VERIFYBefore you quote it

Most core announcements here are confirmed against primary Cisco sources (newsroom, blogs, product data sheets, Splunk) and corroborating trade press. Several figures are vendor-stated, stage-delivered or directional — notably the 450% traffic figure, the 88% case-routing and "35–40 minute" Cisco IQ claims, the 2,036 customer count, and the tokenomics math — and a handful of original-briefing claims were corrected here (Cloud Control's GA date, the 92%→97%/Amsterdam SOC figure, the tokenomics attribution, the $9B as earnings guidance, DefenseClaw's RSAC origin, Centoni's title, and silicon-family vs SKU). The Mythos/Claw threat ecosystem is presented as an illustrative scenario, and all field angles are suggested positioning to validate with a Cisco SME.

• Cloud Control availability: Cisco Cloud Control entered Controlled Availability in the US on June 2, 2026, with "global availability to follow" and no date given. Do not say "GA July 2026" for Cloud Control — that July 2026 belongs to Cisco IQ capabilities (Quantum Ready Assessments, peer benchmarking). Cloud Control Studio = "later in 2026"; Cisco Data Fabric GA = "over the next ~2 months"; Network Actions beta = June 2026.
• Agentic SOC "92%" is contradicted. No source states 92%. The real, primary-sourced figure is 97% (176 of 179 incidents), and it is from Cisco Live EMEA Amsterdam (Feb 2026) NOC/SOC anecdote — not the US event, which gave no percentage. The SOC product (Splunk's six agents + Instant Attack Verification with MITRE ATT&CK mappings + per-decision reasoning log) is confirmed; the 92% figure is uncorroborated.
• Tokenomics is Patel's illustrative example. The $200/wk ≈ $10K/yr → ~$400M at 40K / ~$900M at 90K math was delivered by Jeetu Patel (not Ranjan) as a back-of-envelope example in the fireside with AMD CIO Hasmukh Ranjan; Ranjan's line was "that line item never existed a few years back." Patel conceded it is "still in phase one." Directional — not AMD's actual spend.
• $9B is earnings guidance. The ~$9B is Cisco's raised FY2026 full-year AI-infrastructure ORDER guidance (up from ~$5B) from the May 13, 2026 Q3 earnings — a financial backdrop, not an event announcement and not a single order.
• DefenseClaw origin + scenario nature. DefenseClaw launched at RSAC 2026 (March 23) and was extended (not debuted) at Cisco Live. The Mythos/Claw ecosystem (DefenseClaw, OpenClaw, OpenShell, "ClawHavoc") is presented as a largely illustrative/fictionalized agentic-threat scenario; the "AMD uses DefenseClaw" linkage is uncorroborated.
• 450% traffic is single-vendor. The 450% figure comes from Cisco's published report "AI Impact on Wide Area Networks" (~May 22, 2026; Gurudatt Shenoy & Javier Antich), stated on stage by Patel. The wording/attribution are confirmed, but it is a single-vendor measurement with no independent validation.
• Attendance is not a primary Cisco count. "More than 20,000 attendees from over 75 countries" is corroborated by Futurum's post-event write-up; the precise "75 countries" was not in the Cisco-primary statement fetched — treat as reported.
• Field angles are suggested positioning. Every fieldAngle/talk-track here is a suggested framing to validate with a Cisco SME, not Cisco-endorsed guidance.
• Meraki/Digital Twin dates are trade-press, not primary. "Agentic Actions (Meraki) beta June 2026" and "Digital Twin alpha July 2026" come from Network World, not the primary Cisco release (which uses softer "signal to action" language and no dates).
• Liz Centoni's title is EVP & Chief Customer Experience Officer (since March 2024) — not "Chief Technology and Strategy Officer" as the original briefing stated; one Cisco TL;DR recap and a SiliconANGLE piece used inconsistent/outdated titles.
• Silicon families vs SKUs. The 8100 and Nexus 9300 are system families on shared Silicon One (G300 ASIC), not silicon SKUs; Catalyst 9550 uses campus-class E100/E104, distinct from the data-center G/P series. G300 (102.4 Tbps, 3nm) first shown at EMEA Feb 2026; P200 (51.2 Tbps, scale-across ~1,000 km) announced Oct 2025.