Thanks so much for another engaging and timely discussion! "...bleed data by design..." will be a resounding concern in our upcoming conversations. For organizations considering running on their "own metal" for the most sensitive workloads, often their startup partners and some of their target markets sit in jurisdictions with conflicting data-localization requirements. How does one architect a confidential/trusted AI system when the policy requirements of two sovereign environments are in direct tension? Especially when the sovereign compute being sought is inside the "great firewall" where supply-chain considerations limit us to locally manufactured silicon. Lastly, can we really consider "mutual attestation" to suffice, or is it as reassuring as a mutual NDA? Thanks again for the insights.
Thanks so much for another engaging and timely discussion! "...bleed data by design..." will be a resounding concern in our upcoming conversations. For organizations considering running on their "own metal" for the most sensitive workloads, often their startup partners and some of their target markets sit in jurisdictions with conflicting data-localization requirements. How does one architect a confidential/trusted AI system when the policy requirements of two sovereign environments are in direct tension? Especially when the sovereign compute being sought is inside the "great firewall" where supply-chain considerations limit us to locally manufactured silicon. Lastly, can we really consider "mutual attestation" to suffice, or is it as reassuring as a mutual NDA? Thanks again for the insights.