Key Notes:
- The partnership aims to expand its investment to $100 billion, including debt financing.
- Investments will primarily focus on U.S. data centers and energy infrastructure, with some funds allocated to U.S. partner countries.
- NVIDIA will provide expertise in AI data centers and factories.
BlackRock, Microsoft, Global Infrastructure Partners (GIP), and Abu Dhabi-based investor MGX have teamed up to launch the Global AI Infrastructure Investment Partnership (GAIIP). This new initiative aims to raise $30 billion in private equity to invest in expanding data centers and building new energy infrastructure to meet the growing demand for AI computing power, Microsoft said.
See more – Microsoft Copilot Gets an Upgrade with AI: Faster and Better for Businesses
The fund is designed to support the rapid expansion of AI technologies by enhancing the infrastructure needed to power them. Most of these investments will be concentrated in the United States, with additional funding directed to U.S. partner countries.
According to the Financial Times, the group aims to raise $30 billion of initial capital and eventually raise $100 billion, including from debt financing.
“We are committed to ensuring AI helps advance innovation and drives growth across every sector of the economy,” said Microsoft CEO Satya Nadella, in a statement. He said the initiative brings “together financial and industry leaders to build the infrastructure of the future and power it in a sustainable way.”
Is This Partnership a Solution to AI’s Challenges?
As the demand for AI grows, many tech companies ramp up their investments in data centers equipped with NVIDIA graphics processing units (GPUs).
These GPUs are crucial for running generative AI models, such as OpenAI’s ChatGPT. The rising demand for these powerful components is straining energy resources and creating bottlenecks in establishing new facilities.
The Rising Energy Demands
Tackling energy consumption in AI is becoming a significant challenge for the tech industry. A Goldman Sachs report from earlier this year noted that data centers currently use 1-2% of global energy, which could climb to 3-4% by 2029.
This rise is primarily attributed to the increasing demand for AI services requiring significant power. The report highlighted that a ChatGPT query consumes nearly ten times more electricity than a typical Google search.
Governments are also taking notice. Last week, Data Center Knowledge reported that the U.S. Department of Energy (DOE) has partnered with industry experts to address the growing energy needs of AI and digital infrastructure.
The DOE’s recent report, “Recommendations on Powering Artificial Intelligence and Data Center Infrastructure”, emphasizes the significant power demands posed by AI-driven data centers.
Hence, artificial intelligence (AI) models, especially those for deep learning and large-scale data processing, require a lot of computing power, increasing energy usage. To meet these growing computing needs, tech companies connect thousands of specialized chips in clusters, increasing the demand for high-performance data centers.
In response, BlackRock and Microsoft announced the Global AI Infrastructure Investment Partnership. The initiative aims to strengthen AI supply chains and improve energy sources to support the development of advanced AI technologies.
Researchers predict that by 2030, data centers could boost U.S. power demand by 8%.
How Are Data Centers Adapting to AI Needs?
The demand for data center capacity is soaring as advancements in AI and digital transformation continue to drive the need for greater computational power.
A report noted that Major tech companies like Amazon, Google, Microsoft, and Meta are looking for data center operations. They are building new facilities and leasing additional space to meet anticipated future needs, focusing on increasing physical space and power supply.
In AI, there are two main operations to consider: ‘training’ and ‘inference.’ Training involves building and refining AI models over time, while inference refers to using these pre-existing models to generate responses.
While inference requires significantly fewer computing resources than training, it is expanding rapidly. As AI-powered applications become more widely adopted, the volume of inference requests to data centers is expected to grow substantially.
Key Players in AI and Energy Solutions
Microsoft is leading this push, significantly boosting its infrastructure to support its Azure public cloud services, which include AI clients like OpenAI.
In July, the company reported a substantial capital expenditure of $19 billion for its fiscal fourth quarter. This figure encompasses investments in assets acquired under finance leases, highlighting Microsoft’s commitment to expanding its AI infrastructure.
In a strategic move to enhance infrastructure capabilities, BlackRock announced plans to acquire Global Infrastructure Partners (GIP) for $3 billion in cash and approximately 12 million shares of BlackRock common stock in January. BlackRock recently confirmed that the acquisition is expected to be finalized on October 1.
MGX, a new AI and energy sector player, was established in March with Abu Dhabi’s Mubadala and AI firm G42 as founding partners. This initiative aims to drive advancements in AI infrastructure and energy solutions globally.
Future Directions
This partnership reflects a broader trend where private capital is joining forces with corporations and governments to tackle infrastructure needs.
BlackRock, known for its extensive corporate network and long-term investment strategy, and Global Infrastructure Partners (GIP), with its expertise in managing large-scale infrastructure assets, are at the forefront of this movement.
Together, they are poised to lead significant investments in data centers and related infrastructure, driving their clients’ economic growth, job creation, and long-term benefits.
Co-founder and Chief growth officer