The privacy technology startup landscape in 2026 bears little resemblance to its 2020 predecessor. Five years ago, “privacy startup” almost exclusively meant compliance tooling – companies building dashboards to manage cookie consent, automate data subject access requests, and generate GDPR documentation. That market has matured, consolidated, and largely been absorbed into broader enterprise software platforms. The companies worth watching in 2026 operate at the architectural layer: building infrastructure where privacy is a structural property rather than a policy overlay.
This is a curated list, not a comprehensive directory. We have selected 18 companies across six categories based on technical differentiation, funding trajectory, market traction, and the structural significance of the problem they address. Companies that merely add privacy features to existing architectures are excluded. The threshold for inclusion is architectural: does this company change the fundamental relationship between the infrastructure and the data it processes?
Encrypted Infrastructure
Evervault (Dublin, Ireland)
Evervault has built an encryption-as-infrastructure platform that allows developers to encrypt sensitive data at the point of collection and process it in encrypted form through Evervault’s relay network. The core technical innovation is “Cages” – isolated processing environments where encrypted data is decrypted, processed, and re-encrypted without the plaintext ever being accessible to Evervault’s infrastructure or the customer’s broader application stack.
Founded in 2019, Evervault raised a $22 million Series A led by Molten Ventures in 2022 and a $38 million Series B in late 2024. Revenue is estimated at $15-20 million ARR, primarily from fintech and healthtech customers who need to process sensitive data (card numbers, health records) without exposing it to their own servers.
The significance of Evervault’s approach is that it shifts encryption from a storage concern to a processing concern. Traditional encryption protects data at rest and in transit but requires decryption for any computation. Evervault’s Cages create a third state – data that is decrypted only within an attested, isolated environment and never in the customer’s or Evervault’s general infrastructure. The architectural similarity to confidential computing is deliberate, but Evervault achieves it at the application layer rather than the hardware layer, making it accessible to startups that cannot afford dedicated TEE infrastructure.
Zama (Paris, France)
Zama is building open-source fully homomorphic encryption (FHE) tools that allow computation on encrypted data without decryption. The company’s Concrete framework provides a compiler that transforms standard programs into FHE-compatible equivalents, dramatically lowering the barrier to implementing homomorphic encryption.
Zama raised $73 million in a Series A led by Multicoin Capital and Protocol Labs in 2023, one of the largest early-stage rounds in European privacy tech history. The company has since released Concrete ML (for encrypted machine learning) and fhEVM (for confidential smart contracts on Ethereum), establishing itself as the leading commercial FHE implementation.
The practical limitation of FHE – computational overhead that makes it 1,000-10,000x slower than plaintext computation – remains significant, but Zama’s benchmarks show the gap closing rapidly. Their 2025 benchmarks demonstrated encrypted logistic regression at 47x overhead (down from 820x in 2023), suggesting that practical FHE for common workloads may be feasible within two to three years. If that trajectory holds, FHE will make it possible to use cloud infrastructure without ever exposing plaintext data to the cloud provider – a structural shift that the hyperscalers’ revenue model is not designed to accommodate.
NymTech (Neuchatel, Switzerland)
Nym is building a mixnet – a decentralized network that provides network-layer privacy by routing packets through a series of mix nodes that shuffle, delay, and re-encrypt traffic to prevent traffic analysis. Unlike VPNs (which centralize trust in a single provider) or Tor (which is vulnerable to traffic correlation by well-resourced adversaries), Nym’s mixnet is designed to provide privacy against global passive adversaries capable of observing all network traffic simultaneously.
Nym raised $32 million across multiple rounds, with the most recent $18 million round in 2024 led by Polychain Capital. The network launched its mainnet in 2024 and is integrated into several privacy-focused applications and wallets. The NYM token provides economic incentives for mix node operators, creating a self-sustaining infrastructure layer.
Nym’s relevance extends beyond cryptocurrency. Any application that needs to prevent metadata surveillance – including private AI interactions – benefits from network-layer privacy. The combination of application-layer encryption (which protects content) with network-layer mixing (which protects communication patterns) creates a privacy stack that addresses both the data and the metadata. This is relevant to any architecture, including zero-knowledge systems, that seeks to prevent adversaries from inferring information from traffic patterns even when content is encrypted.
