Modelyo

**Product and the concrete problem it solves.** Modelyo builds a confidential-computing cloud platform it calls the **Confidential Cloud** — infrastructure that lets organizations run sensitive workloads on shared or public cloud without ever exposing plaintext data at any point in its lifecycle. The concrete problem it attacks is the last unguarded stage of the data-protection chain. Encryption at rest and in transit is a solved problem, but the moment a workload actually executes, data must be decrypted into memory where the host operator, a compromised hypervisor, a malicious insider, or a cloud provider under legal compulsion could in principle read it. For a bank, a hospital, a defense ministry, or an AI company processing proprietary models on regulated data, that "data-in-use" exposure is the reason genuinely sensitive workloads still sit on-premises and never migrate to cloud economics. Modelyo's pitch — branded "The Cloud You Control. The Trust You Verify" — is to close that gap with hardware-enforced isolation plus cryptographic verification, so a customer can mathematically attest that their workload is running unmodified inside a protected enclave and that even Modelyo's own operators cannot see inside it. The result is positioned as a way for high-assurance and regulated organizations to adopt cloud, GPU-accelerated AI, and managed services without surrendering control of their most sensitive data.

**Core technology and how it actually works.** Modelyo's differentiation is architectural rather than a policy or software-overlay wrapper. The platform runs workloads inside **Trusted Execution Environments (TEEs)** — publicly described as Intel SGX (Software Guard Extensions), Intel TDX (Trust Domain Extensions), and NVIDIA H100 Confidential Computing for GPU workloads — so that memory is hardware-encrypted and isolated from the host operating system and hypervisor. Around that core sit several reinforcing mechanisms per the company's materials: **continuous remote attestation**, where cryptographic quotes are generated on a short cadence (stated as roughly every 60 seconds) and verified against the Intel Attestation Service, giving customers a live, verifiable proof that the enclave is genuine and the code unaltered; **customer-controlled encryption keys held in Hardware Security Modules (HSMs)** rather than on Modelyo's infrastructure, enforcing an operator-exclusion model; and a confidential platform stack built on **OpenStack and Kubernetes with encrypted etcd**, plus managed confidential databases (PostgreSQL/MySQL/Redis), a GPU-accelerated AI runtime, and a robotics control plane. Crucially, Modelyo is building a *whole cloud* — an IaaS/PaaS layer — not merely a software shim that lifts existing apps into enclaves on someone else's hyperscaler. It also offers **sovereign deployment modes**: on-premises, air-gapped, and multi-cloud. The honest calibration is that most of these capabilities are vendor-stated on the company's own site; there are no published third-party benchmarks, penetration-test results, or independent security audits in the public record yet, and confidential computing's real-world assurance depends heavily on correct implementation of attestation and side-channel mitigations.

**Market, customers, and go-to-market.** Modelyo targets five regulated verticals it names explicitly: **financial services** (banking, insurance, capital markets), **healthcare**, **defense and government**, **telecommunications**, and **automotive/robotics**. Its go-to-market leans on the compliance tailwind: the site references an unusually long list of regimes it is designed to help satisfy — DORA, Basel III, SOX, HIPAA, FDA, ISO 13485, FedRAMP, DISA STIG, ETSI, GSMA/NESA, EU NIS2, ISO 26262, and IEC 62443 — which is a tell that the intended buyers are compliance-driven CISOs and regulated-industry architects rather than general developers. The macro timing is favorable: sovereignty mandates (EU data-sovereignty rules, national cloud requirements), the explosion of AI workloads on sensitive data, and rising distrust of hyperscaler jurisdiction are all pushing "confidential" and "sovereign" cloud from niche to procurement checklist item. The realism check is that selling infrastructure into banks, hospitals, telcos, and defense ministries is a long, reference- and certification-gated sales cycle, and Modelyo is doing it as a very young company against both hyperscaler-native confidential offerings and better-capitalized specialists.

**Traction, funding, and third-party validation.** Public validation is real but early and largely non-financial. Modelyo was **welcomed as a start-up member of the Confidential Computing Consortium (CCC)** — the Linux Foundation body that stewards the confidential-computing ecosystem — in February 2026, a credible signal of technical seriousness and standards alignment. It was named to **Calcalist/CTech's "50 Most Promising Israeli Startups" for 2026**, and selected for **CTech's 2026 "Dream Team" delegation** of Israeli startups (described there as enabling "sensitive core systems in fully encrypted and verifiable environments"). It has exhibited at **Cybertech Global Tel Aviv**. What is *not* publicly disclosed is equally important for calibration: there is no announced funding round, valuation, revenue, headcount, or named customer in the public record. The recognitions establish that credible ecosystem gatekeepers take the team and thesis seriously; they do not establish commercial traction, and the absence of any disclosed capital for a capital-intensive "build-your-own-confidential-cloud" ambition is the single largest open question in the record.

