Cisco to partners: trust, not tech, decides who wins at AI

Only 52 per cent of organisations across Cisco's APJC region are confident they can detect and respond to compromised AI agents, according to Cisco research shared at Cisco Live Melbourne. Nine in 10 believe their current vulnerability management processes cannot handle AI-driven threats. The numbers frame Cisco's argument that trust, security and infrastructure readiness — not the models themselves — will decide which enterprises actually profit from AI.

Cisco APJC president Ben Dawson opened the event by calling AI "the largest transformation of our lifetime" and said the industry has moved past experimentation towards delivering measurable business outcomes.

"We do know it's going to have a transformative impact on countries, economies, societies and the businesses in which we operate," Dawson said. "The next wave of innovation is going to come from understanding what AI can do for us, how do we take advantage of it, how do we integrate it into our business in a meaningful way and do it safely and securely."

How fast does the workforce shift?

Dawson put hard numbers on adoption. Today, roughly 2 per cent of the global workforce qualify as AI "power users" who genuinely embrace AI and AI agents. Cisco expects that share to balloon to around 30 per cent within 12 to 18 months.

That growth loads new demands onto infrastructure, security and networking — which suits Cisco's portfolio. Executive vice president of global sales Oliver Tuszik called APJC the company's most significant long-term growth opportunity despite being its smallest geographic region. He credited a regional case of "FOMO".

"The fear of missing out is about being first, not being second, not third," Tuszik said. "Standing still means moving backwards."

Organisations across APJC, he said, show a greater willingness to experiment with AI and embrace transformation rather than fixate on regulatory or operational risk.

What does agentic AI do to the network?

Cisco SVP of portfolio strategy and marketing Jeff Schultz argued that agentic AI forces enterprises to rethink their infrastructure architectures. Chatbots run intermittently; agents work continuously and increasingly collaborate with each other.

"Agents fundamentally operate at a completely different scale, at machine speed and scale," Schultz said.

Cisco is already seeing "swarms" of tens of thousands of AI agents working together, generating significantly more traffic than human users. That volume raises a cost question Cisco brands "tokenomics": whether to route workloads to frontier AI models or to self-hosted models in an enterprise's own data centres.

"Each and every one of us needs to be thinking about what we call tokenomics," Schultz said. "How do we optimise for the use case and send traffic either to the frontier models when necessary or to our own models that are potentially hosted in our own data centres?"

Autonomy also blurs the boundary between malfunction and attack. "The line between observing an agent and securing an agent is beginning to blur," Schultz said. "You can no longer tell the difference between a behavioural issue and a security threat."

What does this change for customers?

For buyers, the short answer is services. Tuszik said recurring revenue opportunities remain strong across software and services, particularly where organisations need help securing existing environments and deploying AI safely.

"Whatever customer you take, it might be a multi-billion international or smaller SMB one. They all have the same challenge right now," he said. "They know that their installed base might be vulnerable, and might not be able to manage what comes next with agentic AI."

He also flagged emerging AI consumption models, including GPU-as-a-service and AI inferencing services. "A company is not only buying compute by the hour. They need the integration, services, sovereignty and security."

Asahi Beverages offered the customer view. CDIO John Kelvie said the beverage giant is moving beyond experimentation towards larger, business-wide initiatives, having already found value in demand sensing, forecast optimisation, trade promotion optimisation, back-office automation, chatbots and voice bots.

"But I think really for us the next stage is where we're going to get real growth," Kelvie said. He located the obstacle in data quality, not technology: "AI is going to amplify your strengths, but it's also going to amplify your inconsistencies. The real challenge for us is around data."

What happens to distribution?

Cisco will extend its Cisco 360 strategy to distributors, with a dedicated framework slated for FY28. SVP global partner sales Tim Coogan told ARN the company is working with distributors on a model aligned to the broader Cisco 360 transformation.

"The old program worked well for everybody and it worked well for us, but it wasn't going to support what the market was telling us was needed for the future," Coogan said.

Distributors, he said, want the same outcomes as partners: pricing consistency, greater access to training and technical resources, and support to build expertise across Cisco's wider portfolio.

"We want every partner to make the journey with us," Coogan said. A neat summary of the whole event, really — the destination is agreed; the readiness statistics suggest most of the caravan is still packing.