Quantum Sells 1,000 ETH, Hyperscale Borrows BTC to Fund AI Data Centers

iconChainGPT
Share
AI summary iconSummary
On-chain data shows two listed firms are using crypto treasuries to fund AI infrastructure. On July 30, Tokyo-listed Quantum Solutions sold 1,000 ETH at $1,903 per token, netting $1.903 million. U.S.-listed Hyperscale Data monetized about 100 BTC and set up a Bitcoin-backed credit facility. Quantum’s ETH sales support its AI Infrastructure Data Center (AIDC) business, including GPU equipment and data-center contracts. Hyperscale is using BTC proceeds to develop a Michigan AI data-center campus under a master services agreement. Both companies reflect a trend of crypto treasury holders converting assets into capital for AI and high-performance computing infrastructure amid shifting inflation data.

Crypto treasuries are shifting from hodling to funding AI infrastructure, with two listed firms converting digital assets into capital for data centers. What happened - On July 30, Tokyo-listed Quantum Solutions and U.S.-listed Hyperscale Data each tapped crypto holdings to fund AI data-center plans. - Quantum’s GPT Pals Studio sold 1,000 ETH at $1,903 per token, netting $1.903 million after fees. The company expects an accounting loss of ¥17 million (about $100,970) because the sale price was below its May 31 carrying value of $2,003.97 per ETH. - This follows a June 16 disposal of 904 ETH for roughly $1.61 million. Combined, the two sales brought in about $3.51 million and reduced Quantum’s ETH treasury from 6,668.8 ETH to 4,764.8 ETH — a 28.6% decline. - Quantum raised its cumulative sale authorization from 1,875 to 4,375 ETH through October 30, meaning it could sell another 2,471 ETH, though the company stressed this is not an immediate decision and future sales will depend on ETH prices, funding needs and progress on its AI Infrastructure Data Center (AIDC) business. - Of Quantum’s remaining 4,764.8 ETH, 3,050 ETH are pledged as collateral to a Singapore-based financial services firm; 1,714.8 ETH remains in GPT Pals’ trading account. The unused sale authorization exceeds freely held ETH by 756.2 ETH, implying that selling the full authorized amount would require releasing collateral, buying ETH, or another arrangement — none of which Quantum has announced. - The disposals also appear to have cost Quantum its status as Japan’s largest listed ETH holder: Def Consulting reported 4,976 ETH on June 30, about 211.2 ETH more than Quantum’s post-sale balance (reporting dates differ). Why Quantum is selling - Quantum says proceeds will support data-center contracts, GPU equipment and preparations for its AIDC business. In June it signed a nonbinding memorandum of understanding with Hong Kong-based Integrated Capital to explore financing and resource cooperation for a Japanese AI data center geared toward NVIDIA B300 and GB300 GPUs. The MOU is nonbinding, and Quantum says specific investment amounts, financing terms and timing remain undecided — so the ETH sales provide liquidity and flexibility rather than confirm a finished deal. Hyperscale’s approach: sell and borrow - Hyperscale monetized about 100 BTC and set up a Bitcoin-backed credit facility with an expected variable interest rate of roughly 4.5%–5%. The company did not disclose the lender, borrowing limit, maturity, collateral ratio or the exact amount of BTC pledged. - Hyperscale had reported 1,106.0467 BTC on July 27 (about $71.7 million). Subtracting the approximate 100 BTC sale would leave roughly 1,006 BTC, though the company hasn’t confirmed post-transaction balances. - The proceeds are being directed to a Michigan AI data-center campus under a definitive master services agreement with a neocloud provider. Initial deployment is roughly 20 megawatts with a ten-year term and two optional five-year extensions; Hyperscale estimates the maximum term could generate more than $1.2 billion in revenue. The customer also has an option for another 32 MW, which Hyperscale says could push potential contract revenue above $3 billion if exercised. Those are conditional forecasts, not booked revenue. Bigger picture - These moves highlight a growing trend: companies holding crypto treasuries are becoming active funding sources for AI and high-performance computing (HPC) infrastructure, instead of passively holding assets. Quantum’s ETH sale is aimed at early-stage project financing, while Hyperscale paired a BTC sale with collateralized borrowing to back a signed customer project. - Similar activity has been seen elsewhere: Core Scientific sold 2,385 BTC in Q1 to fund AI capex, and listed miners have disclosed more than $70 billion in announced AI and HPC contracts while selling Bitcoin to finance development. What to watch next - Quantum’s expanded ETH sale authorization expires October 30; it has pledged to disclose any additional sales and expects to record the ¥17 million loss in its fiscal second quarter. - Hyperscale has said it will provide further construction, financing and operational updates. Investors will be watching for full credit-facility terms, a clearer post-transaction BTC balance, and delivery progress on the Michigan capacity. Takeaway Crypto treasuries are being used as operational capital for an AI buildout — through outright sales, and by pairing sales with crypto-backed lending. The approach reduces on-paper crypto exposure but accelerates real-world AI infrastructure deployment, underscoring a strategic pivot among crypto-linked infrastructure companies.

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.