TTN Summary of 17:00ET Earnings Call: Guides Q2 EPS $1.61-1.65 ($1.58-1.62 cc) v $1.60e, Rev +14-16% (+12-14% cc) v +15.2%e; Substantially raises FY26 Capex
Tuesday, September 9, 2025 5:12:40 PMEST
- FX $0.03 positive impact on EPS and 1% on Rev in Q2
- Expect RPO to grow to exceed $500B (from current $450B)
- Expect operating income to grow midteens % in FY26 and higher still in FY27
- cloud RPO was up nearly 500% in Q1 y/y
- Guides Q2 cloud Rev +33-37% (+32-36% cc)
- Raises FY26 Capex $35B (prior: >$25B)
- No update to FY26 cloud Rev growth (prior: >+40% (cc))
- No update to FY26 Rev (prior: $67B+)
- Founder Ellison: We are not doing bad in the AI training market, and we are well positioned to be in winner in the AI inferencing market, which is much bigger
- Customers are running out of AI inferencing capacity
- Oracle can give you a fully featured private version of the Oracle cloud on three racks for $6M (called 'The Butterfly'); The closest hyperscaler has 100 times that cost
- OCI capacity tightness: consumption revenue +57%; demand “dramatically outstrips supply,” with one large GPU data hall accepted by a customer in one week.
- Oracle AI Database vectorizes enterprise data and integrates with all major LLMs—enabling secure inferencing over private + public data for “ask any question” analytics.
- Multi-cloud database revenue surged 1,529% in Q1; Oracle now live in 34 embedded data centers on AWS/Azure/GCP and will add 37 more (total 71).
- “Butterfly” private Cloud@Customer: full-feature Oracle Cloud in three racks for $6 million (≈1% of hyperscaler cost), fueling dedicated-region deals with global enterprises.
- AI application generator automates ERP/CRM development—new apps are generated as AI agent workflows, slashing build time and embedding AI at no extra licensing fee.
- Oracle GPU superclusters differentiate on network throughput (“twice as fast, half the cost”), securing a moat as AI inferencing demand scales far beyond training.
~
- Expect RPO to grow to exceed $500B (from current $450B)
- Expect operating income to grow midteens % in FY26 and higher still in FY27
- cloud RPO was up nearly 500% in Q1 y/y
- Guides Q2 cloud Rev +33-37% (+32-36% cc)
- Raises FY26 Capex $35B (prior: >$25B)
- No update to FY26 cloud Rev growth (prior: >+40% (cc))
- No update to FY26 Rev (prior: $67B+)
- Founder Ellison: We are not doing bad in the AI training market, and we are well positioned to be in winner in the AI inferencing market, which is much bigger
- Customers are running out of AI inferencing capacity
- Oracle can give you a fully featured private version of the Oracle cloud on three racks for $6M (called 'The Butterfly'); The closest hyperscaler has 100 times that cost
- OCI capacity tightness: consumption revenue +57%; demand “dramatically outstrips supply,” with one large GPU data hall accepted by a customer in one week.
- Oracle AI Database vectorizes enterprise data and integrates with all major LLMs—enabling secure inferencing over private + public data for “ask any question” analytics.
- Multi-cloud database revenue surged 1,529% in Q1; Oracle now live in 34 embedded data centers on AWS/Azure/GCP and will add 37 more (total 71).
- “Butterfly” private Cloud@Customer: full-feature Oracle Cloud in three racks for $6 million (≈1% of hyperscaler cost), fueling dedicated-region deals with global enterprises.
- AI application generator automates ERP/CRM development—new apps are generated as AI agent workflows, slashing build time and embedding AI at no extra licensing fee.
- Oracle GPU superclusters differentiate on network throughput (“twice as fast, half the cost”), securing a moat as AI inferencing demand scales far beyond training.