Tinfoil · Jul 2026
Tinfoil launched cached prompt pricing for the Inference API and explained how client-side cache secrets prevent even Tinfoil from probing customer cache namespaces.
Tinfoil makes your AI workflows secure, verifiable, and private.
Company research
Public company, workplace, funding, and market signals
Overview
Tinfoil is a San Francisco-based private AI company that builds verifiably private chat, an OpenAI-compatible inference API, and confidential-computing containers that run AI workloads inside secure hardware enclaves so users can verify the operator cannot access their data.
Primary product
Private Inference API
Founded
2024
Headquarters
San Francisco, California, United States
Team size
1-10 employees
Industry
Software Development
Sub-industry
Private AI / confidential computing / AI infrastructure
Offices
San Francisco, California, United States
Los Angeles, California, United States
Business model
subscription
usage-based
enterprise SaaS
Funding
Stage
seed
Total raised
$7.8M
Latest round
Venture · May 2026
People & investors
Leadership
Co-founder & CEO
Co-founder
Co-founder
Founding Engineer
Founding Engineer
Head of Operations
Work & culture
Mission-driven, technical, open-source, and unusually transparent; the company emphasizes building in public, independent builders, and security-critical systems that are auditable rather than based on trust.
Compensation
No public employee salary bands were found. Public pricing signals are product-side only: Private Chat is $20/month, the inference API is usage-based, and Tinfoil Containers are $20/month plus usage.
Product & market
Pricing
Consumer subscription for chat, usage-based pricing for the inference API, and subscription plus usage for containers; enterprise plans are also available.
Differentiators
- —Hardware-verifiable privacy enforced by attestation
- —Open-source security-critical stack
- —Client-side verification on every connection
- —OpenAI-compatible drop-in API
- —Multi-GPU confidential computing support
- —Transparent supply-chain security via Sigstore
- —Docker workloads inside secure enclaves
- —Client-side secrets for secure prompt caching
Technology
NVIDIA confidential computing GPUs
Hopper GPUs
Blackwell GPUs
AMD SEV-SNP
Intel TDX
secure enclaves
Ubuntu-based confidential VM images
Sigstore
dm-verity
HPKE
TLS key binding
Docker
OpenAI-compatible API
Python SDK
JavaScript SDK
Go SDK
Rust SDK
Swift SDK
Customers
Workshop Labs
UC Berkeley
The Open Anonymity Project
Pour Demain
Competitors
Azure Confidential Containers
local/on-prem AI deployments
general-purpose cloud AI providers
Apple Private Cloud Compute
Scale & contact
Estimated revenue
~$0.65M-$2.0M ARR (third-party estimate; public sources vary)
Estimated monthly visits
36.8K
Traffic estimate as of Jun 2026
Latest news
Tinfoil · Jun 2026
Tinfoil published benchmark results for confidential computing overhead on Hopper and Blackwell across inference and training workloads.
Tekedia · May 2026
Tekedia Capital announced an investment in Tinfoil and framed privacy-preserving AI as an important infrastructure layer for the AI economy.
Tinfoil · Mar 2026
Tinfoil launched Containers for end-to-end verifiably private AI services, extending enclave protections from inference to application backends and training pipelines.
Tinfoil · Jan 2026
Tinfoil described its private web-search design, including enclave-based query generation and PII safeguards for searches triggered from chat.
Jobs and careers at Tinfoil
0 jobs at Tinfoil
No open positions
Check back later for new openings