Verizon closed a $1 billion dark fiber contract with Google, routing dedicated optical capacity to the hyperscaler's AI data centers. CEO Dan Schulman disclosed the deal Tuesday and stated the carrier expects to announce infrastructure agreements worth multiple billions of dollars before December, all targeting AI compute connectivity. The deal shifts Verizon from bandwidth commodity supplier to specialized infrastructure partner.
The dark fiber model delivers unlit optical strands Google can activate with proprietary transmission equipment, bypassing Verizon's switching layers entirely. This architecture supports the 400-gigabit and 800-gigabit wavelengths AI training clusters require for inter-campus parameter syncing. Google controls upgrade cycles. Verizon collects recurring lease revenue without capex exposure to rapid transceiver obsolescence. The contract runs multi-year, though neither party disclosed term length or payment structure. Schulman confirmed the deal contributes to 2025 revenue, implying cash conversion begins within six months.
The announcement matters because it confirms enterprise telecom margins are migrating toward infrastructure-as-a-service models where hyperscalers own the intelligence layer. Verizon's existing fiber footprint—built for metro business connectivity and 5G backhaul—becomes an annuity stream without incremental route construction. The multiple billions Schulman referenced suggests a portfolio approach: contracts with Meta, Microsoft, Amazon, and potentially Oracle or xAI by year-end. If the average deal size matches Google's $1 billion, Verizon is guiding to $3 billion-plus in AI infrastructure bookings for 2025, roughly 4-5% of projected total revenue. That concentration creates single-vertical dependency risk, but it also establishes Verizon as the connectivity layer for the U.S. AI capital formation cycle.
The timing intersects with hyperscaler buildout acceleration. Google announced $75 billion in 2025 capex in February, with 60% earmarked for AI infrastructure. Microsoft's Azure expansion plan targets 50-70 new data center regions by 2027. Both operators need point-to-point fiber connecting training sites to inference edge nodes, bypassing public internet latency. Verizon's metro and long-haul fiber assets—previously underutilized after enterprise SD-WAN adoption reduced dedicated circuit demand—now match the geographic distribution hyperscalers require. The deal converts stranded infrastructure into cashflow without new permitting or construction delays.
Operators and allocators should track Verizon's Q2 earnings call in late July for disclosed contract count, average deal size, and margin profile on the infrastructure segment. Schulman's December guidance creates a six-month window to announce 2-4 additional hyperscaler deals. Watch for similar dark fiber partnerships from Lumen, Zayo, or Crown Castle—each holds metro fiber density in AI cluster markets like Northern Virginia, Phoenix, and Dallas. If contract terms include minimum commit escalators tied to cluster expansion, Verizon's infrastructure revenue compounds without sales friction. The risk surface is procurement consolidation: hyperscalers standardizing on 1-2 fiber providers per region to simplify operations, which could lock out late-moving carriers.
Verizon's stock traded flat post-announcement, suggesting the market already priced enterprise infrastructure repositioning into the low-teens forward multiple. The multiple billions language was specific enough to move estimates if confirmed at scale.