EdgeAI

In January 2025, we committed to generating 10GW of compute and have already identified over 8GW of that. Now, we're planning for 30GW of compute by 2030. A milestone that scales with the rapidly accelerating demand for intelligent systems. Image generated by @ChatGPTapp Images 2.0 😉 #YouCanJustBuildThings

It’s 2026 and people are still solving problems by throwing more GPUs at them. Real world reality check: • Edge nodes in Africa • Intermittent power in Southeast Asia • Strict power walls on industrial sites AI doesn’t become efficient just because you scaled up. It becomes efficient when you respect constraints. #EdgeAI

"When anyone can generate content for free, what matters most is verifiable provenance — knowing where it came from and whether you can trust it"

It’s a start. Vastly more will be needed to extend consciousness to the stars.

We reduced blockchain storage by 69% on a 1GB node — without downtime. Here’s how we built a memory-safe cold storage system for EdgeAI 👇 1/ Most blockchain nodes assume “more data = bigger machines.” We did the opposite. Running on a 1GB RAM / 40GB disk node, we hit a wall: 👉 ~990MB transaction index in RocksDB 👉 OOM risk + disk pressure So we redesigned the system from first principles.

12/ Everyone is building for scale. We’re building for constraints. Because in the real world: 👉 resources are limited 👉 costs matter 👉 inefficiencies break systems The next generation of AI + blockchain infra won’t just scale up — it will scale efficiently #EdgeAI

11/ Key lesson: Scaling isn’t always about adding resources. Sometimes it’s about: - understanding data access patterns - respecting system constraints - designing around reality, not ideal conditions

🐾🤝 @MarsCat_Global × @EdgeAI_xyz We’re excited to partner with #EdgeAI_xyz — a decentralized AI infrastructure bringing intelligence to the edge through blockchain and privacy-preserving computation. #EdgeAI_xyz enables local data processing, federated learning, and on-chain value capture — turning edge devices into autonomous intelligence networks. Together, we’re exploring how privacy-first P2P infrastructure and edge AI can power a more scalable, real-time Web3 ecosystem.

Safety evaluations may therefore need to examine not just behavior, but the origins of models and training data and the processes used to create them.

Grayscale Q2 2026 Assets Under Consideration: Digital assets we're considering for future investment products include: $CC, $CELO, $MNT, MegaETH*, $MON, $TON, $TRX, $ENA, $HYPE, $JUP, $KMNO, $SYRUP, $MORPHO, $PENDLE, $ROBO, $FLOCK, $GRASS, $KAITO, $KITE, Nous Research*, Poseidon*, $VVV, $VIRTUAL, $WLD, $2Z, $GEOD, $HNT, $JTO, $ZRO, $W Read the full report: grayscale.com/blog/general-u…