WhiteLine Daily gathers the thinking of the Wu Shuo team to provide readers with the most valuable information and analysis of the day, and seize the changing trends in the AI ​​era. 1. Amazon plans to sell and lease back $8 billion of Nvidia chips According to the Financial Times, citing people familiar with the matter, e-commerce and cloud services company Amazon is discussing with investors to transfer Nvidia's Grace Blackwell chips worth about $8 billion into a special purpose vehicle (SPV) and then lease them back for use. The vehicle plans to introduce external funds through bond issuance to provide financing for chip assets. The chips involved were previously purchased or leased by Amazon and are currently distributed in more than ten data centers in five states in the United States. The deal is still being negotiated, and Amazon and Nvidia have not yet responded to Reuters' requests for comment. One-sentence analysis: Amazon hopes to introduce external funds while retaining the right to use computing power, but the cost of easing the pressure on purchasing funds is the subsequent continued payment of rent. 2. Broadcom-related banking group begins to raise US$60 billion in chip financing A group of Wall Street banks related to U.S. semiconductor and infrastructure software company Broadcom is starting to raise $60 billion to fund the acquisition of chips and AI infrastructure for companies such as Claude developer Anthropic, Bloomberg reported, citing people familiar with the matter. The package is divided into debt in different repayment orders: the bank is preparing to raise $42 billion in senior secured debt; alternative asset manager Blackstone is leading another $18 billion in subordinated debt, with its funds committing to provide $9 billion of it, and the rest is planned to be raised from other investors. This plan has been under negotiation before, and the latest progress is that it has entered the fundraising stage and disclosed the debt stratification and specific funding commitments; it has not yet been officially announced, and the overall fundraising has not been completed. One-sentence analysis: Broadcom connects customers' chip procurement needs with external debt financing, and the progress of fundraising will affect the pace of customers' expansion of computing power. 3. Toshiba plans to invest 60 billion yen to expand production and double its AI data center hard drive production capacity target According to the "Nihon Keizai Shimbun", Japanese electronics and infrastructure company Toshiba plans to invest approximately 60 billion yen to expand its factory in the Philippines, aiming to double the production capacity of mechanical hard disks (HDD) for AI data centers in fiscal 2027. This will be the company's first large-scale investment in the HDD business in about five years. In addition to adding production lines, the expansion also plans to introduce new products with higher single-disk storage capacity and increase the automation of inspection and clean room operations. Toshiba's current HDD market share in terms of storage capacity is just over 10%, with a mid-term goal of increasing it to 30%. One-sentence analysis: AI storage demand is driving additional investment by HDD manufacturers, but Toshiba’s production expansion target is pointed at fiscal year 2027, and supply improvement will still take time. 4. Google data center power supply contract is approved and must bear the cost of new power supply The Michigan Public Service Commission conditionally approved Search's 20-year power supply contract with cloud services company Google and local power company DTE Electric for Google's local data centers. Google will bear the development costs of up to 1,600 MW of renewable energy and 480 MW of battery storage, with facilities built, owned and operated by DTE. The contract also requires a minimum billing of 80% of the agreed electricity demand; even if terminated early, Google will ultimately be responsible for minimum monthly charges for at least 15 years. Regulators require that project-related costs and financial risks must not be passed on to other electricity users. One-sentence analysis: While Google obtains power supply arrangements, it also assumes new facilities and minimum payment obligations. Even if there is insufficient computing power demand in the future, the corresponding costs still need to be absorbed by the company. Today's main line AI infrastructure needs are bringing new financing and capacity investment. Chip leaseback and long-term power supply contracts can alleviate immediate financial or resource constraints, and will also increase future fixed expenses. Project returns are more dependent on whether computing power can continue to be used. Whether financing can be obtained and when production capacity can be realized will also affect the pace of subsequent expansion. If demand is lower than expected, rent, debt service and minimum electricity bills may still continue to occur, and expansion and profit growth may not be synchronized.