Core Scientific laid the footprint for strong growth in HPC (colocation) revenue by the end of the year as it begins to bring online billable capacity for CoreWeave and subsequently rapidly ramp capacity of 250MW. This is the riskiest stock in our portfolio when you consider the current fundamentals are (at face value) of poor quality with revenue declining (55%) YoY and (16%) QoQ to $79.5 million this quarter compared to $179.3 million in the year ago quarter. The company reported an operating loss of $42.6 million and negative adjusted EBITDA of ($6.1) million.

We’ve covered Core Scientific transitioning from Bitcoin mining operations to data centers in more detail here. The company noted that it remains on track to deliver 250MW of billable capacity to CoreWeave by the end of 2025, with the first 8MW tranche to be delivered by the end of May, expanding 5x to 40MW by the end of Q2. According to the earnings call, the Denton facility is a site CoreWeave is working on with Open AI with a recent $12 billion investment.

On the call the company stated: “At full scale, the site will represent around 260 megawatts of billable capacity. To put that in perspective, we broke ground in January, and in just roughly four months, we've achieved meaningful progress. It's a powerful demonstration of our ability to execute quickly and at scale, with before-and-after pictures included in our updated investor presentation.”

Refresher on Core Scientific's Business Model

Before we go into the earnings report, I think it’s appropriate to pause and review Core Scientific's business model given it’s quite unique, and to also help translate what the company is setting up to do in terms of future revenue.

Core Scientific was a major Bitcoin miner that is transitioning to power AI data centers with their primary customer being CoreWeave worth about $10.2 billion when fully realized over a 12-year period for 590 MW of HPC infrastructure. Roth MKM sees CORZ having about $1.6B in revenue by 2027, up from $473M expected today.

CoreWeave, the primary customer, fronts the capex which allows Core Scientific to grow capacity, without which, the business model would not work as CORZ would struggle to raise the level of capital required to acquire more sites and modify the existing infrastructure.

Here is what was shared regarding CoreWeave putting up the capex costs in the most recent earnings call: “From a capital perspective, CoreWeave is funding virtually all of the CapEx associated with these deployments. Our only direct capital outlay on the contract is the $104 million associated with the 70-megawatt expansion we announced during our last earnings call. That structure significantly reduces our capital burden, keeps our balance sheet leverage like compared to peers and gives us the flexibility to use debt more strategically for future growth. We believe this approach sets us apart and creates a clear path to long-term value creation.”

Revenue

Revenue continued to be impacted by Bitcoin’s halving and Core Scientific’s operational shift from Bitcoin to HPC. While the ramp in AI/HPC revenue approaches, crypto self-mining remained the primary driver of the YoY decline in Q1 as it contributed nearly 85% of revenue. Q1 revenue declined (55.6%) YoY, missing estimates by nearly 7%.

Crypto self-mining revenue declined (55.2%) YoY to $67.2 million, impacted by a (75%) decline in Bitcoin mined and shift to HPC
Crypto hosted mining revenue declined (87%) YoY to $3.8 million, again impacted by the HPC shift.
Colocation (HPC) revenue was approximately flat QoQ at $8.6 million.

Here is what was stated on the call about the revenue decline: “The sequential revenue decline was primarily driven by mining disconnections and relocations as we continue converting sites to support high-density colocation. More specifically, we earned 719 Bitcoin in the first quarter compared to 974 in the fourth quarter.”

Quickly ramping capacity for CoreWeave in Q2 through year-end is expected to drive significant growth in Colocation revenue. Management says they expect the 250MW will allow them to enter 2026 with annualized colocation revenue of ~$360 million ($90 million per quarter), up more than 10x from its $8.6 million ($34.4 million annualized) in Q1.

This ramp is expected to drive a significant rebound in Core Scientific’s revenue growth, with analysts currently expecting the company to exit 2025 at ~$160 million in revenue, approximately double Q1’s level. This would correspond to growth of 68.3% YoY, a more than 120 point acceleration as the year progresses.

