Roger Barga, SVP of AI & ML at Oracle Cloud, is a senior infrastructure leader with prior roles at Microsoft Azure and Amazon Web Services. He is interviewed by George Hoyem, formerly of In-Q-Tel. 

Key takeaways

  • AI infrastructure scale and economics 
    • AI capability is increasingly defined by access to electricity and GPU-powered compute rather than traditional cloud metrics. 
    • Oracle positions itself as operating at comparable scale to hyperscalers, with rapid expansion of multi-gigawatt data centre capacity. 
    • Differentiation lies in performance-oriented infrastructure (e.g. bare metal, low-latency networking), delivering lower training costs for large models. 
  • Constraints shaping the AI market 
    • Four key bottlenecks: energy, chip supply, data availability and latency. 
    • Chip manufacturing is viewed as the most binding constraint due to supply chain concentration. 
    • Data scarcity (especially high-quality text) may drive greater reliance on synthetic data. 
  • Enterprise adoption and workforce impact 
    • AI is reshaping tasks rather than eliminating roles; developers increasingly focus on testing, documentation and system understanding. 
    • Productivity gains can unlock new projects rather than reduce headcount. 
    • Effective upskilling is peer-led (one-to-two levels above), rather than top-down transformation. 
  • AI posture management as a new risk frontier 
    • Enterprises lack visibility over employee AI usage, data flows and agent deployment, creating new security vulnerabilities. 
    • Governance frameworks (analogous to cloud or data security posture management) are still immature. 
  • Shifting software and business models 
    • Core systems (ERP, CRM) remain durable, but interfaces are changing via agents and natural language layers. 
    • Emerging models focus on delivering outcomes rather than software licences, enabled by multi-agent workflows. 
  • Implications for corporate venture investors 
    • Strong opportunity in AI security, governance and “posture management” tooling (e.g. red teaming, sandboxing). 
    • Startups are likely to outpace enterprises in addressing emerging AI risks, creating a clear investment gap. 
    • Infrastructure differentiation (cost, latency, scale) remains a critical competitive axis, with partners such as OpenAI and NVIDIA shaping demand. 

This is an AI-generated summary, which has been lightly edited by GCV staff.