Phala Network - 2024 Year in ReviewAbout PhalaRecent PostsSimilar Posts

Phala Network - 2024 Year in Review

INTRO

2024 was a year where AI, blockchain, and privacy met their match in a fusion of engineering brilliance. It is a game-changing year for Phala Network, as Phala’s innovative blockchain-TEE hybrid design sets the stage for decentralized AI computing. By seamlessly blending secure offchain processing with onchain smart contracts, Phala has carved out a niche as the go-to AI coprocessor for the decentralized web.

In this Year in Review, we’ll take you on a journey through Phala’s achievements in 2024 - how we scaled, innovated, and set the benchmark for integrating AI into Web3. If you thought the blockchain world couldn’t get any smarter, think again.

Let’s go.

PHALA NETWORK KEY FEATURES

It has become clear that by the end of 2024, Phala Network has risen to become the go-to infrastructure for decentralized AI projects, with its standout feature being the 'Phala TEE'.

What is TEE?

TEE stands for Trusted Execution Environment and has become a cornerstone of decentralized AI due to their ability to provide strong security guarantees in a decentralized setting. TEEs allow sensitive computations to be carried out in isolated environments, ensuring that data remains confidential and protected from malicious actors - even within untrusted environments. In the context of decentralized AI, TEEs are crucial because they enable private data processing while maintaining the integrity and authenticity of AI models. This ensures that organizations and individuals can confidently use decentralized AI systems without sacrificing privacy, creating trust in the system’s ability to handle confidential and proprietary information securely.

Phala’s TEE technology goes beyond traditional hardware-only or software-only solutions by combining both software and hardware-based TEEs, enhancing security and privacy for decentralized AI. Phala’s Key Management System (KMS) introduces an extra layer of protection by constantly rotating management keys, preventing long-term data exposure and enhancing overall security. Together with the decentralized Root-of-Trust DePin network, which distributes trust across various nodes, this technology safeguards both data and computation, making it ideal for AI applications where confidentiality is paramount.

Stepping into Phala 2.0.

Phala Network is stepping into a transformative era with Phala 2.0, marking a major leap forward in decentralized, secure computation. This evolution is powered by the GPU TEE network, an upgrade from the previous CPU-based TEE infrastructure, enabling higher processing power and efficiency to handle demanding AI workloads. Phala has also integrated nVIDIA Confidential Computing Technology (with TDX), which ensures that sensitive GPU-based computations remain isolated and protected, even if the host machine is compromised. This provides developers with an unparalleled combination of high computational performance and robust security, making Phala the ideal platform for AI and privacy-preserving applications.

Alongside this upgrade, Phala will launch the a Layer 2 rollup on Ethereum, a groundbreaking achievement made possible through strategic partnerships with Succinct Labs and Conduit. This rollup delivers fast finality, low-cost transactions, and cryptographic security. By unifying TEE infrastructure with scalable ZK proofs, Phala is redefining Ethereum's scaling landscape while providing developers with tools to deploy private, AI-ready applications through Phala Cloud. With simplified deployment, real-time monitoring, and trust verification, Phala Cloud makes accessing GPU and CPU TEEs seamless and scalable.

These advancements solidify Phala’s position as a leader in decentralized cryptographic computing, offering developers cutting-edge tools for AI, zk-TLS, and enterprise-grade private computations. Phala 2.0 not only bridges Web2 and Web3 but also sets a new standard for scalable, secure, and verifiable computation solutions in the blockchain space.

Dstack for Developers

Phala’s Dstack is an experimental and rapidly evolving SDK designed to facilitate the deployment of Docker-based TEE applications. It leverages Intel’s Trusted Domain Extensions (TDX) to run Confidential Virtual Machines (CVMs), ensuring a secure environment for sensitive workloads. The SDK integrates several services, including teepod for CVM management, tproxy for secure TLS connections, and KMS for key management. Dstack enables developers to securely deploy and scale TEE applications using a simple Docker Compose setup.

Dstack also provides an interface for managing certificates, passing encrypted environment variables, and securely handling secrets. Dstack is designed to streamline the deployment and management of TEE applications while ensuring secure communication and scalability, with the goal of making it a production-ready solution in the future.

Phala’s TEE-as-a-Service

Phala now offers TEE-as-a-Service, providing decentralized AI projects with secure, scalable computing infrastructure. This service integrates various TEEs such as TDX, SGX, or GPU-based solutions, offering end-to-end support for seamless TEE deployment. By leveraging Phala's TEE, developers can integrate secure, decentralized computing into their backend, with three primary deployment models: Dstack for self-managed server deployments, Phala assisting with TDX-based TEE deployments, and GPU TEE deployments with Phala's support.

Phala's TEE-as-a-Service has already been utilized by several notable projects such as Flashbots, Near, and Nillion, among others. Partners like 0g, Ritual AI, ai16z and Morpheous have also implemented these solutions to enhance their applications. This service enables developers to easily adopt secure, decentralized computing while benefiting from Phala’s expertise in managing and deploying TEE technology.

