How Ecosystem Partnerships Turn AI Ambition Into Business Value

How can organizations move beyond AI experimentation and deploy scalable solutions that deliver measurable business outcomes? At Horizon, Gordon Mackintosh, SVP, Global Partner Sales & Programs at Equinix, leads a conversation with Kevin Wollenweber, SVP, Data Center and Internet Infrastructure at Cisco; Kaushik Shirhatti, VP, AI Factory at NVIDIA; and Robert Kim, Chief Technology Officer at Presidio. Together, they explore the infrastructure, operational, and strategic challenges enterprises face as they move AI initiatives into production. The panel discusses why successful AI adoption requires more than compute alone. It takes an ecosystem that brings together networking, data center infrastructure, models, software, integration expertise, and flexible deployment options. From prioritizing high-value use cases and navigating hybrid architectures to simplifying AI factory deployment, the speakers share practical perspectives on how organizations can accelerate time to value while maintaining control, flexibility, and choice. Key topics include: • Why enterprises must prioritize the AI use cases that can deliver the greatest return on investment • How AI agents are changing network and infrastructure requirements • What an AI factory enables for businesses and end users • The role of hybrid AI across cloud, private cloud, colocation, and on-premises environments • Why tokens are a cost input, not the ultimate measure of business value • How Equinix, Cisco, NVIDIA, and Presidio are working together to simplify AI deployment • The importance of compute, networking, software, power, cooling, and ecosystem integration • How Presidio Path Labs helps customers test concepts and accelerate AI adoption Chapters: 00:00 Why no single company can deliver enterprise AI 02:05 Why enterprise AI stalls: agents, hybrid choices and prioritization 05:16 From cloud AI pilots to running your own infrastructure 06:34 What an AI factory enables, and why tokens are the bill 13:05 Power, cooling and the partner handoff 20:36 Testing AI before production FAQs: Q: What is an AI factory? A: An AI factory brings together applications, models, infrastructure, chips and power to produce tokens that power real business outcomes. It doesn't have to be giga-scale. It can be a single system running a model, and it can run in the cloud, in colocation or on-premises. Q: Why do AI pilots stall before production? A: Teams often have far more AI ideas than they can fund, so prioritizing the few projects with the best ROI comes first. Many proofs of concept are also built on curated cloud AI services that hide the infrastructure underneath. When teams move to their own or hybrid infrastructure, they have to host and broker models, connect model services to GPUs, and plan for power and cooling, and the cost per benefit can fall apart. Q: Are tokens a measure of AI value? A: No. Tokens are the bill, the raw material of an AI factory. Value comes from measuring the outcome the tokens produce, then choosing the AI methods and infrastructure that deliver it most efficiently. Q: How do AI agents change network requirements? A: Cisco analysis shared in the session found that an agent performing the same function as a user can take 450% of that user's network bandwidth, and it works 24/7. That changes how enterprises plan network and infrastructure capacity. The discussion reinforces a central idea: no single organization can deliver every component required for enterprise AI. By combining complementary technologies, infrastructure, and expertise, ecosystem partners can make complex AI systems easier to deploy, operate, and scale. Equinix, Inc. (Nasdaq: EQIX) shortens the path to boundless connectivity anywhere in the world. Its digital infrastructure, data center footprint and interconnected ecosystems empower innovations that enhance our work, life and planet. Equinix connects economies, countries, organizations and communities, delivering seamless digital experiences and cutting-edge AI—quickly, efficiently and everywhere.