AI Privacy and Governance
Opaque Systems (Berkeley, California)
Opaque Systems has commercialized technology from the UC Berkeley RISELab to enable confidential analytics and AI on encrypted data using a combination of hardware TEEs and cryptographic techniques. The platform, built on Apache Spark, allows organizations to run SQL queries and machine learning workloads on data that remains encrypted throughout processing – even from the infrastructure operator.
Opaque raised a $22 million Series A in 2022 and a $35 million Series B in 2024. The company has landed several significant enterprise customers in financial services and healthcare, where the regulatory requirement to analyze data without exposing it to third parties aligns directly with Opaque’s capability.
What distinguishes Opaque from other confidential computing approaches is its focus on collaborative analytics – scenarios where multiple parties need to analyze combined datasets without any party revealing its raw data to the others. Pharmaceutical companies collaborating on drug discovery, financial institutions sharing fraud detection signals, and healthcare providers combining patient populations for research all represent use cases where the data is too sensitive to share but too valuable not to analyze. Opaque’s encrypted analytics make the collaboration possible without the exposure.
CalypsoAI (San Antonio, Texas)
CalypsoAI has emerged as one of the best-funded AI security platforms, with a $100 million Series B in 2024 at a reported $500 million valuation. The company provides an AI model security platform that monitors, audits, and controls how AI models interact with sensitive data. Core capabilities include prompt injection detection, data leakage prevention for AI interactions, model behavior monitoring, and policy enforcement across multiple AI model providers.
CalypsoAI’s initial customer base was heavily weighted toward US defense and intelligence community clients, where the classification level of data interacting with AI models creates extreme sensitivity. The company has since expanded into financial services and healthcare, and its platform now supports integration with all major foundation model providers (OpenAI, Anthropic, Google, Meta) as well as self-hosted models.
The company’s growth trajectory reflects the broader corporate AI espionage concern: as organizations deploy AI tools that interact with their most sensitive data, the need for a security and governance layer between the organization and the AI model becomes urgent. CalypsoAI occupies this layer, providing visibility and control that the AI model providers themselves do not offer.
Private AI (Toronto, Canada)
Private AI has built a PII detection and redaction engine specifically optimized for the AI pipeline. While general-purpose PII detection tools exist, Private AI’s system is designed for the unique challenges of AI interactions: detecting sensitive information in natural language prompts, redacting it before the prompt reaches the AI model, and re-injecting the original values into the response for display to the user.
The company raised $12 million in Series A funding in 2023 and has grown to an estimated $8 million ARR. Private AI’s detection engine supports over 50 entity types across 53 languages and runs as a containerized service that can be deployed on-premises or at the edge, avoiding the need to send sensitive data to yet another cloud service for PII detection.
Private AI’s architecture maps closely to the PII proxy concept that is foundational to privacy-preserving AI interactions. The AI training tax – the risk that sensitive data in AI prompts becomes training material – can be substantially mitigated by stripping PII before the prompt reaches the model. Private AI provides this capability as a standalone service, but the concept is equally applicable as an embedded component within zero-knowledge AI architectures where PII detection runs client-side as a WebAssembly module.
Nightfall AI (San Francisco, California)
Nightfall provides a cloud-native data loss prevention (DLP) platform purpose-built for AI and SaaS applications. Unlike traditional DLP systems designed for email and endpoint monitoring, Nightfall integrates directly with AI tools (ChatGPT, GitHub Copilot, Google Gemini), cloud storage, and collaboration platforms to detect and prevent sensitive data from entering AI systems or leaving organizational boundaries.
Nightfall raised a $40 million Series B in 2023 and has expanded its platform to cover over 30 SaaS and AI integrations. The company reports scanning over 1.5 billion API calls per quarter, detecting sensitive data in approximately 8.3% of AI-related interactions – a figure that underscores the scale of the unintentional data leakage problem in enterprise AI usage.
Decentralized Identity
Spruce Systems (New York, New York)
Spruce is building decentralized identity infrastructure based on W3C Verifiable Credentials and Decentralized Identifiers (DIDs). The company’s SpruceID toolkit enables organizations to issue, verify, and manage digital credentials that users control without relying on centralized identity providers.
