Phala Network (PHA)

Overview

Phala Network is a project that brings private computing to blockchains and AI systems. Its core idea is simple: run code in secure hardware so data stays private, then attach cryptographic proof that the code really ran as promised. The network’s token, PHA, powers this compute layer and its related services. Today, Phala’s stack focuses on “confidential AI” using Trusted Execution Environments (TEEs) such as Intel TDX on CPUs and NVIDIA’s Confidential Computing features on GPUs. These TEEs let apps process sensitive data while keeping it hidden from operators, clouds, and even machine administrators, and then output an attestation—essentially a hardware-signed receipt—that others can verify. (intel.com)

Phala originally launched in the Polkadot ecosystem with a privacy-first parachain design. In November 2025, the project executed a planned migration: core token and governance functions moved to Ethereum and an L2, while confidential compute continues to be delivered through Phala’s open-source “dstack” and managed “Phala Cloud.” The move simplified access to PHA on EVM wallets and exchanges and aligned Phala’s compute layer with a broader AI developer base. (forum.phala.network)

At a high level, Phala presents itself as “private compute, public proof.” Developers can deploy AI models, agents, or data-processing jobs without exposing inputs or keys, and downstream users can verify what ran via the project’s public Trust Center. This “prove, don’t just promise” stance is central to the design. (phala.com)

History & Team

Origins and milestones

Phala’s roots trace back to 2018–2020, when the team began building a decentralized TEE network and shipping the first infrastructure to turn hardware attestation into a product surface. In 2024, the work expanded into GPU TEEs and AI workloads. The later 2025 migration anchored governance and token operations on Ethereum while keeping verifiable compute at the center. (phala.com)

Founders and leadership

Phala was co-founded by Marvin Tong (CEO) and Hang Yin (CTO). Tong previously held senior product roles at Tencent and DiDi, while Yin is a former Google engineer and PhD with deep experience in distributed systems and cryptography. Their current public team page also highlights roles across engineering, growth, and operations that support the confidential-compute stack and ecosystem. (phala.com)

From parachain to Ethereum + L2

Historically, Phala operated a Polkadot parachain, with Khala serving as the canary network on Kusama. During the Polkadot crowdloan period, the project attracted over 610,000 DOT contributions to secure its slot. In November 2025, Phala paused the parachain and opened an asset claim portal to migrate balances to Ethereum (PHA) and to the Phala L2 (vPHA), where staking and GPU-based “mining” now continue. (guide.kusama.network)

Investors and ecosystem support

Phala has received backing from notable web3 investors. In January 2024, DWF Labs announced a strategic investment to “power-boost” Phat Contracts adoption, and databases such as CBInsights list earlier backers including IOSG Ventures and Waterdrip Capital. This mix of community crowdloan support and venture capital helped the project scale from parachain origins to a broader confidential AI platform. (phala.com)

Technology & How It Works

Confidential compute with attestation

Phala’s core technology runs applications inside hardware-protected TEEs. On CPUs, the project builds around Intel TDX; for GPUs, it supports NVIDIA H100/H200-class Confidential Computing. During execution, the TEE produces an attestation quote—cryptographic evidence binding the machine, runtime, and code hash. Anyone can verify this proof, either programmatically or through Phala’s Trust Center. This makes claims about “what ran” independently checkable by counterparties, auditors, or downstream contracts. (intel.com)

dstack and Phala Cloud

Developers can self-host using dstack, an open-source framework for confidential containers, or choose Phala Cloud, the managed option. dstack handles remote attestation, key management, and lifecycle tasks for deploying Dockerized applications into TEEs. It’s designed to be portable across clouds and hardware, letting teams keep their runtime public while secrets remain sealed inside enclaves. Phala publishes guides, audits, and templates to speed up deployment for AI inference, agents, and data services. (github.com)

Off-chain logic and “Phat Contracts”

Phala popularized “Phat Contracts,” a way to run off-chain logic in a TEE that can talk to on-chain contracts and web APIs. Developers typically write this logic in familiar languages (TypeScript/JavaScript), wire it to EVM or Substrate chains, and return verifiable results on-chain. This approach offloads heavy or private computation while keeping deterministic state changes on a base chain. (phala.com)