2024 KEY METRICS

Network Metrics

• AI Agent Contract Execution: In 2024, Phala Network experienced tremendous success with its AI Agent Contract, which saw impressive growth in adoption and usage. Starting with approximately 150 contract executions at the beginning of the year, the number surged to around 4,500 by mid-May, remaining stable ever since. This is partially driven by the release of Phala's Dstack, an open-source, TEE-powered stack that enables developers to seamlessly integrate secure computing into their projects. As of today (Dec 10, 2024), the daily execution count for AI Agent Contracts has reached a remarkable 758,270 times. This achievement underscores Phala's pivotal role in advancing secure, decentralized AI technologies and offering robust tools for developers. ( https://dune.com/queries/3576989/6193020 )

• Workers: Phala Network's decentralized computing ecosystem is powered by its extensive network of workers (nodes), which are responsible for executing secure computations within the Trusted Execution Environments (TEEs). Starting the year with 34,997 workers, the total reached 37,650 by December 10, 2024, where 20,885 of them are hosted on Khala while 16,765 workers on Phala mainnet. ( https://dune.com/queries/3572546/6013980 )

• DAU: Phala's on-chain addresses demonstrate consistent growth across both the Khala and Phala Networks, contributing significantly to the total DAU. On the Khala Network, the number of on-chain addresses started at 30,000 at the beginning of the year and has steadily increased to 36,000 by now. Meanwhile, on the Phala Network, addresses grew from 5,000 at the start of the year to approximately 6,900 currently. Combining these figures, the total on-chain addresses across both networks now stand at 42,900. ( https://khala.subscan.io/tools/charts?type=total_account , https://phala.subscan.io/tools/charts?type=total_account )

Financial Metrics

• Rewards for Operators (USD): As of December 10, 2024, the total mining rewards for 2024 amount to 110,642,177.83 $PHA. Currently, the daily reward is approximately 135,000 $PHA. With 20 days remaining in the year, the total is expected to reach around 113,342,200 $PHA by year-end. ( https://dune.com/queries/3572546/7427382 )

• Total Staked (USD): At the beginning of the year, the total delegation of workers was valued at $280 million, with a robust stake ratio of 41.57%. However, the two halvings of $PHA in 2024 have had quite an impact on the Phala Network's total delegation and stake ratio. Following the halvings, which effectively reduced the reward incentives for staking, the total delegation has dropped to $183 million. This decline in worker delegation also led to a decrease in the stake ratio, which now stands at 23.85%. The reduced staking rewards have likely discouraged participants from maintaining their stakes, however, the halvings of $PHA is an essential step towards the decentralization and sustainability of Phala Network. ( https://dune.com/queries/3572546/6013894 )

• $PHA Token Holders: The number of $PHA token holders across the Phala ecosystem, including Ethereum, has demonstrated steady growth this year. On the Phala chain, token holders increased from 4,100 to 5,300, while the Khala chain saw an increase from 23,000 to 26,000. Additionally, there are currently 23,555 $PHA token holders on Ethereum. Combining these figures, the total number of $PHA token holders across Phala, Khala, and Ethereum now stands at approximately 54,855, reflecting the expanding reach and adoption of the $PHA token across multiple chains. ( https://phala.subscan.io/tools/charts?type=holder , https://khala.subscan.io/tools/charts?type=holder )

• $PHA Inflation Rate: Mining rewards account for most of $PHA inflation, estimated at approximately 17.2% in 2024.

PHALA 2024 MILESTONES

nVIDIA H100 Benchmark Study

In September 2024, Phala released its Performance Benchmark Study on the impact of enabling TEE on NVIDIA H100 GPUs for large language model (LLM) inference tasks. The primary focus of the study is on the overhead introduced by TEE, specifically the bottleneck caused by CPU-GPU data transfers via PCIe, which is the main performance limitation.

Through this study, we found out that while enabling TEE introduces minimal computational overhead within the GPU itself, the performance penalty stems from data transfer times. For typical LLM queries, the overhead remains under 5%, and for larger models and longer token sequences, the penalty is almost negligible. The performance impact is most pronounced with smaller models and shorter sequences but becomes trivial for larger models.

The report further examines the scalability of TEE with growing input sizes and model complexity, showing that larger models like Meta-Llama-3.1-70B and Phi-3-14B-128k exhibit significantly lower overheads. In particular, the throughput efficiency improves as the model size increases, with IO-related overhead becoming less significant. The efficiency of TEE increases substantially for larger models due to longer GPU processing times compared to IO time, which results in a nearly 99% efficiency in high-computation scenarios. This highlights the scalability of TEE mode, making it well-suited for secure, high-performance AI tasks in decentralized or enterprise applications.

nVIDIA H200 Benchmark Study

Following the success of the TEE-enabled GPU benchmark on NVIDIA H100, Phala Network, io.net, and Engage Stack conducted another study on H200 Hopper GPUs for large language model (LLM) inference tasks. The study found that while TEEs provide essential data security by isolating computations, they have a manageable impact on performance, especially for high-demand applications. The H200 showed a more noticeable impact on Time To First Token (TTFT) and Inter-Token Latency (ITL) compared to the H100, but both GPUs maintained high performance with minimal overhead.