Spruce raised $34 million in a Series A led by a16z crypto in 2022 and has secured contracts with the State of California (for mobile driver’s licenses), the US Department of Homeland Security, and several major healthcare systems. The company’s Sign-In with Ethereum implementation is used by thousands of Web3 applications and is the reference implementation for the EIP-4361 standard.
Spruce’s relevance to the broader privacy landscape extends beyond Web3. Decentralized identity, implemented through verifiable credentials and zero-knowledge proofs, enables authentication without data collection – the ability to prove who you are (or what attributes you possess) without revealing your underlying identity data to the verifier. This is the authentication model that privacy-first infrastructure requires: proof without disclosure, identity without surveillance.
Dock.io (Remote / Swiss Entity)
Dock provides a verifiable credentials platform that enables organizations to issue tamper-proof digital credentials (academic certificates, professional licenses, employment verifications) that individuals can store in their own digital wallets and present to verifiers without the issuer being involved in the verification process.
Dock raised $20 million across multiple rounds and has processed over 40 million credential verifications since launching its production platform. The company’s blockchain-based credential registry provides a decentralized source of truth for credential status (valid, revoked, expired) without requiring a centralized database that tracks who is verifying what.
The privacy advantage of Dock’s approach over traditional credential verification is structural: in a conventional system, verifying a credential requires contacting the issuing authority, which reveals to the issuer when and by whom the credential is being checked. Dock’s architecture eliminates this surveillance vector by making verification a local computation against a public registry, with no callback to the issuer.
Anonyome Labs (Salt Lake City, Utah)
Anonyome Labs builds consumer privacy products that allow individuals to compartmentalize their digital identity. The company’s MySudo platform provides users with multiple digital “Sudos” – each with its own email address, phone number, virtual credit card, and browsing profile – enabling users to interact with different services under different identities without linking those identities together.
Anonyome raised $18 million in a Series B in 2024. The company reports over 2 million active users and processes approximately 50 million privacy-preserving transactions per month across email, voice, messaging, and browsing.
Anonyome’s approach is consumer-facing rather than enterprise-focused, but it addresses a fundamental architectural problem: the linkability of digital identities. Even when individual services protect user data adequately, the ability to correlate identities across services – through email addresses, phone numbers, or device fingerprints – creates a comprehensive surveillance profile. Anonyome’s compartmentalization model breaks this correlation by providing users with genuinely distinct identities for distinct contexts.
Confidential Computing
Anjuna Security (Palo Alto, California)
Anjuna provides a runtime security platform that enables applications to run inside hardware-based trusted execution environments (TEEs) – Intel SGX, AMD SEV, and ARM TrustZone – without modification. The platform abstracts the complexity of confidential computing, allowing organizations to deploy existing applications with hardware-enforced data protection against the infrastructure operator, cloud provider, and even privileged system administrators.
Anjuna raised $30 million in a Series B in 2023 and has focused on financial services and government customers. The company’s key technical contribution is reducing the engineering effort required to deploy applications in TEEs from months to hours, by providing a runtime that handles memory encryption, attestation, and secure key management transparently.
The significance of confidential computing for the broader privacy technology stack is that it provides hardware-enforced guarantees that software alone cannot. A zero-trust architecture built on confidential computing does not rely on the cloud provider’s promise not to access customer data – it relies on hardware that makes access physically impossible without detectable attestation failure.
Edgeless Systems (Bochum, Germany)
Edgeless Systems builds open-source confidential computing frameworks, including EGo (for developing SGX applications in Go), MarbleRun (a service mesh for confidential microservices), and Constellation (a Kubernetes distribution where all workloads run inside confidential VMs). The company’s mission is to make confidential computing the default for cloud-native applications rather than a specialized capability for high-security use cases.
Edgeless raised EUR 16 million in a Series A in 2024. The company’s open-source approach has generated significant community adoption, with over 5,000 GitHub stars across its repositories and deployment in production environments at several European financial institutions.
Edgeless’s Constellation product is particularly notable because it extends confidential computing from individual workloads to entire Kubernetes clusters. This means an organization can run its complete microservices architecture inside hardware-enforced encryption boundaries, with the cloud provider unable to access any workload data even at the hypervisor level. The combination of Constellation with a zero-persistence data layer creates a defense-in-depth architecture where data is protected both cryptographically and through hardware enforcement.
