Core Scientific, Inc. (NASDAQ:CORZ) announced a major infrastructure partnership with Advanced Micro Devices, Inc. (NASDAQ:AMD) on July 28, giving the chipmaker’s ecosystem access to more than 500 megawatts of U.S. data-center capacity beginning in 2027. The arrangement can expand to 2.5 gigawatts. Core Scientific shares rallied in premarket trading.

The agreement is more substantial than the initial announcement suggested. Core Scientific’s earnings release described 15-year agreements covering approximately 530 megawatts across five sites, with more than $14 billion of potential base contracted revenue. Its regulatory filing provided an important distinction: AMD directly leased 377 megawatts, while an unnamed neocloud leased another 152 megawatts under agreements that give AMD certain equipment protections and rights if that customer defaults.

The larger story, however, began nine months earlier. Core Scientific shareholders rejected an all-stock acquisition by CoreWeave whose announcement-date implied equity value was approximately $9 billion. The fixed exchange ratio valued CORZ at $20.40 per share when the transaction was announced in July 2025, but the value shareholders would have received at closing was not fixed and moved with CoreWeave’s share price.

In January, Gullane Capital Partners founder Trip Miller, who had opposed the sale, predicted that Core Scientific would secure new AI customers. “I expect them to announce deals for AI with third parties other than CoreWeave,” he told Business Insider.

The new agreements appear to deliver precisely that customer diversification. Taken together, AMD’s 377-megawatt direct lease and the neocloud’s 152-megawatt lease exceed Miller’s roughly 400-megawatt expectation, although AMD itself did not directly lease the full 529 megawatts. The question is whether the agreements prove that shareholders were right to preserve Core Scientific’s independence, or merely give the company a large, capital-intensive opportunity whose ultimate value remains uncertain.

BULL CASE

AMD agreement validates the central argument shareholders made when they rejected CoreWeave: Core Scientific’s power portfolio could attract major customers beyond its existing tenant.

Before the new agreements, CoreWeave remained Core Scientific’s only meaningful high-density colocation customer. Core Scientific had leased approximately 590 megawatts to CoreWeave, and a single customer generated 77% of the company’s first-half revenue.

The new leases almost double total leased customer power capacity to approximately 1.1 gigawatts. Core Scientific said that capacity now represents more than $24 billion of potential contracted revenue, including the existing CoreWeave relationship and the new agreements connected to AMD.

This matters because the independence case depended on more than rising demand for AI infrastructure. Core Scientific had to demonstrate that its sites, power access, construction capabilities, and delivery record were attractive to counterparties other than CoreWeave.

The AMD ecosystem provides that validation. AMD’s direct 377-megawatt commitment is spread across sites in Pecos, Muskogee, and Hunt County. The additional 152 megawatts leased by a neocloud across two other sites will support AMD equipment, with AMD receiving contractual protections surrounding that equipment.

Together, the initial 529 megawatts of critical IT capacity are already comparable with Core Scientific’s entire 590-megawatt CoreWeave relationship. If AMD converts its reservation rights for another 1,925 megawatts into leases, the partnership could eventually dwarf CoreWeave’s position.

The agreement also strengthens Core Scientific’s transition away from bitcoin mining. Colocation generated $136.7 million of the company’s $164.2 million in second-quarter revenue. The business is increasingly becoming a data-center landlord rather than a bitcoin miner attempting an AI pivot.

From this perspective, the shareholder vote preserved exactly the upside that CoreWeave wanted to acquire. Core Scientific retained ownership of its power pipeline, found another major technology partner, and secured long-term contracts without surrendering the company.

BEAR CASE

The bear case is that validating the strategic premise behind independence is not the same as proving that rejecting the sale created more value for shareholders.

