to explore how projects are using @libp2p + @IPFS for:
• peer-to-peer agent coordination
• offline-first AI systems
• resilient communication across constrained networks
• verifiable data + model artifacts
• portable infrastructure across cloud and edge

Let's dive into the projects that presented.

[libp2p @libp2p]

4/5 AgentKit with @coinbase x402

Extending Universal Connectivity with multi-agent AI systems leveraging AgentKit coordinating over libp2p + IPFS.

Sumanjeet demoed edge AI agents exchanging state and inference outputs over libp2p while storing checkpoints and artifacts on IPFS.

For product owners this means:
• AI systems that work even with poor connectivity
• less reliance on centralized infrastructure
• portable AI workloads across cloud, edge and devices
• verifiable model outputs and artifacts
• easier coordination between autonomous agents

[TET @stevenexusz]

TET Network sprint 3 milestone:

Founder wallet now controlled by mnemonic (not hardcoded dev key).
2.5B TET allocated, signed transfers working end-to-end on local testnet.

Found and fixed a critical bug pre-launch: dual genesis_hash
implementations in tet-core. Single source of truth shipped.

Phase 0 ship target: end of June. 🌱

#Rust #postquantum #libp2p @libp2p

[libp2p @libp2p]

@mudler_it @LocalAI_API @jichiep @sgl_project @__tinygrad__ @no_stp_on_snek Thank you so much :) Appreciate your initiative and great efforts.

[Juan Benet @juanbenet]

Here is Episode 2 of my new podcast dedicated to conversations on the future of neurotech, computing, intelligence, and more.

My guest Dr. Ben Rapoport is co-founder and CSO of Precision Neuroscience (@PrecisionNeuro_), Assistant Professor of Neurosurgery at the Icahn School of Medicine at Mount Sinai, and Scientific Director at Mount Sinai. Previously, he co-founded Neuralink and Simbionics (acquired by Apple).

Precision is building a minimally invasive brain-computer interface (BCI) that reads from thousands of points on the cortex without penetrating it. The Layer 7 device is implanted through a one-millimeter slit in the skull rather than the larger borehole other approaches require. It is also fully removable.

Precision seeks to help the 5 million people living with severe paralysis in the US (including 800,000 new stroke cases per year). In March 2025, Precision received FDA clearance for a temporary wired version of the system. Over 85 patients have been implanted with and used the device in clinical studies (50 at the time of our conversation). Wireless implants are planned for 2027.

We go deep on the history of Neurotech from the 1980s to the ML inflection points that triggered Neuralink's founding, why surface ECoG was a contrarian bet that's now paying off, the path to treating paralysis and stroke at scale, and why Ben believes neural data is at the same inflection point genomic data was in 2000 — a whole class of biological problems about to become tractable as computer science problems.

Chapters
00:00:00 Introduction
00:04:39 Paralysis as a lens to understand the brain
00:05:36 The 1980s breakthrough: population encoding and the birth of BCI
00:14:36 Google Translate, ML, and the founding of Neuralink
00:23:08 What is the long-term vision of Precision Neuroscience
00:31:56 Layer 7 and why transformative technology looks impossible at first
00:50:21 The surgery: a slit in the skull, not a borehole
00:55:19 The clinical program: who are the patients
01:04:16 FDA clearance and the path to wireless implants in 2027
01:08:32 The patient population: paralysis and stroke at scale
01:16:26 Neural data as the new genomics
01:30:06 BCIs, AI, and the future of the human-machine interface
01:31:22 From medical necessity to lifestyle technology
01:40:36 Precision as a platform — and an optimistic vision

If you're interested in these kinds of discussions, subscribe to the podcast. And if there’s anyone you’d like to see or hear on the podcast, reply with your suggestions.

Full Episode 2 here and in other platforms below.

[soishell @lla_dane]

Built a secure transport layer today for my p2p stack.

Raw streams -> negotiated -> encrypted (Deffie Hellman + ChaCha20Poly1305).

Feels good seeing the pieces come together.

https://t.co/eVnw0fBtp7

[soishell @lla_dane]

Did an RTT benchmark in my P2P stack:

Sequential ping → ~200–300μs (1s gap)
Burst ping → ~30μs

The gap was because of pure system behavior:
scheduling reset + cache locality + warm vs cold execution paths

https://t.co/xpI6VYzO61

[soishell @lla_dane]

rnet: https://t.co/RkkH6TVYwQ

[soishell @lla_dane]

The project was about
- distributing the compute needed in training an ML model,
- over a p2p network,
- for faster throughput,
- minimal hardware requirements

https://t.co/XQxWh4ev2M

[soishell @lla_dane]

Re-ran this old one after months. Terminal looks beautiful after training an ML model over a p2p network using 2 trainer nodes.

[soishell @lla_dane]

Did an RTT benchmark comparing my experimental p2p stack vs @libp2p

rnet → ~220–300μs
rust-libp2p → ~130–250μs
py-libp2p → ~300–400μs

Same machine, loopback
libp2p: TCP + Noise + Yamux
rnet: TCP + custom DH-based security + Mplex

~60μs implementation overhead in rnet

[thedao.fund @thedaofund]

Ethereum security also means understanding what your system reveals while it runs.

LeakDetect by @manusheel helps teams test for hidden information leakage across wallets, AI agents, and multi-chain systems before adversaries can learn from it.

Learn more: https://t.co/g1Lpt2UVSm

Support: https://t.co/nv1rpPClLz