AI infrastructure

Nvidia’s CEO Jensen Huang announced at the GTC conference in Washington D.C. that Blackwell GPUs are now in full production in Arizona, marking a shift away from solely Taiwanese manufacturing. This move aligns with demands for domestic chip production, including a request from former President Trump. Nvidia also announced a partnership with Nokia to bolster U.S. 5G/6G infrastructure with Nvidia chips and collaboration to advance U.S. quantum computing capabilities. Export restrictions have impacted the company.

OpenAI urges the U.S. to invest heavily in new energy infrastructure to compete with China in AI. Projects like “Stargate” demand enormous power, straining the U.S. grid. OpenAI advocates adding 100 gigawatts of new energy capacity annually to avoid an “electron gap” and secure leadership in AI. They highlight China’s rapid power expansion as a concern. Increased AI power demands could spur investment in renewable energy and grid infrastructure.

POET Technologies (NASDAQ: POET) secured approximately $150 million in gross proceeds through an oversubscribed registered direct offering of common shares, attracting new institutional investors. Closing is expected around October 28, 2025, pending standard conditions. POET intends to use the net proceeds for acquisitions, R&D, and operational expansion. Pro-forma cash is expected to exceed $300 million post-closing. While the offering provides financial runway, the share issuance may cause dilution.

Anthropic’s plan to deploy up to one million Google Cloud TPUs, valued at tens of billions of dollars, highlights a shift toward diversified AI infrastructure strategies. This expansion, aiming for a gigawatt of capacity by 2026, supports Anthropic’s growing customer base, especially among Fortune 500 companies, signaling a move to production-grade Claude implementations. Anthropic leverages Google’s TPUs, Amazon’s Trainium, and NVIDIA’s GPUs, urging enterprises to avoid infrastructure lock-in and prioritize flexibility for varying AI workloads, while focusing on cost-efficiency and responsible AI deployment considerations.

Applied Digital (APLD) secured a $5 billion, 15-year lease with a U.S. hyperscale data center operator for 200 MW capacity at its Polaris Forge 2 campus, bringing its total leased capacity to 600 MW. Despite the significant deal, APLD’s stock fell over 7%, although it’s still up nearly fourfold year-to-date. The agreement underscores the growing demand for AI infrastructure, with major cloud providers investing heavily. CEO Wes Cummins highlighted Microsoft, Meta, Oracle, Amazon, and Google as potential partners. Applied Digital’s 4 gigawatt active pipeline suggests further growth.

CoreWeave’s CEO downplays the necessity of acquiring Core Scientific amidst shareholder opposition. ISS recommends voting against the deal, citing potential undervaluation. While CoreWeave won’t raise its offer, viewing the acquisition as a “nice to have”, not a “need to have”. A major Core Scientific shareholder also opposes the deal. The vote is scheduled for October 30. CoreWeave’s expansion through AI-related acquisitions aims to diversify service offerings beyond GPU power, capitalizing on the AI boom.

HIVE Digital Technologies is expanding its Paraguay site with a 100 MW hydroelectric-powered Phase 3, bringing the site’s capacity to 400 MW and HIVE’s global renewable energy footprint to 540 MW. The company aims to reach 35 EH/s Bitcoin mining capacity by 2026, with Phase 3 construction starting in early 2026 and commissioning in Q3 2026. HIVE has also partnered with Bell Canada to integrate BUZZ HPC AI infrastructure, projecting a 5x growth in HPC/AI capacity by 2026, driven by GPU clusters and a Toronto buildout.

Credo, a semiconductor company specializing in high-speed connectivity, has seen its stock surge due to increasing demand for its active electrical cables (AECs) in AI infrastructure. Its AECs are crucial for connecting AI servers in data centers, with major clients like Amazon and Microsoft utilizing Credo’s signature purple cables. The company’s revenue doubled in fiscal year 2025, driven by the AI boom and hyperscalers’ data center expansions. While facing competition, Credo is expanding its product portfolio and collaborating with hyperscalers to meet the insatiable demand for AI connectivity solutions.

Oracle’s shares rose after projecting significant revenue growth driven by AI infrastructure. The company forecasts $20 billion in AI-powered database revenue by fiscal year 2030, a substantial increase from previous estimates. Oracle secured $65 billion in new cloud infrastructure commitments this quarter, including a major deal with Meta. The company is expanding its cloud infrastructure, competing with Amazon and Google, and offers its database solutions on rival platforms. Oracle also secured a commitment from OpenAI valued at over $300 billion. They project adjusted earnings per share of $21 on $225 billion in revenue for fiscal 2030.

Nvidia, Microsoft, BlackRock, and xAI are set to acquire Aligned Data Centers for $40 billion, the largest data center deal to date. This acquisition, led by the Artificial Intelligence Infrastructure Partnership (AIP), signals a surge in investment into the physical infrastructure required to support accelerating adoption of AI. Aligned, a design and operator of data centers, currently owned by Macquarie Asset Management, has 50 campuses with over 5 gigawatts of power capacity, making it an ideal initial investment for the AIP. The deal is expected to close late next year.