However, it’s important to note that estimates have come down sharply over the past three months:

Q2 revenue was estimated to decline just (10%) to $126.6 million at the end of February, but is now seen declining (34%) to $93.1 million.
Q3 revenue was estimated to rise 84.4% to $175.8 million, but is now expected to rise 27.5% to $121.2 million.
Q4 revenue was estimated to rise 110.5% to almost $200 million, but is now expected to rise 68.3% to $159.8 million.

Despite these changes in estimates, management reiterated they are on track to reach their capacity goals: “Looking ahead, I'm even more confident than I was just two months ago in our ability to hit our milestones, 250 megawatts by the end of this year, inclusive of Austin and 590 megawatts by early 2027.”

The company pointed toward growth potential as well, stating there are additional opportunities to add the following capacity: “On the organic side, we remain confident in our ability to add approximately 300 megawatts of billable capacity across our existing sites by the end of 2027. Looking ahead, we also continue to believe there are significant opportunities to grow into new geographies, and we're targeting an additional 400 megawatts of billable capacity through new site development over the next three years.”

Margins

Gross margin expanded sequentially, but operating margin widened to more than (50%) as rising costs bit into weaker revenue. While Colocation promises to bring substantial revenue streams and strong tailwinds to growth through year-end, margins at the moment are minimal, even with power costs being passed through to CoreWeave.

Gross margin was 10.3% in Q1, expanding from 5% last quarter but well below the 43.3% margin from the year-ago quarter due to the Bitcoin halving and operational shift.
Operating margin was (53.6%), widening from (41.9%) last quarter and a stark contrast to the 30.3% margin from a year ago. The significant YoY difference was primarily caused by a more than (89%) YoY decline in gross profit and a 137% increase in SG&A expenses.

By segment:

Crypto self-mining gross profit margin was 9%, down from 49% a year ago, impacted by the shift to HPC and a (75%) decline in BTC mined, partially offset by a 74% increase in the average price of BTC and a 33% decrease in power costs.
Crypto hosting gross profit margin was 46%, up from 32% a year ago, primarily due to lower power costs.
Colocation gross profit margin was 5%.

EPS

Core Scientific benefited significantly from a $621.5 million mark-to-market adjustment on its warrants, and as a result, it reported $580.7 million in net income. This represented $1.25 in EPS, which is not comparable to the ($0.12) estimate due to the warrant impact. Stripping out this impact, net income would be ($40.8) million.

Core Scientific is currently expected to record losses through the rest of the year, shrinking each quarter from ($0.11) in Q2 to ($0.03) by Q4.

Cash and Balance Sheet

Core Scientific burned through a substantial chunk of cash in the quarter as it continues on its operational shift.

Operating cash flow was ($40.6) million for a (51.1%) margin.
Free cash flow was ($129.0) million for a (162.3%) margin, as Core Scientific’s capex rose 177% YoY to $88.4 million.
Cash and equivalents totaled $697.9 million, with Core Scientific burning through $138 million in cash in the quarter.
Debt totaled $1.12 billion. Management also shared long-term debt leverage targets on the call — per the CFO: “And over time, we believe our net debt to adjusted EBITDA leverage can and should trend toward approximately 4 times, consistent with peers in the space.”
Adjusted EBITDA was ($6.0) million for an (8%) margin, down from $88 million or a 49.1% margin in the year ago quarter.
Core Scientific also recognized $42 million in prepaid colocation license fees as deferred revenue in the quarter.

Earnings Call Q&A:

No New Customers Yet; but Enterprise Customers on the Horizon

The market will reward Core Scientific if the company can add more customers. In our previous write-up we stated the company’s goal is to have CoreWeave customer concentration to be 50% or less by 2028. This remains the goal with no updates on customer concentration improving:

“Now, to be clear, we haven't signed a new customer yet, but our sales pipeline is expanding. It includes a healthy mix of hyperscale and large enterprise customers, and we are actively negotiating with multiple customers today […] We currently have several non-hyperscale deals in our pipeline, ranging from 50 megawatt to 100 megawatt customers. These are substantial deployments, and they come with a return profile that's attractive […] I'm more confident than ever in our ability to build a customer base that is more diverse, more balanced, and more strategically aligned with our long-term vision. Our target remains the same, to have Core represent less than 50% of our billable capacity by the end of 2028.”

