Alphabet has finalized a complex capital strategy to secure high-density artificial intelligence infrastructure, combining an $84.75 billion convertible note issuance with a multi-year hardware lease agreement. A Securities and Exchange Commission filing dated June 5 confirms the company will deploy these funds to maintain its competitive position in the global cloud computing market.
The financing structure involves the sale of 167.5 million Series A and 167.5 million Series B depositary shares, each representing a one-twentieth interest in a 6.25% convertible note. These notes carry a liquidation preference of $1,000 per share and are scheduled to pay quarterly dividends starting August 15, 2026.
The deal, which officially closed on June 5, includes an overallotment option that banks fully exercised on June 3. Alphabet implemented private capped-call transactions to mitigate the potential dilutive impact of the convertible notes on common shareholders.
These contracts establish conversion price caps at $532.67 for Class A shares and $527.80 for Class C shares. The company clarified that these derivatives are independent of the depositary shares, meaning investors in the notes hold no claims on the underlying capped-call agreements.
The capital raised is specifically earmarked for the expansion of global computing capacity, a necessity driven by demand that currently exceeds available supply. According to a June 1 filing, Alphabet projects 2026 capital expenditures to fall between $180 billion and $190 billion.
This aggressive spending trajectory is designed to support the rapid growth of Google Cloud, which reported a 63% revenue surge to $20 billion in the first quarter of 2026. A significant portion of this investment is directed toward a newly disclosed cloud agreement with SpaceX.
Under the terms of this contract, Alphabet gains access to approximately 110,000 NVIDIA graphics processing units, alongside essential CPUs and memory components. The lease agreement mandates monthly payments of $920 million to SpaceX, spanning from October 2026 through June 2029.
The contract includes specific performance safeguards to protect Alphabet’s interests in the event of supply chain delays. If SpaceX fails to deliver the stipulated GPU capacity by September 30, 2026, Alphabet retains the right to terminate the agreement after a one-month cure period or negotiate a proportional reduction in fees.
Ownership of all data, AI models, and proprietary content remains exclusively with Alphabet throughout the duration of the lease. The scale of this hardware commitment highlights the extreme capital intensity required to maintain leadership in the current data center sector.
While Alphabet reported total revenue of $109.9 billion for the first quarter of 2026, the $35.7 billion spent on property and equipment during the same period illustrates the heavy cash requirements of modern data center operations. The company maintains an operating margin of 36.1%, yet the pressure to sustain this growth while balancing shareholder dilution remains a primary concern for market analysts.
Investors are now evaluating whether the long-term utility of the secured GPU capacity justifies the financial leverage Alphabet has assumed. The company itself has acknowledged that the hedging activity associated with the capped-call partners could introduce volatility into its stock price.
Market participants will look toward upcoming macroeconomic indicators, including the June 10 consumer price index release and the Federal Reserve meeting on June 16-17, to gauge the broader cost-of-capital environment. The ultimate success of this infrastructure strategy hinges on the company’s ability to convert the massive GPU allocation into sustained revenue growth.
As the September 30 delivery milestone approaches, the effectiveness of the SpaceX partnership will serve as a key indicator of Alphabet’s operational agility. Future disclosures will likely focus on the integration of this hardware into existing data center architectures and the resulting impact on model inference latency and throughput metrics.