Secure Communications
Element (London, United Kingdom)
Element is the commercial entity behind the Matrix protocol, an open standard for decentralized, end-to-end encrypted communications. Element provides Matrix-based messaging, voice, and video calling for government and enterprise customers, with deployments in the French government (Tchap), the German military (BwMessenger), and NATO.
Element raised $46 million in a Series B in 2024. The company is differentiated from other encrypted messaging platforms by its federation model: organizations can run their own Matrix servers while maintaining interoperability with the broader Matrix network, eliminating the single-point-of-failure and single-point-of-trust that characterizes centralized encrypted messaging platforms.
The federation model is significant because it addresses a limitation that even the most privacy-focused centralized messaging services cannot overcome: the provider is always a potential point of compromise or compulsion. Even Signal, which implements best-in-class encryption, operates centralized servers that handle message routing and could theoretically be compelled to modify client software. Matrix’s federated architecture distributes this trust across multiple independent server operators, making systemic compromise substantially harder.
Wire (Berlin, Germany / Zug, Switzerland)
Wire provides end-to-end encrypted messaging, voice, and video for enterprise customers, with a particular focus on regulated industries and government. Wire’s encryption is based on the Messaging Layer Security (MLS) protocol, which provides post-compromise security and efficient handling of large group communications – capabilities that older encryption protocols like Signal’s Double Ratchet handle less efficiently.
Wire raised $28 million in a growth round in 2024 and reports approximately 1,800 enterprise customers. The company is notable for its dual domicile (operational headquarters in Berlin, holding company in Zug) and its commitment to open-source clients, allowing security auditing by external researchers.
Specialized Privacy Infrastructure
Transcend (San Francisco, California)
Transcend has evolved from a consent management platform into a data mapping and privacy engineering platform that embeds privacy controls directly into an organization’s data infrastructure. Rather than sitting on top of data systems and auditing them periodically, Transcend integrates at the infrastructure layer – connecting to databases, SaaS applications, and data warehouses to provide real-time data lineage, automated deletion, and purpose-limitation enforcement.
Transcend raised a $40 million Series B in 2024 and reports integration with over 100 data systems. The company’s Data Mapping feature uses automated schema analysis and data sampling to classify sensitive data across an organization’s infrastructure without requiring manual inventory – a significant improvement over the spreadsheet-based data mapping that most organizations rely on.
Drata (San Diego, California)
Drata has built a compliance automation platform that continuously monitors an organization’s security and privacy controls against frameworks including SOC 2, GDPR, HIPAA, and ISO 27001. The platform integrates with over 85 cloud services and infrastructure providers to automatically collect evidence of compliance, reducing the manual effort of audit preparation.
Drata raised $200 million in a Series C in 2022 at a $2 billion valuation and has continued to grow aggressively, reporting over 5,000 customers by late 2025. While Drata is primarily a compliance platform rather than an architectural privacy tool, its significance lies in making compliance continuous rather than periodic – shifting from point-in-time audits to real-time monitoring that can detect privacy control failures as they occur.
The Pattern Across Categories
The 18 companies profiled here share a common architectural principle: they move privacy from the policy layer to the infrastructure layer. Whether through homomorphic encryption, confidential computing, decentralized identity, federated communications, or zero-knowledge proofs, each company is building technology that makes privacy a structural property of the system rather than a contractual commitment by the operator.
This architectural shift is the most significant development in privacy technology since the GDPR. Compliance tools manage the privacy risk created by data-accumulating architectures. The companies on this list are building architectures that do not create the privacy risk in the first place.
The Stealth Cloud Perspective
We track these companies because they are building the same future we are building – from different angles, with different technical approaches, but with a shared conviction that privacy must be architectural. The Stealth Cloud architecture incorporates elements from several of the categories represented here: client-side PII detection (Private AI’s domain), zero-knowledge authentication (Spruce’s domain), ephemeral infrastructure (a principle shared across categories), and network-layer privacy considerations (Nym’s domain).
No single company will build the entire privacy infrastructure stack. The companies worth watching in 2026 are the ones that recognize this and build composable, interoperable components that other privacy-first architectures can integrate. The privacy tech funding data confirms that capital is flowing to this thesis. The companies on this list represent the best current execution against it. We expect several of them to become foundational infrastructure within three years, and we are building Stealth Cloud to be composable with the stack they are collectively assembling.