The research highlighted that TEE’s overhead is under 7% for typical LLM queries, with larger models like Llama-3.1-70B experiencing almost no penalty. Despite slightly higher overhead on the H200, both GPUs demonstrated strong throughput, handling nearly 130 tokens per second for medium inputs. These findings confirm that TEE can secure real-time AI applications, making it viable for industries like finance, healthcare, and decentralized AI, where data protection and scalability are essential.

The Release of Dstack

The release of DStack marks a significant milestone for Phala in 2024, reinforcing its position as a leader in confidential computing. As a VM-level confidential compute solution, DStack enables developers to offload sensitive workloads into a secure environment, running TEE-protected computations in Docker containers with cryptographic verifiability. It provides a cloud-like experience akin to AWS EC2 but with enhanced security, leveraging Intel's Trusted Domain Extensions (TDX) to power Confidential Virtual Machines (CVMs).

This experimental and rapidly evolving tool not only streamlines the management of TEE applications but also ensures secure communication and scalability, paving the way for production-ready deployment in the future. With DStack complementing Phala's AI Agent Contract, the network now offers a versatile framework for running APIs and offloading computation securely, making it a cornerstone of Phala's 2024 advancements.

Phala Moving To Ethereum

In December, 2024, Phala Network proposed sunsetting the Khala chain and migrating Phala Network to Ethereum, signaling a strategic move to align with cutting-edge technologies and streamline operations, making it a major milestone in 2024 for Phala. With Intel's planned discontinuation of SGX-IAS support in 2025 rendering Khala workers obsolete, migrating to Ethereum ensures Phala can leverage advanced platforms like Intel TDX and Nvidia GPUs for secure and high-performance computations. This transition consolidates Phala’s ecosystem into a unified infrastructure, eliminating redundancy and enhancing scalability.

By deploying a Layer 2 (L2) network and a TEE marketplace on Ethereum, Phala strengthens its interoperability and deepens collaboration with the Ethereum ecosystem, while the introduction of an Ethereum Vault contract modernizes staking and asset management. The migration also includes a seamless process for transferring assets, ensuring users can easily reclaim their tokens and accrued rewards.

Ai16z TEE Integration

Collaborating with a16z to build Eliza’s TEE framework is a landmark achievement for Phala in 2024, as Eliza has emerged as the leading open-source AI Agent framework for Web3 developers. This partnership highlights Phala’s role in advancing privacy-preserving AI by integrating its Trusted Execution Environment (TEE) technology into Eliza’s multi-agent framework. Eliza enables the deployment of autonomous AI agents with consistent personalities across platforms like Discord, Twitter, and Telegram, offering cutting-edge features like a RAG memory system, media analysis, document processing, and autonomous trading. By leveraging Phala’s TEE, these capabilities gain cryptographic verifiability and enhanced privacy, ensuring secure and trustworthy interactions. The collaboration also aligns with Eliza’s extensible design, as Phala’s framework supports modularity and custom plugins, enabling seamless scaling and new integrations for developers. This milestone solidifies Phala’s position as a leader in secure AI infrastructure while fostering innovation in decentralized multi-agent AI systems.

PHALA 2024 ECOSYSTEM

Demand Side

The rise of artificial intelligence (AI) has brought unprecedented demand for confidential computing, where secure and private data processing is crucial. This surge in demand has highlighted the importance of trusted execution environments (TEEs) for ensuring data integrity, confidentiality, and security in AI operations. Phala's TEE technology provides a robust solution by enabling scalable, secure, and decentralized computation. With its DStack framework, Phala simplifies the adoption of advanced TEEs such as TDX and GPU-based TEEs, making it easier for AI projects to implement secure, high-performance computing environments. Here is an overview of how Phala has influenced the decentralized AI landscape:

• A Decentralized GPU Network with Hyperbolic io.net: In partnership with Hyperbolics, Phala integrates its confidential computing technology to ensure secure deployment and verification of AI models on the blockchain, while with io.net, Phala extends its decentralized AI vision by providing secure access to GPU resources for AI workloads. Both initiatives emphasize the importance of data privacy, security, and the democratization of AI, enabling developers, researchers, and AI startups to access powerful resources without the burden of costly infrastructure.

