What Alphabet’s AI Bet Means
Alphabet’s $80 billion equity raise and its agreement to buy AI computing power from SpaceX together describe a new phase in the AI race where success depends on gigantic investments in specialized data centers, chips, and cloud infrastructure rather than software alone. Alphabet, Google’s parent company, plans to channel the fresh equity into Alphabet AI infrastructure, expanding global capacity for its Gemini models and broader AI services. This move follows an increase in its 2026 capital expenditure outlook to USD 180–190 billion (approx. RM828–874 billion), underlining how capital‑intensive frontier AI has become. At the same time, Google’s decision to pay SpaceX for access to xAI’s data centers shows that even a company with extensive facilities can face short‑term shortages of AI computing power as enterprise demand spikes. Together, these decisions reveal both competitive pressure and a long‑term infrastructure strategy.
Inside Alphabet’s $80 Billion Capital Raise
Alphabet’s new Google capital raise is one of the largest stock fundraising efforts by a listed company. The USD 80 billion (approx. RM368 billion) plan comes despite Alphabet holding more than USD 126 billion (approx. RM580 billion) in liquid assets and remaining highly profitable. The package has three layers: USD 30 billion (approx. RM138 billion) through public offerings, a USD 40 billion (approx. RM184 billion) at‑the‑market share sale program, and a USD 10 billion (approx. RM46 billion) private investment from Berkshire Hathaway. Berkshire will buy USD 5 billion (approx. RM23 billion) of Class A and USD 5 billion (approx. RM23 billion) of Class C shares, lifting its total Alphabet stake above USD 26 billion (approx. RM120 billion). According to The Tech Portal, this marks a notable shift from Alphabet’s past reliance on internal cash, as the company has also raised about USD 85 billion (approx. RM391 billion) in debt over the past year.
The SpaceX Deal: Buying Time and GPUs
Google’s USD 30 billion (approx. RM138 billion) agreement with SpaceX underscores how scarce high‑end AI computing power has become. SpaceX’s filing says it will receive USD 920 million (approx. RM4.2 billion) per month from Google in exchange for access to computing power from xAI’s data centers starting in October and running through June 2029. The deal includes access to 110,000 NVIDIA GPUs, CPUs, and memory, with contractual rights for Google to terminate or pay less if that GPU count is not delivered by September. A Google Cloud spokesperson described the arrangement as “a short-term, timely agreement” to provide bridge capacity for Gemini Enterprise as demand rises faster than the company can build. SpaceX is not tying itself to Google alone; according to filings, Anthropic will pay USD 1.25 billion (approx. RM5.8 billion) a month for access to xAI’s Colossus 1 data center through May 2029.
AI Infrastructure as a Long‑Term Battleground
Alphabet AI infrastructure spending shows how the economics of AI have changed. Training and serving frontier models now demands vast data centers, custom chips such as TPUs, dense networking, storage, and reliable electricity, making AI a capital‑hungry infrastructure business. Industry‑wide AI capital expenditure is expected to reach several hundred billion dollars annually, reshaping how investors assess long‑term value in technology. Alphabet’s raised 2026 capex guidance to USD 180–190 billion (approx. RM828–874 billion) and its deepening partnership with Berkshire signal that AI is not a side project but the core engine for the next decade. The xAI deal, framed as temporary bridge capacity, shows that even as Google builds aggressively, peak demand can still outrun its own data center investment. In this environment, access to compute becomes a strategic weapon, and balance sheets become as important as algorithms.
