Meta to Become the Biggest Nuclear Buyer Among Hyperscalers | Bloomberg Tech 1/9/2026

Bloomberg Tech - January 19, 2024 Summary

Key Concepts:

• AI Energy Demand: The rapidly increasing energy requirements of AI development and deployment, particularly for data centers.
• Nuclear Power Investment: Meta’s significant investment in nuclear energy to secure a low-carbon energy supply.
• Data Center Bottlenecks: The critical limitations in power supply and physical infrastructure hindering data center expansion.
• China’s AI “Tigers”: The emergence of competitive Chinese generative AI startups like Minimax, focusing on cost efficiency.
• Observability Platforms: The importance of monitoring and understanding the performance of complex AI systems and data infrastructure.
• Semiconductor Industry & Government Support: The U.S. government’s investment in Intel and the broader implications for domestic chip manufacturing.
• Streaming M&A: The ongoing negotiations between Paramount, Warner Bros. Discovery, and Netflix, and the valuation challenges.

1. Meta’s Nuclear Energy Deal & AI Power Needs

Meta has entered into a series of agreements totaling 6.6 gigawatts of nuclear energy capacity, making it one of the largest buyers of nuclear power. This investment is driven by the “insatiable demand for energy” required to power its data centers, primarily for AI training and inference (Prometheus and Hyperion data center clusters). Meta is proactively securing its energy supply, even utilizing natural gas plants alongside nuclear, recognizing the risk of existing nuclear infrastructure shutting down. This move is notable as Meta isn’t a traditional “hyperscaler” (cloud computing company) but is independently addressing its energy needs. As Jenson Huang (NVIDIA) stated, “For new energy to emerge, you need energy. It is safe to say we wish we had more energy in the United States.”

2. Data Center Power Supply as a Critical Bottleneck

Paul Meeks (Bloomberg Intelligence) emphasized that power supply is the most critical bottleneck for the AI sector, even more so than demand. While nuclear capacity expansion is positive, it won’t be realized for 4-6 years (2030-2032 at the earliest), necessitating reliance on immediate solutions like natural gas. He highlighted the need to cover the entire supply chain, including utilities, as traditional tech sector analysts are now forced to do. Beyond power, physical constraints like concrete pouring for data center construction are also emerging as bottlenecks. There is currently no adequate federal regulatory framework to support the energy requirements of the AI industry.

3. China’s Generative AI Landscape: Minimax IPO

Minimax, one of China’s largest generative AI startups (backed by Alibaba and Abu Dhabi), successfully went public in Hong Kong, raising approximately $19 million and closing up 109% on its debut. The company focuses on “capital efficiency” and “cost efficiency,” having spent only around $100 million (1-2% of competitors’ spending) on development. Minimax aims to compete with OpenAI by focusing on talent and building a robust product experience. They are also navigating the challenges of accessing advanced accelerators, potentially benefiting from potential approval of NVIDIA H200 chips in China. The cofounder emphasized a focus on ROI when selecting chips, prioritizing performance and accessibility. They reported a gross profit margin of over 55% on their API. Peter Elstrom (Bloomberg Executive Tech Editor) described these Chinese AI companies as “AI Tigers,” offering innovative, low-cost models and aggressive corporate implementation.

4. Snowflake Acquires Observe for Observability

Snowflake is acquiring Observe, an observability platform, for a reported $1 billion. Observability is crucial for monitoring and ensuring the proper functioning of complex systems, particularly AI agents. The acquisition leverages Observe’s existing integration with Snowflake, offering customers faster problem identification and reduced costs. Snowflake CEO Frank Slootman highlighted the importance of covering the entire data lifecycle, from data birth to action, and sees observability as a key component of its platform. He emphasized the need for both internal development and strategic acquisitions to address evolving customer needs.

5. Media M&A: Paramount, Warner Bros. Discovery & Netflix

The potential merger between Paramount and Warner Bros. Discovery, with Netflix as a potential buyer for the streaming assets, remains complex. Warner Bros. Discovery rejected Paramount’s $30/share offer, citing concerns about debt, restrictive covenants, and the valuation of the legacy cable networks (which Paramount values at zero). Matthew Bloxham (Bloomberg Intelligence) believes Paramount needs the assets more than Netflix, while Netflix’s incremental gain is less significant. The January 1st deadline for Paramount’s offer is a key date, and further developments are expected. Tim Cook’s compensation at Apple was held steady at $74 million, reflecting a response to previous criticism of high executive pay.

6. Intel & Government Support

Intel CEO Pat Gelsinger met with President Trump and the Commerce Secretary to deliver a progress report on the company’s chip manufacturing initiatives. President Trump praised Intel’s leadership and the government’s investment in the chipmaker. The U.S. government currently owns approximately $11 billion worth of Intel stock, roughly double its initial investment, but not reaching the “tens of billions” claimed by the President. The government’s continued ownership is contingent on Intel achieving its stated goals.

7. X AI & Deepfake Concerns

Bloomberg reporting revealed that X AI’s Grok image generation tool produced thousands of explicit images of women and children in a 24-hour period, far exceeding the output of dedicated deepfake websites. The tool has since had limited access. Elon Musk responded by stating that users generating illegal content would face consequences. The issue highlights the challenges of content moderation and the potential for misuse of AI-powered image generation.

Notable Quotes:

• Jenson Huang (NVIDIA): “For new energy to emerge, you need energy. It is safe to say we wish we had more energy in the United States.”
• Meta Head of Global Energy: “They started two years ago and were looking at what nuclear needed to keep going and they heard we need investment now into plants that could shutter. This is future proofing.”
• Minimax Co-founder: “It is more about all of the top talents across hours of work that bring the technology to the end users.”
• Paul Meeks (Bloomberg Intelligence): “Of all the bottlenecks it is the most important [power supply].”

Data & Statistics:

• Meta’s Nuclear Deal: 6.6 gigawatts of nuclear energy capacity.
• Minimax IPO: Raised $19 million, closed up 109% on its debut.
• Minimax Spending: Approximately $100 million spent on development (1-2% of competitors).
• Minimax API Gross Profit Margin: Over 55%.
• X AI’s Losses: Spent $9 billion in the first half of the year, revenue doubled to $107 million.
• Grok Deepfake Output: Approximately 6,000-7,000 sexualized images per hour.
• Snowflake Acquisition: Acquiring Observe for approximately $1 billion.
• Intel Government Ownership: $11 billion worth of Intel stock.
• Snowflake Stock Performance: Rose 42% in the past year.
• TSMC Revenue: 20% rise in December revenues.

Conclusion:

The Bloomberg Tech broadcast highlighted the critical intersection of AI development and infrastructure challenges, particularly energy supply. Meta’s proactive investment in nuclear power signals a growing awareness of these limitations. The emergence of competitive Chinese AI companies like Minimax, coupled with the ongoing consolidation in the media landscape, underscores the dynamic nature of the tech industry. Government support for semiconductor manufacturing, exemplified by the investment in Intel, remains a key factor in shaping the future of the industry. The need for robust observability platforms and responsible AI development, as demonstrated by the Grok deepfake controversy, are also paramount concerns.