We hosted a lunch with Zac Smith, CEO of Packet, yesterday. Packet is a micro edge data center company that provides bare-metal servers for enterprise customers. The rise of micro edge data centers has important implications for towers, data centers, and carriers. For the towers, we came away thinking that micro edge data centers could drive upside, though the size of the opportunity remains opaque and likely immaterial to financials over the next several years. For data centers, Packet’s views reinforce our preference for Equinix over the wholesale players. For carriers, Packet’s deployments with Sprint, along with the proposal they submitted for Dish’s network build, suggests there is upside for carriers to deploy services for enterprises at micro edge locations.
Data centers at towers could provide big uplift to revenue per site…: The economic model is still emerging, but Packet spoke to leasing revenue of $8,000 per month at a tower site, which could more than double the amount of revenue that tower companies generate per site on average. Tower locations are optimal for micro edge data centers because they provide real estate, power, fiber, and potential access to low-cost financing; however, it is important to note that micro edge data centers don’t need to be at towers - Packet also leases space from other real estate companies (Brookfield) and other data centers (Equinix).
…but opportunity size still opaque: Packet has proven proof of concept through deployments with Sprint and others; however, the scale of the opportunity remains unclear. It seems like Packet hopes to deploy hundreds of sites over the next several years (cities with more than 100,000 people); even if all of these are deployed across tower companies, it’s unlikely to move the needle for tower companies. To put it into context: the big three tower companies have ~100K sites collectively in the US that generate $8.8BN in site leasing revenue. If the market sees 1,000 micro-edge data centers, and all of them went on tower locations, it would only boost revenue by 1% for the industry. If the micro-edge data center market grew to 10,000 locations, that could boost tower revenue by 10% and would be material; however, we suspect market growth is likely to be closer to 1,000 sites than 10,000.
Micro-edge interconnection unlikely; Equinix will remain dominant: While many speculate around interconnection at micro-edge data centers, Packet was down-beat on this opportunity. They stressed the difficulty in building ecosystems with critical mass at the edge to warrant interconnection. Equinix will likely remain the preeminent interconnection platform for the foreseeable future – “there is no way around them”.
The opportunity with Dish: Packet has bid on the Dish network RFP (along with seemingly everyone else). Interestingly, Packet has already deployed several sites with Sprint to serve enterprises for ~30K per site (80% cheaper than Sprint’s internal estimates for what it would cost to build themselves). Packet could deploy similar functionality with Dish. We think this aligns with Dish’s wireless network strategy (LINK), though unclear at this stage who Dish will select as their vendors.
Hyperscale data centers at risk: Packet thought the major cloud service providers would exert their market power on hyperscalers and beat them down on price over time. Additionally, the flood of infrastructure fund money into data centers may be driving lower pricing as well. Anecdotally, Packet described a recent trip to Ashburn to secure a 500kw as a “feeding frenzy”, with prices well below what they were expecting. This echoes comments by data centers companies over the past few months about oversupply in Northern Virginia, and may suggest that pricing will remain pressured in select markets through the rest of the year.