Your pilot worked; then it made production. The Deloitte-HKU AI Adoption Index 2026 found that only 23% of companies in Hong Kong and mainland China get real financial returns from AI¹. The models were never the hurdle; the underlying data infrastructure was. So the questions for CDOs have now moved down a layer. Can your infrastructure deliver live data fast enough to be useful? Can you hold the cost line once agents multiply every request? And how much of it do you actually control, now that the HKMA expects banks to assume frontier AI helping to scale attacks², and the SFC wants you to treat every document and webpage your agent reads as potentially hostile?³ You don’t have to answer them alone.     
 

Red Hat and CDOTrends will host key data and infrastructure leaders from Hong Kong companies at a private lunch. Join us with cross-industry’s CDOs, CTOs and CISOs working through the same problems and discuss:     
 

• The hidden bill: When you shift from training to inference, GPUs sit idle, and pilots stall on real data, the model is rarely the biggest bill.
• What sovereignty really means: Why ownership, not location, matters more as you increase infrastructure resilience and go multi-model.
• The open source edge: The economic advantage in using a single platform to manage servers and agents without lock-in.     
   

Discussion Leaders:

Felix Tsang, Business Value Director, GCG, Red Hat

Winston Thomas, Editor-in-Chief, CDOTrends

Sources  
<sup>1. Deloitte-HKU AI Adoption Index 2026  
2. HKMA circular, 29 May 2026  
3. SFC circular on intermediaries</sup>