• Decentralized AI Model Training: Sentient leveraged Phala’s TEE technology to train AI models securely and verifiably using a GPU-based TEE environment. By integrating Phala's TEE support with Nvidia's H200 GPU and Intel Xeon CPUs, Sentient enabled confidential and verifiable training of Llama-3.1-8B-Instruct. The process involved setting up a TEE environment using Phala’s Dstack to create a Confidential Virtual Machine (CVM), where the training of AI models occurred without compromising data privacy. The use of TEE allowed Sentient to achieve high accuracy with minimal performance overhead, demonstrating the effectiveness of Phala's TEE technology in securely training large AI models with robust results.

• Decentralized AI Model Inference: Phala's collaborations with 0G and with Morpheus and Lumerin highlight its robust TEE-powered infrastructure for secure and verifiable AI execution. In both cases, Phala leverages its support for nVIDIA GPUs and Intel TDX to provide tamper-proof, confidential environments for AI tasks, ensuring data integrity through RA-TLS-secured communication and verifiable attestation reports. These integrations enable decentralized AI platforms to run large models like Llama3 securely, while cryptographic proofs ensure trust and transparency for users. Additionally, Phala simplifies node onboarding and service registration, demonstrating its role as a critical enabler for decentralized, privacy-preserving AI ecosystems. Similar collaborations are also between Phala and RitualAI, RedPill Near.

• Decentralized Social Media: Decentralized Social Media such as Teleport leverages Phala's Trusted Execution Environments (TEEs) to enable secure, one-time-use account delegation for Twitter, combining Web2 distribution with Web3 programmability. TEEs enforce Teleport's core guarantees, such as allowing only one post per generated link and mandating adherence to predefined content safeguards, which can integrate AI-driven policies like GPT-based filters or external conditions. By providing remote attestation, TEEs ensure verifiable execution, allowing users to trust the platform without requiring blind faith. Representing account links as NFTs adds Web3 composability, enabling these assets to be traded, auctioned, or reused in innovative ways. Teleport exemplifies how TEEs empower a new form of value exchange by bridging Web2's massive user base with Web3's decentralized infrastructure, laying the groundwork for cooperative AI interactions and redefining digital property rights in the information age.

• TEE-Powered Co-processor Combined with Zero-Knowledge Proofs (ZKP): The collaboration between Phala Network, Succinct Labs, and Conduit represents a powerful convergence of technologies, combining TEEs with ZKP to redefine Ethereum scaling and secure computation. Phala Network brings its expertise in privacy-preserving TEE infrastructure, Succinct Labs contributes cutting-edge ZKP capabilities through its decentralized proof generation systems, and Conduit delivers Rollups-as-a-Service for seamless Layer 2 deployment. Together, this partnership enables the first Op-Succinct Layer 2 rollup, offering fast finality, cost efficiency, and cryptographic security while unlocking new possibilities for decentralized AI, zk-TLS, and scalable private computation across Web3. TEE + ZKP = the future of trustless, scalable blockchain infrastructure.

• Decentralized Ethereum Block Building: Phala Network's TEE technology supports Ethereum MEV projects like Flashbots by providing a secure, confidential, and verifiable environment for transaction processing. By encrypting transaction data during block construction, Phala's TEE prevents frontrunning and sandwich attacks while enabling fair and deterministic ordering through cryptographic proofs. This ensures that Flashbots can build trustless, transparent, and efficient blocks, protecting sensitive trading strategies and minimizing centralization risks. Phala's infrastructure empowers Flashbots to tackle MEV challenges effectively, promoting a more secure, fair, and decentralized blockchain ecosystem.

• AI Agent: Phala has also pioneered fully autonomous AI Agents, such as the first fully autonomous AI Agent @TEE_HEE_HE on Twitter, eliminating human interference in AI decision-making processes. Additionally, ai16z built its open-source AI Agent framework, Eliza, based on Phala's TEE network, highlighting its versatility. Experimental applications like Agent Wars and Sxysama - a virtual AI Catgirl CEO - demonstrate the creative potential of Phala's TEEs in emerging AI use cases.

Supply Side

As of today (Dec 10, 2024), Phala Network hosts 37,648 active Worker Nodes and 3,269 delegators who have collectively staked a total of 182,788,105 $PHA. The average APR for 2024 varies between 22.32% and 12.10% ( https://dune.com/phala_network/phala-analytics ). Worker Nodes contribute compute resources, hosting AI Agent Contracts as off-chain programs inside their TEEs. The TEE ensures code integrity and data privacy, enabling a secure computation environment. Miners worldwide can join the network by deploying Worker Node clients on TEE-compatible hardware, which is verified on-chain.

The $PHA token is the native utility and governance token of the Phala blockchain, integral to its ecosystem. It powers AI Agent Contracts, facilitates seamless program execution, and rewards compute providers (Workers) for their contributions with transparent calculations to incentivize participation. $PHA also enables staking and delegation to support network security while earning rewards. Launched in September 2020 with a maximum supply of 1 billion tokens, 769 million (77%) are currently in circulation. The token supports blockchain validation, incentivizing Worker Nodes, application pricing, and governance.

