It's been a busy period for Alphabet (GOOGL 1.66%) (GOOG 1.69%). In mid-November, investors learned that Warren Buffett's investment conglomerate, Berkshire Hathaway, took a $4 billion stake in the internet giant during the third quarter. Subsequently, shares soared to new all-time highs.
Then on Dec. 22, Alphabet announced a definitive agreement to acquire Intersect, a provider of data center and energy infrastructure solutions, for $4.75 billion. As Alphabet continues to push the frontiers of all things artificial intelligence (AI), investors may be wondering how the Intersect deal fits into the company's long-term roadmap.
Let's dig into why this deal represents a savvy move by Alphabet, and unpack what it could mean for the broader AI infrastructure movement.
Image source: Getty Images.
What does Intersect do, and why is it important?
When it comes to AI-related capital expenditures (capex), most investors immediately think of graphics processing units (GPUs), server architecture, and other networking gear hosted inside data centers. While these costs are indeed important, the rising expenses of procuring chips are only part of the equation.
NASDAQ: GOOGL
Key Data Points
Intersect develops utility-scale renewable energy solutions for data centers. The company places various sources of power generation, such as wind, solar, and battery, alongside data centers. This co-location strategy allows Intersect to power its infrastructure efficiently, bypassing long wait times associated with new facilities accessing the U.S. electrical grid.
Why did Alphabet acquire Intersect?
One of the biggest drains on gross margins for AI developers is the rising cost of energy. As AI workloads continue to grow, investors can expect training and inference costs to multiply exponentially from their current levels.
Alphabet is trying to hedge against these headwinds by gaining control over its future energy costs. Tucking Intersect into its ecosystem allows Alphabet to access data center capacity more efficiently than relying on external utility providers. This will become increasingly important as the company brings new products and services online to its Gemini model and Google Cloud Platform (GCP).
Investors should not be surprised by Alphabet's decision to acquire Intersect. The company already employs a vertically integrated business model across its search, cloud, advertising, and consumer electronics businesses.
Whether it's furthering research in quantum computing through DeepMind, accelerating the production of its custom Tensor Processing Unit chips, or refining the algorithms powering search across Google and YouTube, Alphabet's core assets are stitched together by the same thread: artificial intelligence.
Intersect represents another layer of Alphabet's vertically integrated model, providing the company with greater cost visibility as well as critical flexibility around the future of its infrastructure buildouts.
What does the Intersect deal signal about AI infrastructure?
For the last three years, chip companies such as Nvidia, Advanced Micro Devices, and Broadcom have raced to design cutting-edge hardware to serve as the backbone of generative AI development. While demand for next-generation chips isn't waning by any stretch of the imagination, most hyperscalers realize that AI accelerators will eventually become commoditized.
In my eyes, Alphabet's acquisition of Intersect already signals that these dynamics are in motion. Although chip procurement remains a priority, Alphabet is getting a head start in laying the groundwork to lower the overall cost of compute and improve the flow of data within GPU clusters by acquiring Intersect.
This deal suggests that hyperscalers will not be able to afford to rely on incremental improvements from GPU developers and external power providers in the long run. In a broader sense, this move by Alphabet signals that the AI infrastructure era will feature much more than new chips.
Over the next several years, I suspect more players in big tech will follow a similar playbook to that of Alphabet and begin accelerating investment in the energy infrastructure space -- diversifying their AI budgets beyond new chips and interconnect gear.
Ultimately, I believe the biggest winners of the AI revolution will be those who control all aspects of the value chain, from hardware and software design to the costs associated with energy management and infrastructure development.