Architecture notes and earlier design

In its parachain era, Phala separated consensus from computation. The blockchain acted as a canonical, encrypted message queue, while secure “workers” (then based on SGX) fetched requests, executed them in pRuntime, and posted results back. Concepts like Gatekeepers, Workers, and key hierarchies were formalized to guard secrets and isolate cluster-level keys. Those ideas continue to inform Phala’s current TEE design, even as the runtime has moved to modern primitives like TDX and GPU TEEs. (docs.phala.network)

Tokenomics & Utility

What PHA powers

PHA is the utility and governance token associated with Phala’s compute ecosystem. It is used to:

• Secure and prioritize access to compute via staking mechanisms and contract-cluster models.
• Pay for confidential workloads or associated services, either directly in PHA or through cluster-defined payment schemes.
• Reward compute providers (e.g., GPU workers) for delivering verifiable capacity. (github.com)

Phala’s tokenomic literature describes a two-sided market. On the demand side, developers stake to reserve computing power for their contract clusters. On the supply side, miners (now GPU-backed workers) are remunerated based on capacity, attested security level, and demand. Various modules coordinate matching, pricing, and long-term incentives. (github.com)

ERC-20 and L2 mechanics

With the 2025 migration, liquid PHA on the former parachain became claimable to Ethereum as ERC‑20 PHA, while delegated balances and certain rewards migrated as vPHA on an L2. Staking continues on Ethereum L1, and GPU mining/rewards operate on L2, with governance handled through Ethereum-native tools. The team’s notice outlines staking contracts, reward splits, a pledge mechanism for high-performance GPUs, and a Snapshot + Safe-based governance flow. (forum.phala.network)

For reference, the long-standing ERC‑20 contract for PHA on Ethereum is published and verified on Etherscan. Many integrations refer to this address, and Phala’s app surfaces it for convenience. (etherscan.io)

Delegation, vaults, and rewards

Phala documentation (from the parachain era and later updates) details delegation to pools, vault-based management, and how rewards roll over automatically. While the execution venue changed in late 2025, the project has kept the idea that users can supply capital to secure compute and earn emissions tied to real workload capacity. Updated mechanics are reflected in the migration guide and app. (docs.phala.network)

Ecosystem & Use Cases

Confidential AI and agents

A prime focus is private AI inference and agent workloads. Teams deploy LLMs, data jobs, or autonomous agents that must hold prompts, keys, customer data, or model weights privately. Using dstack on TDX + GPU TEEs, the runtime encrypts memory and ties the exact code image to hardware measurements. Receipts are published in a Trust Center, letting partners and auditors verify claims without seeing secrets. This pattern fits healthcare, finance, legal, and enterprise AI SaaS. (github.com)

Off-chain compute for web3

Phat Contracts connect smart contracts to the outside world. A dApp on Ethereum (or another chain) can push a request to a TEE program, which fetches APIs, performs heavy or private computation, and then responds on-chain with a verifiable result. Starter kits include oracles, VRF templates, and integrations with popular web3 developer tools. (github.com)

Industry integrations and tooling

Phala provides developer guides, examples, and audits for confidential workloads. The stack emphasizes portability—self-hosted or managed—so different teams can meet their own compliance and performance needs. NVIDIA documentation and Intel’s TDX materials illustrate the underlying hardware foundations for this kind of computing. (github.com)

Advantages & Challenges

Advantages

• Verifiable privacy: Phala’s Trust Center surfaces live attestations so others can check what ran, where, and under which code hash—beyond simple “trust the provider” claims. (trust.phala.com)
• Developer-friendly deployment: dstack allows Docker-based apps to run in TEEs, with key management, attestation, and templates packaged for quick starts. (github.com)
• AI-ready infrastructure: Support for TDX CPUs and NVIDIA GPU TEEs enables confidential AI inference and agent workloads at modern performance tiers. (images.nvidia.com)
• Mature off-chain patterns: Phat Contracts make it practical to combine on-chain logic with off-chain, privacy-preserving computation and web connectivity. (phala.com)
• Compliance posture: The project highlights SOC 2 Type I and HIPAA status for its cloud offering, useful in regulated settings. (phala.com)

