Perle

Over 99% of the electricity powering AI is thrown away as heat. That's not a rounding error, it's a structural flaw in how we build AI infrastructure. Ahmad El Shiekh just published new research tackling this head-on: "AI Compute at the Thermodynamic Limit: Energy Scarcity, GPU Heat, and Functional Quantum Materials." The core insight: the AI energy crisis isn't just an engineering problem. It's a thermodynamic systems problem   and the solution may already exist in the Earth's crust. The paper proposes using Bornite (Cu₅FeS₄), a naturally occurring mineral with a unique quantum structure, to convert GPU waste heat back into usable electricity. Instead of the current one-way energy model: ⚡ Energy → GPU → Heat → Gone Ahmad El Shiekh propose a circular one: 🔄 Energy → GPU → Heat → Thermoelectric Generator → Recovered Electricity Why Bornite? → Abundant and non-toxic (copper, iron, sulfur) → Enters a superionic state at ~when heated   ideal for thermoelectric generation → After 10–20 years of service, it remains high-grade copper ore for recycling The next AI scaling breakthrough won't come from larger models alone. It will come from smarter energy flows. Read the full paper and blog post on the Perle AI website. https://lnkd.in/d4Tx3DBh #AI #EnergyEfficiency #Sustainability

Benchmarks drift; business reality doesn’t. Winning teams trust us to build evals around real workflows and edge cases. https://lnkd.in/eunVGMr7

ETHDenver 2026 Recap: Building the Sovereign AI Data Layer The Perle Labs team was on the ground at ETHDenver last week, and one thing was clear: the AI conversation is maturing. Here’s where we showed up: At Multichain day, our CCO Arthur Ding spoke on AI infra onchain alongside teams from 0G Labs, Kite AI, and Pi² Labs. The discussion focused on coordination — what breaks when AI systems span chains and execution layers, and where crypto meaningfully changes the infrastructure stack. At RoboCon by PrismaX, our Founder & CEO Ahmed Rashad joined Sentient and PrismaX to talk about scaling human intelligence for AI. The discussion focused on incentive design, verification mechanisms, and why human-in-the-loop systems need to be architected as infrastructure. At Proof of AI by Kite, the team joined discussions on autonomous agents, open-source innovation, and how to move agent systems from chat interfaces to real execution engines alongside the team from Avalanche, The Block, and others. And lastly at Open AGI Summit by Sentient Labs, Ahmed delivered a keynote: “Trust is the New Compute: Why Sovereign Data Wins the AI Race.” The core message was simple: AI is moving from generating content to making high-stakes decisions — medical diagnoses, autonomous systems, defense applications. If the training data is wrong, the decisions are wrong. The next AI race won’t be won by bigger models. It will be won by those who can verify and control the data layer. That’s the layer we’re building at Perle Labs. ----- Thank you to everyone who stopped by our booth and contributed to the growing conversation around human-verifiable data. See you all at the next one!

Perle Labs CEO Ahmed Rashad on Why AI Needs Verifiable Data Infrastructure AI agents dominated ETHDenver 2026, from autonomous finance to on-chain robotics. But as enthusiasm around “agentic economies” builds, a harder question is emerging: can institutions prove what their AI systems were trained on? BeInCrypto spoke with Ahmed Rashad, Founder & CEO of Perle, on the sidelines of ETHDenver 2026 Rashad previously held an operational leadership role at Scale AI during its hypergrowth phase. In the conversation, he discussed data provenance, model collapse, adversarial risks and why he believes sovereign intelligence will become a prerequisite for deploying AI in critical systems. Read the full interview: https://lnkd.in/dDV_UzxM

LLMs don’t fail loudly. They fail quietly through tool calls. Governance-grade evaluation of agent actions is now table stakes for enterprise AI. Mind the gap isn't just for the tube or train. Trust Perle. https://lnkd.in/gu5CASEN

AI governance isn’t a checklist. It’s a living system. If you can’t produce versioned datasets, traceable evaluations, and evidence on demand, you don’t have compliance, you have hope. AI leaders count on Perle. https://lnkd.in/e4Je7aKJ

Columbia AI Hackathon 2025 Wrap-Up! Last week, we sponsored the AI Hackathon at Columbia University alongside Anthropic, ElevenLabs, Lovable, BizCrush, and more. Over 150 builders came together for 12 hours of rapid prototyping, collaboration, and ambitious AI ideas. The room was filled with focus, creativity, and the kind of momentum that defines the NYC builder scene. Congratulations to the winning teams and to everyone who brought their creative visions to life in a single day. That level of execution is what pushes the ecosystem forward. Proud to support the next generation of AI builders, and big appreciation Iterate and Columbia Business School for organizing, and Simon Lorenzo, Yoan Di Cosmo, and Taewan Yoon, CFA for making it all happen 🙌

The AI training AI loop is accelerating. But without human-verified anchor data, it becomes an echo chamber: models recycling their own outputs until everything converges into generic, repetitive intelligence. Perle is building the infrastructure to prevent that through three core layers: 1. Expert Anchoring: Real specialists define ground truth, keeping the system tied to reality instead of probability. 2. Closing the Data Flywheel: Capturing expert decisions + rationales in real time, so feedback becomes compounding intelligence, not static labels. 3. Decentralized Wisdom: Verification needs provenance. Perle makes human intelligence auditable and reputation-based, so datasets can be trusted long-term. The moat isn’t more data. It’s verifiable, expert-anchored corpora that keeps AI from collapsing into itself. Details in blog ↓ (Authors: Sajjad Abdoli PhD, Founding AI scientist, Drew Mailen, Perle Labs) https://lnkd.in/gHhXCpS4

Human-in-the-loop is not a feature. It is an operational system. At scale, it requires incentive design, verification mechanisms, structured feedback loops, and coordination across distributed contributors. At RoboCon @ ETHDenver, our second panel focuses on that infrastructure layer: Coordinating Human Intelligence for AI 🗓 February 18 🕐 2:05 – 2:30 PM Featuring: ➡️ Oleg Golev, Head of Product at Sentient Labs ➡️ Ahmed Rashad, Founder & CEO at Perle ➡️ Bayley Wang, Cofounder & CEO at PrismaX Moderated by Gil Rosen, Managing Partner at Blockchain Builders We will examine how supervision scales, where coordination breaks down, and what infrastructure is required before autonomy becomes viable in real-world systems. RSVP → https://luma.com/2v9ghwum

When regulators ask. When clients probe. When your board leans in. Perle builds the human intelligence layer behind AI with expert-led data, rigorous evaluation, and traceable decisions you can stand behind. This is AI you can defend.