The 500MW Bait: Core Scientific's AMD Deal and the Hidden Dilution Trap
Core Scientific stock popped 20% on the news. AMD is building a 500MW AI data center together. The market cheers. I read the fine print.
30 million warrants. That's the number everyone ignores. Not the megawatts. Not the partnership. The warrants.
Context: The Mining Pivot Playbook
The Bitcoin mining industry is bleeding. Hashprice is down 60% from 2021 peak. Public miners are desperate for a new narrative. Enter AI. The script is simple: convert cheap power capacity into high-value compute for AI training. Core Scientific, fresh out of bankruptcy restructuring, is the first to throw a punch of this scale.
500 megawatts. That's enough to power 400,000 homes. Or one of the largest AI data centers in North America. AMD will supply the chips. Core Scientific will supply the power, the racks, the cooling, the ops. And as part of the deal, Core Scientific hands AMD 30 million warrants—the right to buy CORZ shares at a predetermined price.
This is not a standard supplier agreement. This is a strategic partnership disguised as a financial instrument.
Core: The Warrant Mechanics
30 million shares. At current trading volume, that's roughly 20-25% of the fully diluted float. If AMD exercises those warrants at, say, $5 per share (a common structure), they would inject $150 million into Core Scientific's treasury. But here's the catch: the warrants are long-dated, probably 5-7 years. AMD isn't paying cash upfront for the chips; they are being compensated in equity that only has value if CORZ stock rises.
From a forensic risk perspective, this is a bet on Core Scientific's future stock price. AMD is effectively saying: "We believe in your AI pivot so much that we'll take our payment in shares rather than cash." That sounds bullish. But it also means Core Scientific is diluting its existing shareholders by up to 25% to secure a chip supplier. The question every trader should ask: is the AI revenue potential worth the dilution?
Let's run the numbers. Assume the data center generates $200 million in annual revenue at full capacity (conservative for 500MW). At a 10x revenue multiple (typical for data center REITs), that implies a $2 billion market cap add. Current CORZ market cap is ~$800 million. Add $2 billion, you get $2.8 billion. Divided by 180 million shares (post-dilution from warrants), that's ~$15.5 per share. Today CORZ trades at $4.50. The upside is real—if the execution hits.
But execution is where the trap lies.
Contrarian: The Blind Spots
I've audited enough smart contracts to know that fine print kills narratives. Here, the fine print is threefold.
First, conversion timing. If AMD is smart, they will exercise warrants after the stock rallies on positive customer announcements. That means they buy shares at a discount to market, then dump or hold. Either way, the overhang depresses price. Smart money will front-run this—sell before the dilution hits.
Second, technology risk. Converting a Bitcoin mining facility to an AI data center is not plug-and-play. Bitcoin mining uses ASICs; AI training uses GPUs. Different power distribution requirements (GPUs are more sensitive to voltage fluctuations), different cooling (liquid cooling vs. air), different network topology (high-bandwidth InfiniBand vs. simple Ethernet). Core Scientific has operational expertise in mining, but AI data centers require engineers who understand cluster scaling, job scheduling, and customer SLAs. Hiring is a bottleneck.
Third, competition. Core Scientific is not the only miner pivoting. Riot Platforms is building its own 1GW facility for AI/HPC. Hut 8 just signed a deal with a cloud provider. And then there's the cloud giants: AWS, Azure, Google Cloud are all spending billions on AI infrastructure. Core Scientific's 500MW is a rounding error to them. Their differentiation—low-cost power—is not unique. Power is a commodity. The real moat would be lock-in with application frameworks (e.g., CUDA for NVIDIA chips), but AMD's ROCm is not there yet.
This is where my 2020 DeFi liquidity sprint experience kicks in. I remember when Uniswap launched on Optimism in 2021. Everyone rushed to provide liquidity for the "next big L2." Total value locked hit $200 million. Then the incentives dried up, and the TVL dropped 80% in three months. The yield was bait; the exit liquidity was the hook. Here, the yield is the AI revenue narrative. The exit liquidity is the warrant dilution and the eventual capital raise to fund construction.
Takeaway: What to Watch
The market is pricing in a successful pivot. But the probability of flawless execution is low. The smart money will watch two signals:
- First customer announcement. If Core Scientific lands a large AI lab (e.g., Mistral, Stability AI, or a Fortune 500 enterprise), that validates the partnership. If not, the data center will sit empty, burning CapEx.
- Warrant exercise price and timing. If AMD exercises early, that signals they see a near-term catalyst. If they hold, they are betting on long-term appreciation—but that also means the overhang persists.
Until we see concrete revenue, this is a story stock with a 30M share anchor. Patience is for traders; timing is for killers. I'll wait for the first quarterly earnings where AI revenue is broken out. Until then, I'm watching the order book for the exit.