The advantages of enterprise customers were discussed further in the call, with Core Scientific likely needing to first prove it’s been able to stand up Blackwell systems before demand increases from a broader set of customers.

“Adam Sullivan

Yeah, it's a great question. Thanks Darren. Large enterprises, the timeline to get into final contract details are definitely faster than on the hyperscale side. There's a natural inclination to move towards hyperscale from the broader perspective of their creditworthiness. But the large enterprises that we're looking at today are I think $75 billion market cap plus and represent a creditworthiness that we find very acceptable in the return profile of these are higher than hyperscale deals as well. So, as we evaluate potential multitenant build-outs going forward, large enterprises could represent significant anchor tenants for those new sites to allow us to begin development in new geographies and start building out new sites”

The CEO reiterated again they are getting close to signing more enterprise customers:

“And so, we're currently evaluating a number of different deal structures that we're in discussions with clients. And I would say we're excited about the return profiles the large enterprises represent because they do have the capability, based on their scale and their size, that they're demanding today to represent an acre tenant for us to open up a new site location. And so, we will continue to evaluate deals going forward, and we're excited about the continuously growing pipeline in large enterprise channel.”

In time, however, Core Scientific believes it will prove itself to other hyperscalers, stating: “I think our delivery and execution is only going to breed confidence kind of as this year goes on. And again, we're probably one of the only data center providers right now that's going to launch 250 megawatts in a single calendar year of what essentially is going to be more than 100,000 of GB200.”

Tariffs

It wouldn’t be a Q1 earnings report if I did not address tariff commentary from the call. Core Scientific stated they have procured components to deliver on time this year and into the first part of 2026, although there seemed to be some hesitation when looking further out. Although management put a positive spin on it, there’s a scenario that should be monitored which is that data center builds become more expensive as we approach 2026, and thus, budgets could tighten.

Paul Golding:

Thanks so much. Just quickly a housekeeping question I wanted to ask regarding digital asset mining. I think previously you'd mentioned that the digital asset mining hosting the capacity was you were going to exit that by year end. Just wanted to confirm that that was still the plan since we saw some revenue come through this quarter for that.

And then, my main question is around long lead time items. Just referencing your commentary around 2025 goals, the equipment being acquired to meet those goals. Wanted to ask about ’26 hearing about long lead times for step down infrastructure and other components, particularly around electrical equipment, and so just wanted to see how that was progressing as well? Thank you.

Matt Brown:

Yeah. I think the way to look at this is 2025, we've already secured all that equipment and most of it is already sitting in the ground either in warehouses adjacent to the projects that are ongoing or at the project site themselves. So, 2025 is locked in from an equipment standpoint.

2026, we have a good read-through on both the availability and cost for all that equipment. A lot of that through the first half of ’26 has already been procured, and a few portions of that will start taking delivery towards the second half of this year for projects that are going to extend that are going to start launching in 2026.

With that said, I think this part of this touches on tariffs and equipment availability. With our strong relationships with our suppliers that I would say we have really, really good insights in the availability and that we're not really too concerned right now about not being able to take receipt of equipment and meeting our dates or any buy dates to meet our delivery goals.”

Conclusion:

Core Scientific is the highest risk stock in our portfolio as it takes a leap of faith that the partnership with CoreWeave is setting a standard in terms of standing up and powering up data centers very quickly. This quarter the company is starting to transition toward AI revenue rather than bitcoin revenue (i.e., primarily Bitcoin losses). According to analyst estimates, CORZ looks to be returning to revenue growth by Q3 which gives you a good idea as to when AI should be leading the market again as CoreWeave is a strong proxy for when Nvidia will resume its product cadence.

We feel confident taking on the challenge of owning CORZ although it will require an active stance with risk management controls in place.

Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.

Recommended Reading:

Core Scientific Q1: Expects 250MW of Billable Capacity to CoreWeave by Year-End