In 2024, Phala completed two halvings: the 5th on June 6 and the 6th on December 3, reducing daily rewards for miners from 238,578 $PHA to 178,332 $PHA and from 177,892 $PHA to 134,930 $PHA, respectively. ( https://dune.com/queries/3572546/7427382 )

PHALA TREASURY

The Phala Treasury serves as a pivotal financial reservoir for the network, accumulating funds from various sources, including transaction fees, slashing penalties, and a 20% allocation from Secure Worker mining rewards. Designed to foster sustainable growth and innovation, the Treasury finances proposals aimed at improving infrastructure, bolstering network security, supporting ecosystem collaborations, advancing software development, and engaging the community. Proposals undergo a rigorous governance process, requiring a 5% deposit from stakeholders, which is either refunded or slashed based on the proposal's approval status. Unspent Treasury funds at the end of each budget period are currently exempt from burn penalties, maintaining the Treasury’s balance while allowing deflationary mechanisms to remain an optional governance tool. Ultimately, the Council wields the authority to allocate these funds, ensuring that the network’s strategic priorities align with its long-term vision.

In 2024, Phala Network's total spending on infrastructure amounted to 5,674,308 $PHA (USD 771,747). The community-focused initiatives had a total expenditure of 3,209,520 $PHA (USD 442,231), while marketing efforts saw a significant investment of 7,444,290 $PHA (USD 893,300). Phala’s 2024 spendings on Infrastructure, Community and Marketing, therefore, account for 37%, 20% and 43% respectively.

Treasury Proposals on Infrastructure

• Subscan Explorer Services:
• Proposals request funding for the operation and maintenance of Subscan Explorer services, including:
• Chain data collection and parsing, database storage, network egress bandwidth, monitoring, and technical support.
• Cost structures based on usage, with pricing tiers depending on data volume (up to 1 TB).

• Public RPC Services:
• Proposals from OnFinality and Dwellir focus on providing high-performance, scalable, and reliable public RPC endpoint services for Phala and Khala:
• OnFinality, as Polkadot's largest infrastructure provider, offers API services for both networks with millions of responses and high uptime (99.87% for Khala and Phala).
• Costs for providing these services are requested in advance with discounts for upfront payments (15%).
• The services help reduce costs for developers by providing access to low-latency infrastructure and autoscaling solutions.

• SubQuery Services:
• SubQuery provides enterprise-level managed services for hosting and maintaining data for Phala’s projects, ensuring reliable production data with 99.9% uptime.
• The proposal requests funding for operations covering the cost of compute hours, indexing services, and database access.

• Node Snapshots:
• RadiumBlock has been maintaining Khala snapshots and proposes a public snapshot service to speed up node syncing.
• This service will reduce the time required for new validator nodes to sync, from several days to a few hours.

• Runtime Upgrades:
• Various proposals outline runtime upgrades for Khala and Phala, which include:
• Adding fee support for inDEX, updating DCAP support, fixing bugs, and enhancing tokenomics (e.g., budget balancer updates).
• Improvements to the Khala Network’s functionality, including better performance and security for vault owners and gas fee management.

• General Infrastructure Support:
• A broader Infrastructure Support and Operational Costs proposal requests reimbursement for the Phala team's past investments in Gatekeeper nodes, Collator nodes, and public RPC services, ensuring the continued smooth operation and security of the network. The proposal covers hardware, data center, network, cloud services, and maintenance costs for both Phala and Khala.

Treasury Proposals on Community

• Koolex Treasury Reward:
• Community member Koolex identified a DoS attack vulnerability.
• Koolex is eligible for a reward of 8,000 USD (29,520 $PHA).
• Aimed to incentivize security contributions and protect Phala's network integrity.

• Ambassador Program
• 3.18M $PHA proposed and approved for the Ambassador Program in the year of 2024.
• The 3.18M $PHA was allocated for rewards to existing and new ambassadors, performance-based incentives, ambassador-led events, bounties, and ensuring operational stability to expand Phala's global community and boost engagement.

Treasury Proposals on Marketing

• Events: A total budget of 6,384,267 $PHA was proposed to be spent on events in 2024. That includes participation in EthDenver, Token 2049 Dubai, Coindesk Consensus, EthGlobal Hackathons, and Devcon in Bangkok. Received budget was spent on event sponsorships, event booths, travel, accommodations, SWAGs, and operational costs.

• Public Relations: A total of 798,717 $PHA was spent on funding for Messari's research services, including reports and analysis to boost Phala’s brand visibility.

SOCIAL MEDIA GROWTH

Phala Network experienced remarkable social media growth in 2024, showcasing its expanding reach and community engagement. Posts on X generated an impressive 5.6 million impressions, 318,200 engagements, 35,500 profile visits, 92,300 likes 15,200 reposts throughout the year. A highlight occurred on June 6, when Phala launched the popular AI Agent on Agent Wars, achieving a single-day record of 95,651 impressions, 1,327 likes, 4,460 engagements, and 12,279 media views. Other standout moments included the release of the Benchmark Study results on NVIDIA H100, H200, and TDX for LLMs, as well as partnership announcements with innovative collaborators like io.net and 0G.