The headline comparison between more than $14 billion of potential contracted revenue and a rejected $9 billion acquisition is tempting but misleading. The $9 billion figure was the announcement-date implied equity value of the proposed stock consideration, with the final value dependent on CoreWeave’s share price at closing. The AMD-linked figure is potential base revenue collected over 15 years before construction costs, operating expenses, financing costs, taxes, and the time value of money.

The full 2.5-gigawatt opportunity also should not be treated as contracted capacity. The signed leases cover 529 megawatts of critical IT load. AMD has only a reservation right for the additional 1,925 megawatts through December 2028, subject to specified timing and other conditions.

Even the initial capacity requires qualification. AMD is the direct tenant for 377 megawatts. The remaining 152 megawatts is leased to an unnamed neocloud. AMD can cure certain defaults and has protections over its equipment, but those arrangements are not identical to AMD directly guaranteeing every payment under the neocloud leases.

The warrants create potential dilution, although exercise would also provide cash to Core Scientific. AMD received the right to purchase up to 30 million Core Scientific shares at $23.47 per share, with vesting tied to contracted capacity. Warrants covering approximately 6.5 million shares vested and became exercisable when the initial agreements were signed; the underlying shares were not issued at that point.

The shares underlying the vested portion represent roughly 2% of Core Scientific’s 321.3 million shares outstanding on July 23. The full 30-million-share warrant equals about 9.3% of the current share count, or approximately 8.5% of the post-exercise total if exercised in full. A cash exercise of all 30 million warrants at the stated strike price would also provide Core Scientific with approximately $704 million.

The capital requirements may be more important than the potential dilution. Core Scientific had approximately $4.4 billion of borrowings at the end of June, including $3.3 billion of 7.75% senior secured notes issued in May. It was also committed to approximately $1 billion of future construction and purchase expenditures, of which only $264 million was expected to be passed through to a customer.

Core Scientific’s existing CoreWeave conversions are funded almost entirely by that customer. The company warned that new sites, including Hunt County and Muskogee – two locations involved in the AMD leases – do not have the same customer-funding structure. Complete project-funding terms for the new agreements were not disclosed.

The partnership therefore shifts the test from customer acquisition to execution. Core Scientific must build the capacity on time, control construction costs, arrange suitable project financing, and generate returns that justify the construction spending, financing costs, and potential dilution from the warrants issued in connection with the leases.

INSIDER MONKEY’S HEDGE FUND DATA ANALYSIS

Insider Monkey’s hedge fund database shows that 81 hedge funds held positions in Core Scientific, Inc. (NASDAQ:CORZ) at the end of the first quarter of 2026, compared with 76 funds at the end of the preceding quarter.

These holdings reflect positions as of March 31, 2026. They do not capture subsequent trades or investors’ reactions to the AMD partnership.

CONCLUSION

The AMD partnership substantially vindicates the strategic argument behind rejecting CoreWeave. Core Scientific has demonstrated that it can attract another major technology partner, while the 529 megawatts tied to the AMD ecosystem – 377 megawatts leased directly by AMD and 152 megawatts leased by a neocloud – are nearly as large as its entire CoreWeave relationship.

It does not yet vindicate the economic outcome of the shareholder vote. More than $14 billion of potential revenue over 15 years cannot be compared directly with the proposed transaction’s approximately $9 billion announcement-date implied equity value, particularly when complete project-funding terms remain undisclosed, Core Scientific carries substantial debt, and the warrants could dilute shareholders if exercised – although exercise would also provide cash to the company.

The fairest conclusion is that shareholders have been proven right about Core Scientific’s customer appeal, but not yet about the value ultimately available to them.

The next test is whether Core Scientific can begin delivering the initial 529 megawatts in 2027, as management expects, at returns that support the independence case and convert AMD’s remaining 1,925-megawatt reservation into firm leases without disproportionate financing costs or dilution.

READ NEXT: 33 Stocks That Should Double in 3 Years and 15 Stocks That Will Make You Rich in 10 Years 

Disclosure: None. Follow Insider Monkey on Google News.