Nvidia Leads $2 Billion AI Infrastructure Push as Hyperscalers Plan $700 Billion Capex Surge
Nvidia invested $2 billion in AI cloud firm Nebius and an additional $2 billion in photonic companies as hyperscalers prepare to spend approximately $700 billion on AI infrastructure in 2026. The investments come amid surging demand for AI computing resources and the unexpected cancellation of Oracle and OpenAI's Texas Stargate expansion project.
Key Points
- Nvidia invested $2 billion in AI cloud firm Nebius to expand high-performance computing access [4]
- Five largest hyperscalers plan approximately $700 billion total on AI infrastructure in 2026 [1][2]
- Oracle and OpenAI canceled the expansion of their Texas Stargate data center project [3][4]
- CoreWeave secured $11.9 billion contract with OpenAI and $6.3 billion initial order with Nvidia [3]
- Google announced $40 billion investment for three new Texas data centers through 2027 [3]
The AI infrastructure landscape witnessed significant shifts this week as Nvidia announced multiple billion-dollar investments while Oracle and OpenAI unexpectedly canceled their Stargate expansion plans [3][4]. With hyperscalers planning to invest approximately $700 billion in AI infrastructure in 2026, the race to build computing capacity for artificial intelligence workloads continues to intensify [1][2].
Nvidia's Strategic Infrastructure Investments
Nvidia announced a $2 billion investment in AI cloud firm Nebius around March 13, 2026, aimed at expanding high-performance computing access for AI training and deployment [4]. This move directly addresses the surging global demand for AI chips that continues to outpace supply [1][4]. The investment comes just days before Nvidia's GTC 2026 conference (March 16-19), where CEO Jensen Huang is expected to outline comprehensive strategies across energy, chips, infrastructure, models, and applications [1][4]. Additionally, Nvidia committed another $2 billion to photonic firms Lumentum and Coherent to bolster U.S. R&D and manufacturing capabilities for AI infrastructure [3].
Hyperscaler Capex Plans Reach Historic Levels
The five largest hyperscalers are planning unprecedented capital expenditures totaling approximately $700 billion for AI infrastructure in 2026 [1][2]. Amazon leads with projected spending of $200 billion, up from $131 billion in 2025 [2]. Google follows with $175-185 billion, representing a near-doubling from its $91 billion spend in 2025 [2]. Meta plans to invest $115-135 billion in 2026, up from $71 billion in 2025, and has announced an additional $600 billion for U.S. infrastructure through 2028, alongside a $10 billion Google Cloud deal [2]. These investments reflect the exponential growth in AI workload demands and the critical need for expanded computing infrastructure.
Major Deals and Partnerships Reshape Market
CoreWeave emerged as a significant player with two major announcements: a five-year $11.9 billion contract with OpenAI and a $6.3 billion initial order with Nvidia, which includes guarantees for Nvidia to purchase unsold cloud capacity [3]. Microsoft committed up to $5 billion and Nvidia up to $10 billion in Anthropic, with Anthropic pledging $30 billion to Microsoft cloud workloads in return [3]. Amazon is reportedly considering approximately $10 billion investment in OpenAI, though talks remain fluid [3]. These partnerships underscore the intertwined nature of AI model developers and infrastructure providers.
Stargate Cancellation and Regional Expansions
Oracle and OpenAI canceled the expansion of their Texas Stargate data center project, a joint venture with SoftBank that was initially announced in January with up to $500 billion allocated for AI infrastructure [3][4]. No specific power capacity or updated investment figures were provided post-cancellation [3][4]. Meanwhile, Google announced $40 billion for three new Texas data centers through 2027, including facilities in Armstrong County in the Panhandle and two in Haskell County near Abilene, plus expansions in Midlothian campus and the Dallas region [3]. These developments are part of Google's broader strategy encompassing 42 global cloud regions [3].
Energy Solutions for AI Infrastructure
The massive power requirements of AI infrastructure are driving innovative energy solutions across the industry. Microsoft funded a $16 billion restart of the Three Mile Island nuclear facility [5]. Amazon is deploying Small Modular Reactors (SMRs) to address power needs [5]. Google has partnered with Kairos Power on molten salt reactor technology to help alleviate grid strain [5]. These nuclear and alternative energy investments highlight the industry's recognition that traditional power sources cannot meet the exponential growth in AI computing demands.
Market Consolidation Amid Explosive Growth
The cancellation of the Stargate expansion alongside Nvidia's aggressive investments signals a potential consolidation in the AI infrastructure market. With Morgan Stanley estimating approximately $2.9 trillion in global data center construction through 2028 and roughly $3 trillion total AI infrastructure investment by then [6], the market is experiencing both rapid growth and strategic realignment. Oracle's positioning as an AI infrastructure leader [7] may be tested as competitors like CoreWeave secure major contracts and Nvidia deepens its infrastructure partnerships. The emphasis on nuclear and alternative energy solutions indicates that power availability, not just capital, will be the critical constraint for AI infrastructure expansion.
Strategic Realignment Ahead
As Nvidia's GTC 2026 conference approaches, the industry awaits strategic direction on managing the supply-demand imbalance in AI computing resources [1][4]. The hyperscaler capex commitments totaling $700 billion for 2026 suggest continued aggressive expansion [1][2], though the Stargate cancellation hints at potential project reassessments. With energy infrastructure investments in nuclear and SMR technologies [5], the industry is laying groundwork for sustainable long-term growth beyond traditional data center models.