Phala’s #60DaysofTEE campaign from Aug 9, 2024 to Oct 11, 2024 was a 60 days of resounding success, which highlighted the power and potential of TEEs through engaging social media and community activities. The campaign generated 28 impactful tweets that collectively amassed 267.9K impressions, 2,758 likes, 1,378 retweets, and 236 comments, showcasing its strong engagement and widespread reach. Additionally, the campaign integrated a Galxe engagement initiative, attracting 804 participants, with 23 winners celebrated for their contributions. By sparking discussions and driving awareness, #60DaysofTEE cemented Phala’s leadership in advocating for TEE technologies within the blockchain ecosystem.

In 2024, Phala published 51 blog articles covering a diverse range of topics. These included educational pieces explaining concepts like DePIN and Coprocessors, as well as technical guides for developers, such as building FHE Coprocessors on TEE using JavaScript. Mid-year, the focus shifted toward AI-centric content, addressing trust issues in decentralized AI infrastructure from Addressing Trust Issues in Decentralized AI Infrastructure to Phala's defense-in-depth solutions with TEE technology . Other highlights include discussions on the Truth of AI Agents and the innovative aICO (AI Initial Coin Offering), emphasizing Phala’s role in advancing secure and trustworthy AI ecosystems.

2024 Phala Social Media Highlight #1

TEEs are ZK for the Poor : In November 2024, Phala launched the impactful campaign "TEEs are ZK for the Poor", highlighting the significantly lower gas fees for AI developers using TEEs compared to ZK technology for offchain processing. The campaign has been a massive success, generating a remarkable 5.5 million impressions on X and sparking conversations around the cost-effectiveness of TEEs in decentralized computing.

2024 Phala Social Media Highlight #2

The first-ever GPU TEE Benchmark Study : Another major highlight for Phala in 2024 was when Phala shipped the first-ever GPU TEE benchmark, showcasing impressive results with LLaMa 3 and Microsoft Phi models tested on NVIDIA H100 GPUs. The benchmark revealed up to 150 TPS for LLaMa-3-8B models with minimal performance trade-offs of just 0.5% to 5% in TEE mode. This post generated a massive 65.5K views on X.

2024 Phala Social Media Highlight #3

60DaysofTEE : Phala Network’s #60DaysofTEE campaign has become a social media highlight of 2024, bringing together TEE enthusiasts and curious minds alike, generating 267.9K impressions on X. In collaboration with our partners, the campaign features exciting events such as Twitter spaces, livestream workshops and giveaways, all centered around the future of TEE technology in AI and beyond.

2024 Phala Social Media Highlight #4

aICO : Phala’s Co-founder, Marvin, published an article discussing the use of AI in ICOs to prevent human interference, such as rug pulls. This is another example of how AI, combined with TEEs, can enhance blockchain security and trust. 36.4K people viewed and hundreds have engaged with this post.

2024 Phala Social Media Highlight #5

Phala Featured in Messari’s Report : Messari introduces Phala Network as a blockchain and AI innovator, combining a hybrid blockchain-TEE design for secure, scalable computation. The report highlights Phala’s architecture, including decentralized computing, a TEE framework, AI agent contracts, and blockchain nodes, supporting over 4,000 contracts and 36,000 worker nodes. This is sure to be one of the social media highlights of 2024.

EVENTS AND PARTNERSHIPS

Events

• Phala at EthDenver : EthDenver 2024 showcased the vibrancy of the Ethereum ecosystem and its growing intersections with AI, cryptography, and decentralized technology. Phala Network stood out with highlights such as introducing its AI Agent Contract framework, promoting the rising potential of TEEs, and leading discussions on DePIN scalability. The team also made significant strides in community engagement, forging connections with researchers and builders across ecosystems.

• Token2049 Dubai : Phala Network made waves at Token2049 Dubai, showcasing its role as an AI coprocessor for blockchains. Phala Co-founder Marvin Tong participated in key panels on AI x Crypto convergence, decentralized societies, and the role of data in AI models, alongside industry leaders from EigenLayer, Sahara Labs AI, and Animoca Brands. The team also demonstrated AI’s potential in gaming and Web3 at Animoca's portfolio day, solidifying Phala’s position as a leader in privacy-preserving AI and blockchain innovation.

• Consensus2024 : Consensus 2024 was a vibrant event for Phala, showcasing the rise of AI and Web3 synergies. Phala team mentored the EasyA x Consensus hackathon, where developers created innovative projects like AI Agents generating 3D models with Phala’s framework. The buzz around DePIN and advanced tech like TEE, FHE, and ZK highlighted the push for secure, scalable Web3 infrastructure. Consensus 2024 strengthened Phala’s partnerships and reinforced its role in leading privacy-preserving, AI-driven blockchain solutions.

