Core Scientific eyes $25-$30 billion valuation with new AI data center contracts: Bernstein

In an interview with analysts at research and brokerage firm Bernstein, Core Scientific CEO Adam Sullivan said the Bitcoin BTC +3.27% miner can grow “exponentially” over the next few years via its AI data center services.

However, that depends on whether or not the firm can demonstrate the ability to deliver gigawatt scale AI data centers beyond its initial client, CoreWeave. “Signing up another 500MW or 1GW of deals over the course of the next few years puts us in a position where we’re not a $2.5 billion company, we are potentially a $25 to $30 billion company,” Sullivan told Bernstein analysts.

Sullivan, a former investment banker, joined the company in April 2023, about four months into its Chapter 11 bankruptcy process following the collapse of bitcoin’s price in the 2022 bear market, which coincided with soaring energy prices. However, after successfully exiting bankruptcy in January this year, Core Scientific signed a 12-year deal with AI Hyperscaler CoreWeave, potentially worth up to $3.5 billion in total revenue.

Core Scientific previously worked with CoreWeave on Ethereum GPU mining before it switched to a proof-of-stake consensus mechanism, rekindling conversations with the firm at the end of 2023 about the potential for conversions, Sullivan said.

So far, Core Scientific has signed three contracts with CoreWeave, adding 70MW and 112MW deals to its first agreement for 200MW. These contracts are for conversions of existing Core Scientific Bitcoin mining facilities to deliver CoreWeave GPU-ready data centers throughout 2025 and 2026. Sullivan said that some Bitcoin mining rigs at converted sites may be sold, but most will be redirected to other Core Scientific sites.

While the Core Scientific CEO claimed no other data center company is delivering that size, and they have also attracted top talent from the data center industry amid the CoreWeave deal, the niche the firm has carved out for itself is its shorter timelines.

“When we talk about some of our competitors like Digital Reality, Equinix, Switch or Cyrus One, their timelines are like three to five years in terms of when they can deliver to you infrastructure. We kind of play in that sub-three year category. So that’s really the niche that we have carved out for us and now in the existing infrastructure and sites, it's all about being able to deliver on the timeline that’s faster than the other peer set in the data center world,” Sullivan said.

Other Bitcoin miners have yet to compete in AI services

Despite the diversification opportunities of high-performance computing (HPC) services like AI data centers, Core Scientific has not seen another Bitcoin miner in any of the deals it has competed on, according to Sullivan.

Sullivan said a few Bitcoin miners have specific sites that are interesting on the HPC side, sitting along major fiber lines and close enough to major metropolitan areas they could fit well into. However, their teams may lack traditional data center experience, so they would also need to absorb additional talent to be successful in diversifying.

“I think there's still a lot of questions around whether this thesis is real, and having another mining company actually sign a contract is important,” Sullivan noted. “My biggest worry is that they aren't successful in terms of executing, and that hurts us. But I'm very hopeful that another Bitcoin mining company is able to sign a contract on the HPC side.”

Core Scientific's diversification plans are already in full swing. Just 500MW of the 1.2GW the firm has contracted now is only for Bitcoin mining, Sullivan explained, mainly because those are not convertible to HPC.

“Obviously, the revenue and gross margins are much higher on the HPC side, and the more contracts we can sign, the better the long term valuation of the business would be. So for the Bitcoin mining side of the business, we're just very focused on return on cash,” he said. 

“We're not going to deploy a significant amount of capital into refreshes unless there is a very good opportunity and something very economical for us to do. That was the reason why we signed the Block deal for chips. Right now, between 20 to 25% of all revenue generated in the Bitcoin mining side goes to refresh or essentially maintain capex, that is just to maintain your portion of the network without really any growth,” he continued.

To stay competitive in the long term, the firm’s goal is to reduce this cost by shifting from replacing entire shoebox machines to just hashboard replacements, avoiding unnecessary chassis and PDU replacements. “Our goal is to be the best return on capital Bitcoin mining business you can possibly run,” Sullivan added. 

Bernstein analyst Gautam Chhugani raised the point that Bitcoin mining is essentially a hash rate competition, where failing to invest at the pace of hash rate growth results in dilution. In response, Sullivan explained that their focus is on growing hash power through refreshes rather than building large amounts of new infrastructure. They are currently developing some new infrastructure, such as spending $13 million for an additional 100MW in Pecos, Texas, but overall, their strategy is to maximize the economic efficiency of existing sites. By focusing on profitable operations through the next two halvings and maintaining well-performing older machines, they aim to generate strong cash flows without heavy infrastructure investment.

Chhugani also pointed out that the company doesn't keep any bitcoin on its balance sheet, unlike some of its Bitcoin mining rivals. “For the next 12 months, we're an equity risk story until we begin to generate significant cash flows from the HPC business,” Sullivan explained. “So right now, we're focused on how we stabilize our cash flows, given we are at record low hash prices.”

Bernstein currently rates Core Scientific as “outperform,” with a price target of $17 — 70% above the stock’s $10 value at market close on Monday.