**Founders and team background.** Modelyo was **founded in 2024 and is headquartered in Tel Aviv, Israel**. Publicly reported leadership includes **CEO Benny Meir**, described as an IDF lieutenant colonel (reserve) who served in elite units focused on technological operating systems and special operations and who reportedly received a Chief of Staff innovation award in 2021; **Dr. Itzhak Aviv**, who holds a PhD in requirements engineering of complex systems and is based in the Tel Aviv area; and **Artem Barger as VP of R&D**, who is quoted in the CCC announcement framing confidential computing as moving "from an emerging technology to essential infrastructure, especially for government organizations." This is a plausible, security-literate founding profile — elite-unit operational pedigree paired with systems-engineering and deep R&D depth — but public disclosure is thin, the full cap table and engineering bench are not documented, and the founder details rest partly on LinkedIn-sourced reporting that should be confirmed directly in diligence. Team depth in both cryptographic-hardware/attestation engineering and cloud-platform operations is decisive for this category, and the public footprint does not yet prove that bench.

**Competitive dynamics.** Modelyo sits in a crowded, fast-consolidating field and competes on several fronts. (1) Against **hyperscaler-native confidential computing** — Azure Confidential Computing, Google Cloud Confidential VMs, and AWS Nitro Enclaves — it wagers that regulated and sovereignty-sensitive buyers will not accept confidential computing *from the very provider they are trying not to trust*, an argument that is strongest exactly where jurisdiction and operator-exclusion matter most. (2) Against **confidential-computing specialists** such as Israeli-founded **Anjuna Security** (which abstracts enclaves so existing workloads run in TEEs across hyperscalers), **Fortanix** (data-in-use security and key management), and **Edgeless Systems** (confidential Kubernetes/Constellation), Modelyo's distinction is that it is building an entire sovereign cloud stack rather than a portability layer over someone else's cloud — a bigger, more differentiated, and far more capital-intensive bet. (3) Against **sovereign-cloud programs** (European sovereign clouds, national government-cloud initiatives), it competes as a nimble specialist. Its plausible edges are the operator-exclusion/verifiability story, GPU confidential-AI support, and sovereign/air-gapped deployment flexibility; its vulnerabilities are scale, capital, and the risk that hyperscalers simply fold verifiable confidential computing into their default offerings.

**Defense, security, and resilience dual-use relevance.** Modelyo's dual-use relevance is credible and close to its core, though it is enabling infrastructure rather than a fielded weapons or ISR capability. Confidential computing is a genuine national-security primitive: it lets defense and intelligence organizations run classified or sensitive analytics, AI inference, and signals processing on cloud or shared infrastructure while shrinking the trusted computing base and keeping the host operator out of the data — the precise barrier that keeps sensitive government workloads off cloud today. Modelyo names **defense and government** as a target vertical, cites defense-oriented accreditations (FedRAMP, DISA STIG), and its **air-gapped and sovereign deployment modes** map directly onto tactical, classified, and data-sovereignty use. The confidential-AI angle (GPU TEEs) is increasingly strategic as militaries and agencies want to run proprietary models on sensitive data without exposing either. The honest calibration is threefold: this is a security-infrastructure enabler, not a fielded system; the strongest defense uptake requires formal accreditation (FedRAMP authorization, DISA IL levels, national cryptographic certification) that the public record does not yet document as achieved; and dual-use weight is currently constrained by the company's youth, undisclosed funding, and vendor-stated (unbenchmarked) assurances. On the strategic axes of sovereign cloud, critical-infrastructure resilience, and secure/classified compute, however, the technology is squarely on-target.

**Growth stage, trajectory, and key diligence risks.** Modelyo reads as an **early-stage** company: founded 2024, Tel Aviv-based, an early-commercial platform with named target verticals and compliance ambitions, meaningful ecosystem validation (CCC membership, CTech "50 most promising" and "Dream Team," Cybertech), but no disclosed funding, revenue, headcount, or named customers. The trajectory is a bet that sovereignty mandates plus AI-on-sensitive-data demand create durable pull for a verifiable, operator-excluded confidential cloud that hyperscalers cannot credibly provide to the most jurisdiction-sensitive buyers. The principal diligence risks are: (1) **capital intensity vs. undisclosed funding** — building an actual confidential cloud/IaaS is expensive, and no round is public; (2) **unverified assurances** — attestation, isolation, and side-channel claims are vendor-stated without published third-party audits or benchmarks; (3) **hyperscaler incumbency** — Azure/GCP/AWS could make verifiable confidential computing a default, eroding a standalone platform's rationale; (4) **certification gap** — defense/government adoption needs formal accreditation not yet evidenced as achieved; (5) **team-depth opacity** — thin public disclosure of the engineering bench in a domain that lives or dies on correct cryptographic implementation; and (6) **commercial opacity** — no named customers or revenue. Progression toward "mid" would be evidenced by a disclosed funding round, independent security audits/benchmarks, formal accreditations (FedRAMP/DISA), and named regulated or government customers.