• EthCC Phala dAGI Hackathon in Brussels : At EthCC Brussels, Phala made a significant impact, with multiple team members sharing insights and engaging in various panels. At the #dAGIHouse Hackathon, out of 43 submissions, 10 projects leveraged Phala, with the following highlights projects.
• 1st Place: WarpAds – A display ads protocol for Farcaster that uses Phala’s #AIAgentContract to ensure secure, private ad data processing and royalty distribution.
• 2nd Place: Devdock – An AI IDE extension rewarding developers for writing quality smart contracts, using Phala’s AIAgentContract for secure data processing.
• 2nd Place: Jarvis – A multi-agent system using Phala’s AIAgentContract for collaborative AI interactions.
• 3rd Place: Open Market – A decentralized AI marketplace that uses Phala’s AIAgentContract for secure, transparent data and royalty management.

• Token2049 Singapore: EthGlobal Singapore : We had 26 projects submitted to Phala's bounty at the EthGlobal Singapore 2024, and our judges had a hard time selecting the winners! Of all the 26 teams, Hwakeye won 1st place, Grand Theft Nouns secured 2nd, and AuditGPT took 3rd place.

• EthGlobal SF : At ETHGlobal San Francisco, Phala Network shone brightly with 15 impressive submissions, including two that made the top 10 finalists. All 15 teams have incorporated Phala TEE Docker Simulator SDK, in some cases, AI Agent Contract, into their projects. Here are the highlights, starting with Phala’s standout projects:
• Phala’s 1st: Found in Translation - A decentralized system for real-time language translation using AI wearables.
• Phala’s 2nd: Root4.fun A memecoin launchpad offering tools for creating, buying, and selling memecoins with AI-generated designs.
• Phala’s 3rd: SuperPass A secure password management system eliminating the need to store keys on devices.
• Additional Highlights: DAOsaster, Betsy, and Gains Gladiator showcased innovation with AI-powered solutions, winning multiple bounties. Other highlights include dAIp, a composable DApp builder, and Chainsage, an AI-driven DeFi strategy optimizer for cross-chain ecosystems.

• DevCon EthGlobal Bangkok : At ETHGlobal Bangkok, Phala Network’s Trusted Execution Environments (TEEs) shone brightly, with 40 AI and TEE-focused submissions, 10 finalists, and 3 leveraging Phala’s tech. Winning teams included:
• BubbleWars: A multiplayer Telegram game using Phala TEE for secure AI verification and leaderboard commentary.
• Zubernetes (ZK8S): Revolutionizing container orchestration with Phala DStack SDK for secure workload execution.
• Cat In A Box: A DeSci and SocialFi fusion safeguarding private health data with Phala TEE.
• YumeRTS: A medieval multiplayer strategy game ensuring verifiable, fair gameplay via TEEs.
• SocialBounty: An AI-powered marketing tool grading tweets and paying for quality posts using Phala TEE.
• HeptagonAI: Collaborative AI training with verified computations powered by Phala TEE.

Partnerships in 2024 (in chronological order)

• With Flock.io : Phala Network and Flock.io are collaborating to democratize AI by combining Phala's privacy-preserving decentralized infrastructure with Flock.io's modular ML marketplace, enabling community-driven AI creation, ethical monetization, and enhanced accessibility for AI-powered applications.

• With AwesomeQA : Phala Network has partnered with AwesomeQA to receive 24/7 multilingual AI chatbot support on Phala’s Discord and website, enhancing user satisfaction and efficiency.

• With Pond : Phala Network and Pond have partnered to develop advanced AI-powered trading agents, combining Phala's decentralized AI-Agent Contract platform with Pond's innovative Graph Neural Network model to enable precise on-chain predictions and strategies for decentralized finance and Web 3.0 use cases.

• With GM Network : Phala Network has partnered with GM Network to combine their expertise in decentralized AI and AIoT, enhancing data, computing power, and application scenarios for modular, interoperable AI agents in the Web3 ecosystem.

• With Allora : Phala Network has partnered with Allora to enhance secure AI and ML operations, integrating Phala's TEE technology with Allora's decentralized network for private model training, secure inferences, and the creation of self-improving AI agents, setting new standards in privacy, security, and AI performance.

• With Artela : Phala Network has partnered with Artela to integrate its AI coprocessor into Artela’s blockchain, enabling the creation of AI-powered, intent-centric dApps with seamless on-chain and off-chain AI modules, driving innovation in intelligent decentralized applications.

• With Morpheus and Lumerin : Phala, Morpheus, and Lumerin have teamed up to revolutionize decentralized AI, with Phala’s TEE compute capabilities powering the Morpheus platform and integrating into Lumerin’s architecture, enabling secure, scalable, and verifiable Smart Agents for web3 users.

• With Hyperbolic : Hyperbolic has partnered with Phala to combine their strengths in verifiable AI computing, integrating Phala's secure off-chain computation technology with Hyperbolic's decentralized marketplace to enable the secure deployment and verification of AI models on the blockchain.

