Phala
@PhalaNetwork
Followers
143K
Following
7K
Media
2K
Statuses
6K
Empowering 10,000 builders and companies to build scalable, private, and safe intelligence.
San Francisco, CA
Joined August 2019
We love a layer of trust when it’s provable. Verified, hardware-backed confidentiality is the standard enterprise AI actually needs. 🙌 Congrats on the launch of the Confidential AI Gateway for Enterprise.
Following yesterday's momentum, today we're proud to announce another layer of trust for our users. We've partnered with @NEARProtocol to integrate their confidential computing infrastructure into the OLLM Gateway. This is what happens when leaders in private AI join forces.
44
46
150
⏰ Only 7 days left! Complete 3 quick tasks, earn Primus AlphaNet Scores, and compete for a 90,000 $PHA reward pool. Christmas came early this year. 🎅
1/ Primus AlphaNet is still live — come join us! 🔗: https://t.co/V3sddYfDVk Primus AlphaNet gives you control over your own data with better security and fairness, making it easier to monetize. Jump in now and start building your verified Web3 identity! Stage 0 includes
1
4
53
2025 is the year confidential compute starts to become part of the standard AI stack. 2026 is the year privacy becomes the default.
4
12
77
It's almost 2026. AI apps will either: > consult Phala to build their own private infra > or just use Phala there's no third option.
5
10
54
If you’re building local AI, agents, or privacy-first infra, Doubao is a concrete blueprint for what a TEE-first AI phone looks like at scale. We’ve translated and annotated the official whitepaper — https://t.co/1GqHrEzcZk Read the blog:
phala.com
Phala is the new cloud for confidential AI helping build AI people can trust.
1
1
12
This is exactly the architecture we’ve been betting on: > TEEs as the root of trust > confidential computing as the default runtime > clear separation between model capability and data governance > verifiable agents instead of opaque black boxes.
2
0
14
When the phone must call the cloud, the design shifts into confidential AI mode: > end-to-end encrypted links > per-session one-time keys > user & device IDs pseudonymized > inference running in confidential-computing containers > user data not used for training by default
1
0
5
ByteDance built the phone around a full end-to-cloud security model where: > On-device AI runs inside a Trusted Execution Environment (TEE) > Network traffic is fully encrypted + pseudonymized > Cloud inference runs inside confidential-computing containers This is a zero-trust
1
0
4
Doubao Phone is ByteDance’s AI-native smartphone, built around Doubao Phone Assistant. And @TaylorOgan is right: This AI Assistant has COMPLETE control over your phone — How is this even safe? Answer: It's safe because it's built around TEEs. https://t.co/l9s6z4jhBo
Another DeepSeek moment. This is the world’s first actual smart phone. It’s an engineering prototype of ZTE’s Nubia M153 running ByteDance’s Doubao AI agent fused into Android at the OS level. It has complete control over the phone. It can see the UI, choose/download apps,
10
7
63
Thrilled to be highlighted in @MessariCrypto's State of AI 2025 report as the Privacy & Verification layer for AI. Big year ahead! 🦾
6
15
98
Dive in: Pricing: https://t.co/Z3jxTxYnrj Billing: https://t.co/sJcg0spuYK Instance Types:
cloud.phala.com
Explore Phala Cloud's instance types for compute and GPU workloads.
0
2
14
New on Phala Cloud: ☁️ Crystal clear pricing docs ☁️ Revamped billing explanations ☁️ Instance directory with real-time hourly rates, specs, disk charges, etc. See costs before you deploy, understand your bill, pick the right instance without guessing.
6
10
54
Congrats on the launch of V2. Glad to see V2 running laps on Phala’s confidential compute. Honored to be the engine block under the hood. 🦾
Zero-code. Zero emotion. Fully Autonomous. V2 agents aren't an upgrade; they are a different species. Think racecar vs horse and buggy. Dedicated to perps on @HyperliquidX, sharper edges, colder execution, and much more adrenaline. Built on the most robust stack in agentic
8
11
101
We all know it: Telegram is home base, but it's also chaos. Dory fixes that: chat & user management, AI insights, nudges — all happening inside Phala’s TEEs that keep plaintext out of reach, giving you clarity, structure and privacy. What happens in TEEs stays in TEEs.
10
10
87
NEAR tech stack for AI agent developers: - @near_ai cloud for private and verifiable AI inference - Shade Agents for running verifiable agents (services) using @PhalaNetwork and @near_ai cloud - NEAR Chain Signatures - to manage cryptography that belongs to contract / shade
12
27
203
Retention check: 55% of Phala cloud developers deploy a 2nd CVM after the 1st CVM goes live.
13
10
54
DeepSeek v3.2 just landed on Phala. 🧠⚡️ You can now run the latest DeepSeek model as Confidential AI inside GPU TEEs – perfect for research, agents & onchain use cases. Try it on :
redpill.ai
DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a...
8
15
95