Cogent Communications CEO: AI Is Transformational, but Monetization Unclear; Sees Fiber, Layer 1 Upside

Core Insights - The economic value of AI outputs currently does not exceed the costs of inputs, leading to companies losing money on AI outputs, although input costs are expected to decline sharply [1][5] - AI is seen as transformational, with potential benefits accruing more to end users and businesses built on AI rather than the infrastructure providers [2][5] - The internet's historical impact is compared to AI's potential, highlighting a deflationary effect and a shift in value from service providers to consumers [1] Company Overview - Cogent Communications is a multinational Internet service provider specializing in high-speed Internet access and data transport services, operating one of the largest Tier 1 IP networks globally [23] - The company offers services including dedicated Internet access, Ethernet transport, wavelength services, and MPLS-based IP Virtual Private Networks over its fiber-optic backbone [23] AI and Connectivity - AI is expected to drive demand for greater data collection and storage, more fiber buildouts at the edge, and increased demand for low-latency "Layer 1" transport services [4][10] - Cogent carries about 2 exabytes (EB) of data daily, accounting for roughly 25% of global internet transit traffic, and has reported 100% year-over-year growth in its Wavelength business [4][10] Financial Performance and Strategy - The company is targeting multi-year revenue growth of 6-8% and aims to return to approximately 40% EBITDA margins after experiencing a decline post-acquisition of Sprint assets [3][16] - Cogent's leverage rose to about 6.6x due to integration challenges, and the company is focusing on asset monetization strategies, including a novel IP-address securitization that raised $380 million [3][19] Market Dynamics - The AI sector currently lacks a proven monetization model, with significant investments expected to yield returns below the average cost of capital [9] - The demand for fiber investment is anticipated to increase due to AI-driven use cases, although broader fiber proliferation does not guarantee attractive returns for companies [10][12] Operational Challenges - Power availability is a major constraint, with traditional data centers consuming about 2.2% of power in developed regions, while AI adds approximately 2.5%, totaling around 4.7% of grid load [8] - The company is addressing post-acquisition integration challenges and has seen a significant shift towards on-net sales, with 80% of sales in the most recent quarter being on-net [18]