• With Aligned Layer : Phala Network has partnered with Aligned Layer to enhance secure, confidential computation in Web3 by integrating TEE attestation, zero-knowledge proofs, and expanding AI use cases for privacy-preserving applications.

• With theoriq : Phala has partnered with Theoriq to enhance decentralized AI agents by integrating Phala's secure TEE capabilities with Theoriq's protocol, enabling AI agents to seamlessly interact with blockchains and ensuring privacy, security, and efficient deployment in Web3 environments.

• With io.net : Phala has partnered with io.net to enhance decentralized AI by integrating Phala’s secure TEE technology with io.net’s cloud-based GPU resources, enabling high-performance AI workloads with advanced cryptographic protections for secure, scalable, and accessible AI computation.

• With Ava Protocol : Phala Network has partnered with Ava Protocol to integrate Phala’s TEEs into Ava's automation framework, boosting confidentiality and security for DeFi and Web3 applications while protecting against MEV attacks.

• With 0G : Phala Network has partnered with 0G to integrate its Trusted Execution Environment (TEE)-based SDK into 0G’s decentralized AI Operating System, ensuring confidential AI inference, data privacy, and verifiable execution for large language models (LLMs) in a secure, decentralized environment.

• With NeurochainAI : Phala Network is partnering with NeurochainAI to provide GPU TEE compute infrastructure, enabling scalable, secure, and cost-effective AI development and machine learning resources, empowering NeurochainAI to enhance its offerings and accelerate the growth of AI applications.

• With Vana Network : Phala Network and Vana Network are partnering to enhance decentralized data privacy and security by integrating Phala’s TEE-powered solutions with Vana’s user-driven data contribution platform, ensuring secure, verifiable data handling and empowering individuals to monetize their data while supporting decentralized AI advancements.

2025 LOOKOUT

In 2025, Phala Network will focus on building Phala 2.0, an upgraded platform that integrates GPU TEE technology to deliver higher processing power and enhanced security for AI and privacy-preserving applications. This evolution, alongside the launch of the Op-Succinct Layer 2 rollup on Ethereum, will redefine decentralized cryptographic computing with scalable, secure, and verifiable solutions for developers.

Real Code for Real dAGI

Phala Network’s TEE technology has been integrated across all major sectors of blockchain and AI, from decentralized AI model training and inference to secure Ethereum block building and decentralized social media. In 2025, Phala aims to strengthen its market share in the TEE infrastructure space by introducing Dstack to more developers, continuing to drive innovation and ensure privacy, security, and scalability for decentralized applications.

TEE-as-a-Service

Phala is committed to developing Dstack as the best open-source TEE SDK, empowering developers with robust tools for building secure and privacy-preserving applications. Simultaneously, we will launch Phala Cloud, an AI-ready platform designed for effortless deployment and management of TEE applications. With features like one-click Docker deployment, real-time monitoring, trust verification, and flexible billing, Phala Cloud will simplify access to secure computing, enabling developers to focus on innovation without worrying about infrastructure complexities.

Rebuilding Phala’s Success

Phala is rebuilding its success with Phala 2.0, a transformative upgrade built on Ethereum to seamlessly integrate into the vibrant Ethereum ecosystem. By leveraging Ethereum's Layer 2 scalability, Phala 2.0 addresses key opportunities in AI-native use cases, cryptographic solutions, and TEE-based infrastructure. This transition enhances tokenomics by aligning supply and demand, with $PHA becoming the universal currency for AI services, enabling AI agents to autonomously pay for themselves and other services. With advanced TDX support and NVIDIA GPU integration, Phala 2.0 offers unparalleled scalability and performance, ensuring developers can deploy high-demand AI applications efficiently. Ethereum’s strong product-market fit and its ecosystem's compatibility with Phala’s TEE solutions make it the ideal foundation for Phala 2.0, fostering seamless testing, payment, and verification processes for users and reinforcing Phala’s leadership in decentralized, secure computing.

CLOSING

As we look back on an incredible year of innovation, growth, and milestones, we are deeply grateful to our community, partners, and supporters who made it all possible. From pioneering advancements with Phala 2.0 to empowering developers with cutting-edge TEE and AI technologies, 2024 has been a testament to what we can achieve together. But this is just the beginning - 2025 holds even greater promise as we continue to push boundaries, expand our ecosystem, and bring transformative solutions to the blockchain and AI landscapes. We invite you to keep adventuring with us, exploring new frontiers and shaping the future of decentralized, secure computation. Let’s make 2025 another year to remember.

SOURCE

https://khala.subscan.io/tools/charts?type=total_account https://phala.subscan.io/tools/charts?type=total_account

https://etherscan.io/token/0x6c5bA91642F10282b576d91922Ae6448C9d52f4E

https://khala.subsquare.io/treasury/proposals