The Data Center Company Betting Big on AI (w/o taking on GPU risk)

Today, I’m digging into Equinix (EQIX)

They are a company that basically builds and leases out data centers.

There's been a fair amount of chatter out in the market right now about whether or not big tech companies are appropriately depreciating the GPUs they're buying. There are big unknowns about the useful life of a GPU given the advances in NVIDIA technology. So this Equinix piqued my interest.

A few things that stood out to me:

• Their business is incredibly sticky. Once a customer is wired into an Equinix facility, it is a pain to leave.

• Recurring revenue is in the mid-90% range, which is elite for a real-world infrastructure business.

• They do not own GPUs in their AI facilities. Customers bring their own hardware. Equinix is the landlord, not the compute provider.

• Their moat comes mostly from network density and interconnection, not the real estate itself.

• The negative: the capex is brutal. Historically, capex has absorbed basically all operating cash flow, which makes it hard to evaluate the true economic engine underneath.

The two main insights that I took away from this are:

1. I think it's smart of them to not own the GPU/compute. Obviously they would have more pricing power if they did, but I like the strategy of minimizing risk by allowing customers to bring their own GPUs to the data centers.

2. As mentioned above, I think that it would be really interesting to see what their profitability and cash flow would look like if and when they dial back the Capex. The company has an enterprise value north of $90B today, the market is giving them plenty of value, but I remain interested to see what their long-term cash flow would be if they dialed down their growth capex.

It's hard for the average entrepreneur like myself to have a bunch of actionable takeaways from this report. But the big one is: If you're in a space where technology is advancing quickly, be cautious of committing significant capital to buying that technology that could be obsolete in just a couple years' time frame.

With that, I will see you tomorrow.

Nick

TL;DR

• Equinix runs the physical infrastructure where clouds, carriers, and enterprises interconnect.

• The company wins through scale, network effects, and extremely sticky recurring revenue.

• Their moat is strong but expensive to maintain due to heavy capex and growing debt.

• Entrepreneurs can learn about platform design, category reframing, and the tradeoffs of capital-intensive models.

The 30,000-Foot View

Equinix is a global data center and interconnection REIT. Customers rent space, power, cooling, and cross-connects, creating dense digital ecosystems. Revenue comes mostly from recurring charges tied to colocation and interconnection services.

• Business model: global colocation, interconnection, and digital services.

• Revenue mix: about 95% recurring.

• Key metrics:

  • Market cap: $73.8B

  • TTM revenue: $9.06B

  • TTM gross margin: ~50%

  • TTM operating income: $1.53B

  • Net debt TTM: ~$16.7B

  • Employees: 13,600

Company History

• 1998 to 2000: Founded as a neutral colocation hub, rapid early expansion, IPO.

• 2001 to 2008: Survives dotcom crash, expands into Europe and Asia.

• 2010 to 2016: Major acquisitions (Switch and Data, Telecity, Verizon DCs).

• 2017 to 2020: Repositions as Platform Equinix, enters bare metal and xScale.

• 2021 to 2025: Sustainability positioning, AI infrastructure push, heavy Capex plans.

Show Me the Money

Stand Out Features

• Massive and rising capex commitments (avg $4-5B through 2029).

• Capex has exceeded operating cash flow in each of the last four years. This is a pretty big red flag to me. Their operating cash flows/profits have seen a steady increase, so there's obviously an incremental ROI from the investments that they're making. But I'm just surprised to see the lack of free cash flow.

• Leverage increasing as they chase AI-related demand.

• Growth moderating into mid single digits.

Financial Data

The N.O.O.B. Nine — Competitive Powers

The Nerd Out on Business Nine is made up of Hamliton Helmer's famous "7 Powers" of competitive advantage (Scale Economies, Network Economies, Counter-Positioning, Switching Costs, Branding, Cornered Resource, and Process Power) combined with two of my own (Data Flywheel and Distribution Advantage).

Average Score: 3.4/5 - Equinix has a solid, multi-layered moat built on scale and network effects, but it is not untouchable.

Memorable Marketing

Equinix markets itself as a digital infrastructure platform rather than a landlord. The brand focuses on thought leadership, technical content, and partnership driven influence.

Key Tactics

• Platform Equinix: Reframed the category away from space and power.

• Interconnection Oriented Architecture: Architectural playbooks embedded Equinix in design decisions.

• Sustainability positioning: Helped win enterprise deals.

• xScale and hyperscaler co marketing: Strengthened credibility in the AI era.

Tactical Takeaways

• Sell a system, not a SKU.

• Reset your category in your favor.

• Use deep technical content as pre sales fuel.

• Align with external pressures.

• Co market with larger players.

AI Uses & Opportunities

Current Uses

• Predictive maintenance and energy optimization.

• Capacity planning.

• Sales targeting.

• Supporting customer AI deployment architectures.

Future Opportunities

• Dynamic pricing and yield management.

• Automated design for infrastructure deployments.

• AI driven ecosystem matchmaking.

• ESG and compliance automation.

• Managed AI edge services.

Bumps in the Road

• Heavy debt load tied to aggressive capex.

• Forecast volatility due to deal timing.

• Historical short seller scrutiny.

• Regulatory constraints around power and environmental impact.

• Competition from Digital Realty, hyperscalers, and sovereign backed data centers.

Your Swipe File

In all honesty, there aren't a lot of really actionable entrepreneur-focused takeaways for the majority of people from this type of company. If you're looking to build multi-billion dollar data centers, I'm probably not the expert for you. With that said, here are a few insights:

• Capex almost always "pencils" when it's approved/implemented, but it should eventually lead to excess free cash flow.

• There's a fair amount of chatter in the market about how AI-related data centers and servers should be depreciated. My research tells me that the customers bring the GPUs to the Equinix-owned data centers. This lowers the margin profile of the services they provide, but I think it's smart from a risk management standpoint given the advances in AI and the ultimate questions around GPU life and the final business models that are going to get put in place. I like the fact that they do not own the GPUs/compute.

Tomorrow, I have another AI-related data center company coming to you.