Challenges

• Specialized hardware: Relying on TEE-enabled CPUs and GPUs can limit flexibility compared to commodity instances and requires careful operational discipline. (intel.com)
• Technical learning curve: Attestation, enclave boundaries, and key flows are novel concepts for many teams; despite tooling, it takes time to master them. (docs.phala.network)
• Evolving architecture: The 2025 migration simplified access but also introduced a new split (Ethereum L1 staking, L2 GPU mining) that developers must understand. (forum.phala.network)
• Competitive landscape: Other privacy and TEE-based platforms (and ZK approaches) are also progressing, pushing everyone to keep improving model proofs and usability. (en.wikipedia.org)

Where to Buy & Wallets

PHA is available on major centralized exchanges. Phala Network can be purchased on Binance, which announced a PHA listing and supports various buy methods depending on region. PHA is also listed on Kraken and KuCoin. Always check each exchange’s supported deposit/withdrawal networks before transferring. (binance.com)

PHA is an ERC‑20 token on Ethereum. EVM-compatible wallets such as MetaMask and other WalletConnect-enabled wallets can hold, send, and receive PHA. The canonical ERC‑20 address is published on Etherscan and is also surfaced in Phala’s app interface. (etherscan.io)

PHA and vPHA can be moved between Ethereum and Phala’s L2 using the official Phala L2 Bridge. The bridge site and docs provide guided steps for selecting the from/to networks and confirming transactions. (bridge.phala.network)

Regulatory & Compliance

Phala’s compute layer is built with verifiable privacy in mind and is presented with a public Trust Center that shows attestation evidence. For enterprise buyers, the team reports SOC 2 Type I and HIPAA compliance for its managed cloud, and it positions its stack as privacy-by-design to help customers navigate GDPR-like data protection regimes. These claims are supported by a trust page, a dedicated announcement, and downloadable materials. (phala.com)

From a token perspective, third-party MiCA documentation categorizes PHA under “other crypto-assets” (not e-money or asset-referenced), a common bucket for utility tokens providing access to network services. While such documents are prepared by individual trading platforms for their compliance needs, they offer a useful frame for how European rules may view the asset type. (cdn.21mgroup.com)

Regarding Islamic finance, many analysts consider PHA halal because it enables productive, technology-based services—private computation and confidential smart contract execution—rather than interest-bearing lending or speculative instruments without utility. The token supports payment for compute, staking for access prioritization, and governance, all tied to real value exchange in a cloud-like network. This functional alignment is typically cited in rulings that view technology infrastructure tokens as permissible. (github.com)

Future Outlook

Phala’s direction centers on making private AI “feel normal to build, and difficult to fake.” Expect continued work on:

• Better developer ergonomics: faster templates, richer SDKs, and more turnkey data- and model‑handling inside TEEs. (docs.phala.network)
• Stronger, portable attestations: tying domain ownership, OS measurements, code hashes, and GPU/CPU proofs into a single receipt so partners can verify without bespoke setups. (trust.phala.com)
• Scaled GPU capacity and marketplaces: connecting more TDX-capable servers and Confidential GPUs to meet AI inference demand while keeping privacy intact. (github.com)
• Ethereum + L2 alignment: refining staking, L2 mining, and governance processes that fit the habits of broader EVM-native builders and communities. (forum.phala.network)

As confidential computing matures and more clouds expose TDX/SEV-SNP and GPU TEEs, the total addressable market for verifiable private compute is likely to expand—especially in healthcare, finance, and enterprise AI SaaS where proofs reduce due-diligence friction. Phala’s open-source stance with dstack and its public Trust Center positions it well to serve teams that need both privacy and proof. (intel.com)

Summary

Phala Network’s promise is to make privacy a first-class feature of modern apps and AI systems—without asking users to take a provider’s word for it. By running workloads in TEEs (TDX CPUs and Confidential GPUs) and publishing verifiable receipts, Phala turns “trust me” into “prove it.” The PHA token underpins the network’s compute economy and governance, now anchored on Ethereum and an L2 following the 2025 migration. Developers can self-host with the open-source dstack framework or choose the managed Phala Cloud, then connect off-chain logic to on-chain systems using Phat Contracts and standard web tooling. For readers exploring infrastructure tokens, Phala offers a clear, technology-driven story: confidential compute, public proof, and a growing toolkit for building real applications that keep sensitive data private while remaining auditable. (forum.phala.network)