Nvidia's Vera Rubin NVL72, unveiled at CES 2026, introduces rack-scale encryption, marking a potential turning point for enterprise AI security. The new platform encrypts every bus across 72 GPUs, 36 CPUs, and the entire NVLink fabric, offering confidential computing across CPU, GPU, and NVLink domains for the first time at this scale.
This development addresses growing concerns about the security of AI models and the infrastructure supporting them. As AI model training becomes increasingly expensive, with Epoch AI research indicating costs growing at 2.4x annually since 2016, the need for robust security measures intensifies. The Rubin platform aims to provide a more secure environment for these valuable AI assets.
According to Louis Columbus, "For security leaders, this fundamentally shifts the conversation. Rather than attempting to secure complex hybrid cloud configurations through contractual trust with cloud providers, they can verify them cryptographically." This shift is particularly relevant given the increasing sophistication and frequency of cyberattacks, including those launched by nation-state adversaries.
The core issue is that traditional security approaches are struggling to keep pace with the rapid advancements in AI model training. Existing security budgets and methods may not be sufficient to protect the massive investments in frontier training models. The Vera Rubin NVL72 aims to solve this by providing a hardware-based encryption solution that scales to meet the demands of modern AI workloads.
The implications of rack-scale encryption extend beyond mere data protection. It enables organizations to maintain greater control over their AI infrastructure and data, reducing reliance on third-party cloud providers for security. This is especially important for industries dealing with sensitive data, such as healthcare, finance, and defense.
The introduction of the Vera Rubin NVL72 represents a significant step toward securing the future of AI. By providing a comprehensive encryption solution at the rack scale, Nvidia is addressing a critical vulnerability in the AI ecosystem and enabling organizations to develop and deploy AI models with greater confidence. The platform is expected to be available in the latter half of 2026, with further details on pricing and specific configurations to be released closer to the launch date.
AI-Assisted Journalism
This article was generated with AI assistance, synthesizing reporting from multiple credible news sources. Our editorial team reviews AI-generated content for accuracy.
Deep insights powered by AI
Continue exploring
NASA has launched the Pandora mission to enhance the capabilities of the James Webb Space Telescope in the search for habitable exoplanets. Pandora, a smaller satellite, will work in tandem with Webb to analyze the chemical composition of distant planetary systems, seeking signs of water vapor, carbon dioxide, and methane.
Despite explicit Google Play Store policies prohibiting apps that generate non-consensual or sexualized imagery, particularly of children, Elon Musk's Grok AI app remains available with a "Teen" rating. This discrepancy highlights a lack of enforcement by Google, contrasting with Apple's stricter but less explicitly defined app content restrictions, raising concerns about platform responsibility and user safety.
The Supreme Court is set to review the FCC's authority to issue fines, specifically regarding a case where major carriers were penalized for selling customer location data without consent, raising questions about the agency's power and potential Seventh Amendment implications. This legal challenge could reshape the regulatory landscape for telecommunications, impacting how the FCC enforces consumer privacy and data protection rules in an era increasingly reliant on AI-driven data collection and analysis.
Pompeii's public baths, preserved by the eruption of Mount Vesuvius in 79 CE, offer insights into the city's evolving water management. A new study analyzing calcium carbonate deposits reveals a shift from reliance on rainwater and wells to a more complex aqueduct system, reflecting advancements in Roman engineering and urban development. This transition likely improved hygiene and public health in the bustling port city, a key hub in the ancient Mediterranean world.
The Trump administration's EPA is considering a policy shift that would disregard the health benefits of reducing air pollution when making regulatory decisions, potentially reversing decades of established practice that factors in the economic value of human life. This change could have significant implications for public health, as it may lead to weaker regulations on pollutants like ozone and fine particulate matter, both of which are linked to serious cardiovascular ailments. The move raises concerns about the future of environmental protection and the role of AI in assessing the true cost-benefit analysis of environmental regulations.
Semantic caching, which focuses on the meaning of queries rather than exact wording, can drastically reduce LLM API costs by identifying and reusing responses to semantically similar questions. Traditional exact-match caching often fails to capture these redundancies, leading to unnecessary expenses, but implementing semantic caching can increase cache hit rates and significantly lower LLM bills.
Anthropic's Cowork, now in research preview for Max subscribers, simplifies AI-driven file management by allowing Claude to interact with designated folders through a user-friendly chat interface. Built on the Claude Agent SDK, Cowork offers a less technical alternative to Claude Code, opening up possibilities for non-coding tasks like expense report generation while raising considerations for managing AI autonomy.
Pebble's founder, Eric Migicovsky, is launching Core Devices, focusing on a sustainable business model for a Pebble smartwatch reboot and an AI ring, avoiding the pitfalls of traditional venture-backed startups. Core Devices aims for profitability from the outset, leveraging lessons learned from Pebble's acquisition by Fitbit, by carefully managing inventory and foregoing external funding. This approach signals a shift towards long-term viability in the consumer electronics space, prioritizing measured growth over rapid expansion.
Multiple news sources report that MacKenzie Scott donated $45 million to The Trevor Project, a nonprofit supporting LGBTQ youth, marking their largest single donation ever and a critical boost following increased demand for services and the Trump administration's closure of related federal counseling programs. This donation aims to expand the organization's reach and address the heightened mental health challenges and political hostility faced by LGBTQ young people, who have experienced increased suicidal ideation.
Anthropic has unveiled Claude for Healthcare, a suite of AI tools designed to streamline healthcare processes for providers, payers, and patients, mirroring OpenAI's ChatGPT Health announcement. Claude distinguishes itself with connectors that allow access to crucial databases, potentially accelerating research and administrative tasks, though concerns remain about the reliability of AI-driven medical advice.
GoFundMe is facing scrutiny for hosting a fundraiser for an ICE agent who fatally shot a civilian, potentially violating its own policy against supporting legal defenses for violent crimes. This raises questions about the platform's content moderation and the ethical implications of crowdfunding in cases involving law enforcement and civilian deaths, highlighting the challenges of applying AI-driven content policies consistently. The FBI is currently investigating the shooting.
Right-wing influencers are using social media to portray Minneapolis as lawless following a federal agent's shooting, potentially shaping public perception through selective narratives. This coordinated content strategy highlights the power of AI-driven platforms in amplifying specific viewpoints and influencing broader societal discourse.
Discussion
Join the conversation
Be the first to comment