White Fiber (WYFI) H.C. Wainwright 28th Annual Global Investment Conference summary

Market overview and industry transformation

• AI infrastructure is now a leading investment theme, with hyperscalers expected to spend over $1 trillion on AI in the coming year and Bitcoin miners securing $160 billion in AI colocation deals to date.

• Bitcoin miners are rapidly transitioning into AI data center developers, leveraging their access to scalable power and land to meet surging demand for AI compute.

• The economics of AI data centers differ significantly from Bitcoin mining, with build costs rising from $4.5 million to $13 million per megawatt due to labor and equipment shortages.

• Labor, power availability, and supply chain constraints are now the primary bottlenecks for new data center development.

• Companies are pursuing both co-location and cloud service models, often spinning out pure-play AI infrastructure businesses to attract institutional capital.

Recent deals, business models, and execution strategies

• Major multi-billion dollar lease deals have been secured, such as a $6.6 billion, 20-year triple net lease in Georgia, with additional exclusivity on 885 MW in Texas.

• Companies are balancing co-location and GPU-as-a-service models, with some focusing on smaller sites for neocloud offerings and others locking in long-term leases with investment-grade tenants.

• Strategic partnerships and acquisitions, such as acquiring specialized retrofit teams, are enabling rapid conversion of legacy facilities into AI campuses.

• Recent contracts include a 500 MW deal with AMD and pipeline deals for up to 2,000 MW, as well as $500 million in new cloud service agreements.

• Financing structures typically involve 80% debt and 20% equity, with project-level funding and creative risk mitigation through turnkey contracts and supply chain certainty.

Market risks, CapEx, and perspectives on a potential bubble

• CapEx intensity for AI data centers is high, with costs ranging from $10–13 million per MW and significant upfront equity required.

• Labor shortages and supply chain delays are driving up costs and extending timelines, making execution certainty a key differentiator for lenders and investors.

• Most panelists reject the notion of an AI infrastructure bubble, citing persistent and growing demand for compute from both hyperscalers and enterprise clients.

• Some caution that speculative site announcements and inflated demand projections, especially in Texas, create a bubble in perceived—not actual—demand.

• Retrofitting existing facilities and engaging with local communities are seen as effective strategies to address environmental and social concerns.