Open in interactive viewer → charts, metric popovers & call review

Csquare's $114 Million Backlog Isn't an AI Story. That's the Point.

Record enterprise bookings and a 17% renewal spread at a newly public data-center landlord—while the AI-infrastructure tape cracks.
CSQR · Earnings Call · 2026-11-09

The Quarter: Contracted Growth, Not AI Hype

Csquare used its third-quarter call to make a simple argument: this is a contracted, enterprise colocation business that happens to sit near AI, not an AI data-center story that happens to have enterprise customers. The distinction matters because the public tape has spent the last 90 days punishing the latter. The AI data centers basket and data center capacity names are among the market’s sharpest 90-day decliners in the global tape. Csquare, newly public and without a price tape in this dataset, is asking investors to underwrite backlog and pricing instead of AI sentiment. The quarter gives it a credible case—but the balance sheet leaves little room for a sloppy execution story. The headline numbers were strong. Revenue rose to $302.3 million, up 25.6%, and adjusted EBITDA reached $132.1 million, up 44.3%. “Revenue increased 25.6% year-over-year and adjusted EBITDA grew by 44.3% during that same time period.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 Adjusted EBITDA margin expanded to 47.8% from 40.5%, driven largely by the same data-center portfolio: about $16.4 million of incremental EBITDA from growth and $15.5 million from margin expansion, or roughly 79% of the year-over-year increase. GAAP net loss widened to $88.1 million from $29.8 million, but the swing was mostly $58.6 million of IPO-related costs, post-IPO stock compensation and pre-IPO compensation termination. The cash engine, not the GAAP loss line, is the relevant story. The more important operating fact was the contracted backlog jump: from roughly $76 million at the end of Q2 to about $114 million at the end of Q3, with roughly $104 million expected to commence over the next 12 months. Annualized bookings reached $66.1 million, the 14th consecutive quarter of sequential growth. Management also sold 21 megawatts of new capacity at an expected build multiple of 1.6x and a cost below $4 million per megawatt. “we sold 21 megawatts of new capacity ... our overall expected build multiple for that first quarter is at 1.6x and our cost per megawatt is less than $4 million for the totality of the quarter.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 That is the kind of disclosure that makes a landlord story tangible: not a promise about AI demand, but signed deals, a cost basis and a delivery schedule.

The AI Distinction Is the Differentiator

Management spent unusual airtime disclaiming AI. More than 85% of monthly recurring revenue comes from core enterprise customers, and the average customer relationship is longer than 12 years. AI is not the foundation of the model. In Q3, only about 30% of deals by MRR were strictly AI, and Enterprise demand was broad across financial services, health care, technology, communications and network providers. “only about 30% of the deals that we did from an MRR perspective were AI, strictly AI.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 The CEO was even more direct: “We don't need Csquare to become an AI company for AI to be meaningful to our growth.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 That is a deliberate contrast with the consensus AI-infrastructure pitch. Csquare is not a hyperscale or neocloud pure play. Neocloud exposure is roughly 10% or less. “Our neocloud exposure is relatively light, Madison, today, approximately 10% or less.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 The company is instead selling enterprise colocation with inference workloads as an incremental demand driver. If AI capex sentiment stays weak, that mix is a defense. If enterprise inference adoption accelerates inside existing infrastructure strategies, it is offense. Either way, the story is not levered to the most crowded part of the AI trade.

Leverage and the ABS Question

The balance sheet is where the story gets less comfortable. Post-IPO, Csquare repaid about $1.1 billion of debt, fully repaid its revolver and ended the quarter with roughly $1.1 billion of available liquidity. But net leverage is still about 8.4x annualized year-to-date adjusted EBITDA, against a target of 6x to 7x. “Net leverage based on our annualized year-to-date adjusted EBITDA was approximately 8.4x at quarter end.” — Steven Cook, Chief Financial Officer · 2026-11-09 The deleveraging path depends on converting contracted backlog into EBITDA while keeping capex disciplined. That is plausible—but it is not automatic. The capital structure has real strengths. Substantially all debt is fixed rate at a weighted average of about 5.4%, less than 4% of nonrevolving debt matures through 2028, and 99% of asset-backed securitization issuances are investment grade. The CFO also pushed back on the idea that the ABS structure is trapping cash:

We currently maintain a large restricted cash balance on our balance sheet ... We are not in a cash trap scenario. The ABS administration works where cash is swept and then released following debt service.

Steven Cook, Chief Financial Officer · 2026-11-09
That answer addressed coverage and liquidity tests, but not the specific question of whether Brookfield or its affiliates hold any of those notes. The omission is worth noting, because ABS complexity is exactly the kind of thing that can become a sentiment problem before it becomes a cash problem. There is also a subtle definitional point in the customer-quality disclosure. Management says more than 60% of the customer base is investment grade or has investment-grade characteristics; the CFO later clarified that about 40% is actually IG, with the remainder mostly market caps above $1 billion. That is still a solid base, but it is not the same as a 60% IG book. For a leveraged landlord, that distinction matters.

What to Watch: Conversion, Not the Print

The backlog conversion schedule is unusually specific: about $34 million expected online by the end of Q4, $44 million by the end of Q1 and $23 million by the end of Q2. That gives the next two quarters a visible bridge. The risk is that the 17% renewal spread—the highest in five quarters—is extrapolated as a run-rate. It should not be. Excluding the largest customer renewal, the historical spread is closer to 8%. “generally speaking, we see approximately 8% on our renewal spread overall in the system.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 The large renewal was real pricing power, but the CEO said nothing similar is on the near-term horizon. The other tension is bookings normalization. The deck apparently mentioned bookings normalizing from peak, while the CEO simultaneously called demand record-level. “We did talk about normalizing where that normalizes, I don't really know.” — Spencer Mullee, Chief Executive Officer · 2026-11-09 That is not a guide. Investors should watch Q4 bookings and Q1 conversion rather than annualizing the Q3 print. The embedded runway is genuine: nearly 2 gigawatts of utility feeder capacity, about 1 gigawatt of installed transformer capacity, roughly 400 megawatts of installed sellable UPS capacity and only about two-thirds of floor space currently sold. That supports the capital intensity argument—much of the growth can come from existing shells and power, not greenfield builds. The megawatt economics are attractive when deals are sold into existing space. But the same leverage that makes the equity option valuable also raises the cost of a missed commissioning schedule. Csquare is not asking investors to believe an AI narrative. It is asking them to believe a contracted cash-flow story: enterprise demand, pricing power, backlog conversion and deleveraging. The quarter supports that story. The 8.4x leverage and ABS complexity mean the margin for error is thin, and the tape is not offering any benefit of the doubt to data-center infrastructure broadly. That is the real setup.