@ericjackson@FransBakker9812 Curious why no mention of $APLD ? They’ve been touting their ability to curtail power when necessary, and ramp with optionally. Part of the ellendale campus design.
Where AI Projects Usually Die, $APLD Just Raised a Bridge
@APLDdigital just added a new weapon to its playbook, and it’s aimed at the hardest part of the AI data center business: the awkward gap between “we found a great site” and “we have a signed hyperscaler lease.”
On December 18, 2025, $APLD announced a development loan facility with @Macquarie (Commodities and Global Markets) to fund pre-lease development costs for new data center projects. The facility is at the DevCo level and is meant to cover early-stage work like sourcing, planning, development, and construction for new campuses and potential projects.
They also disclosed the first real timeline clue: Applied Digital says it is in advanced-stage negotiations with another investment-grade hyperscaler for multiple campuses, and the initial $100M in draws is intended to support development activities tied to those campuses.
Why this matters
Pre-lease is where projects go to die.
A lot of developers can talk about megawatts. Fewer can lock down land, power paths, permitting, design, and early procurement fast enough to meet a hyperscaler’s internal deadlines. Hyperscalers move with a weird mix of urgency and bureaucracy, and if you miss a window, the tenant does not “wait patiently.” They move to the next site.
This facility is basically a bridge that lets $APLD keep sites moving forward without committing full build capital before the lease is signed. Management framed it as capital flexibility and a disciplined site-development strategy that aligns spend with demand.
It also fits their broader financing stack with Macquarie.
Earlier, Applied Digital announced a preferred equity financing facility of up to $5.0B with Macquarie Asset Management for HPC data centers. At financial close, Macquarie funded $112.5M, with more capital expected to fund as construction and leasing progress.
So you now have a cleaner ladder:
Development money to push sites through pre-lease work, then structured capital to fund the heavy build once leases are secured.
Timeline lens
This press release does not give a full facility size, pricing, maturity, or covenants, so I’m not going to guess. What it does give is intent and sequencing.
The facility was entered on Dec 18, 2025.
The first $100M is tied to “advanced-stage negotiations” with an investment-grade hyperscaler for “multiple campuses.”
In practice, this kind of development capital usually goes into the work that de-risks a campus before a tenant signs or before project financing closes: site control, early engineering, interconnect planning, queue work, permitting, long-lead equipment planning, and early civil packages. If this money is being drawn now, it suggests $APLD is trying to be “ready to move” the moment the customer gives final internal approval.
Total cost of ownership lens
TCO is the real story in AI infrastructure, and it matters differently for each side of the table.
For the hyperscaler or AI cloud customer, the biggest TCO drivers are time-to-power, power efficiency, cooling capability, and execution risk. The cost of delay is brutal because GPUs sitting idle are not just wasted capex, they are missed revenue and missed training cycles. If a developer can consistently deliver powered capacity on schedule, the customer can treat that as a TCO win even if the rent line item looks higher than a self-build spreadsheet.
Applied Digital sells that “purpose-built” angle hard in its North Dakota campuses. For example, their Polaris Forge 2 campus release highlighted projected PUE of 1.18 and near-zero water consumption, built for high density and liquid cooling.
For Applied Digital, TCO is dominated by cost of capital during development, change orders during construction, and the probability-weighted cost of a miss. Pre-lease work is expensive, but failing to do it early is often more expensive because it shows up later as schedule slip, redesign, or equipment lead-time pain.
This facility helps on $APLD’s side of the TCO equation by funding the pre-lease phase without forcing the company to either slow down or lean entirely on equity. It is not “free money,” it is debt that has to be carried, but if it shortens the path to signed leases and keeps builds on track, it can lower the total project cost per delivered MW once the campus is financed and stabilized.
How it benefits the business model
Applied Digital is building an assembly line of large AI campuses. Their recent track record shows what they’re trying to repeat.
They reported fiscal Q1 2026 revenue of $64.2M, up 84% YoY, and noted a big portion of the jump came from tenant fit-out services tied to the HPC hosting business.
On the leasing side, they’ve been stacking long-duration contracts. Their Aug 29, 2025 release said Polaris Forge 1 would deliver 400MW total to $CRWV CoreWeave across three leases, with the first 100MW expected ready in Q4 2025, the second 150MW in mid-2026, and the third building targeted for 2027, with total prospective contracted lease revenue cited around $11B.
Then Oct 22, 2025 they announced an additional hyperscaler lease at Polaris Forge 2 for 200MW and about $5B over an estimated 15-year term, with phased delivery starting in 2026 and reaching 200MW in 2027.
The development loan facility is aimed at feeding that machine. More sites, earlier. More shots on goal. Less waiting around for the lease signature before doing anything meaningful.
My Sentiment
Bullish for execution.
This is not a hype headline, it’s a plumbing headline. Plumbing is what separates “AI infrastructure platform” from “cool idea with a slide deck.” If $APLD can use this facility to consistently turn high-quality sites into signed campus leases, it improves their speed, their negotiating posture, and their long-term economics.
The thing to watch is simple: do we get a named tenant announcement and a clear campus timeline tied to these “multiple campuses,” and do they keep their existing campus deliveries on schedule.
It's not a sprint it's a marathon
My PT: 36$
Patrick Mahomes on 4th down: “This fu***** never fu***** works.”
Tricks the Raiders defense to think he’s not gonna call the play…
Proceeds to snap the ball and get the first down.
Brilliant.
@T_OConnell Hi Tim, I’ve read through your DD on $LQDA and was wondering what risks you see if any on the PDUFA on the 24th? Of would any hurdles be seen with a UTHR appeal before hand? Where do you think it all stands? And upon assumed approval what does that mean for revenues?
Nakamoto is about to go parabolic 🚀
“It will focus on acquiring and holding bitcoin and is set to merge with an existing Nasdaq-listed company in a transaction that’s expected to be announced early next week”
@smatthewschultz road to 50 EH/s… hope we can get there a little early! Apr/May? Would love to see some of that capital from the 0% int convertible put towards expediting 60 EH/s and beyond. Press on. $CLSK
Eagles are rewarding Saquon Barkley with a two-year, $41.2 million contract extension that makes him the highest-paid running back in NFL history, per sources. The deal makes Barkley the NFL’s first $20 million+ per-year running back. Barkley also has the ability to earn an additional $15 million in incentives and escalators. The deal includes $36 million fully guaranteed at signing.
The Eagles are taking the rare step of extending Barkley one year into his three-year deal - and in the same league year in which they signed him to the initial contract.