Nvidia Invests $3 Billion in Lancium to Secure Power for AI Datacenters in Texas
The AI compute build-out accelerates, with power access emerging as a critical bottleneck driving new infrastructure investments and revised market forecasts.
The story
The intense demand for AI compute capacity is increasingly shifting focus to underlying power infrastructure, as evidenced by Nvidia's recent $3 billion investment in Lancium. Nvidia agreed to invest $2 billion initially, with an additional $1 billion committed as Lancium secures more power capacity.
This deal gives Nvidia a roughly 20% stake in Lancium, valuing the power infrastructure developer at approximately $10 billion, and directly addresses the energy needs of the OpenAI and Oracle AI campus in Texas. This move highlights how access to power is now a primary constraint governing AI infrastructure expansion, prompting GPU suppliers to invest upstream in land acquisition and grid interconnection.
The broader market reflects this urgency, with Dell'Oro Group raising its forecast for the global data center IT semiconductors and components market to $1.8 trillion by 2030, citing stronger expectations for capital expenditure and power capacity additions driven by AI. TSMC also lifted its 2026 revenue forecast by over 40% and increased capital expenditure guidance to between $60 billion and $64 billion, propelled by relentless demand for advanced AI chips.
Silicon
9th-generation 2Tb QLC 3D flash memory
Maker: Kioxia and Sandisk
What: Offers improved read/write bandwidth, higher power efficiency, and a 4.8Gb/s NAND interface for data-intensive infrastructure.
For Whom: AI-driven and data-intensive infrastructure requiring growing storage capacity.
Agilex FPGAs (DDR5, LPDDR5, LPDDR5X support)
Maker: Altera
What: Expanded memory support delivering up to 204.8 GB/s of aggregate memory bandwidth.
For Whom: AI, networking, and embedded workloads requiring high-bandwidth memory.
The build-out
| Project | Who | Scale | Where |
|---|---|---|---|
| Power infrastructure for OpenAI and Oracle AI campus | Nvidia (investing in Lancium) | $3 billion investment | Texas, United States |
Supply & policy signals
TSMC raised 2026 revenue forecast to over 40% growth and capital expenditure guidance to $60-64 billion.
Implication: Indicates surging demand for advanced AI chips and a significant commitment to expanding manufacturing capacity.
Applied Materials lifted fiscal 2026 revenue outlook to approximately $10.3 billion.
Implication: Reflects strong demand for chipmaking equipment driven by AI infrastructure, compensating for China export curbs.
Dell'Oro Group raised global data center IT semiconductor market forecast to $1.8 trillion by 2030.
Implication: Highlights increased expectations for capital expenditure and power capacity additions, with AI as the primary driver.
What we'll be watching
- Microsoft's expected unveiling of its next-generation custom AI processor, the Maia 300, as early as September.
- Customer sampling of Intel's new data center GPU, code-named Crescent Island, optimized for inference workloads, expected in the second half of 2026.
- Availability of Nvidia's MGX-compatible 800 VDC power rack for hybrid data center environments, slated for the second half of 2026.
Reporting + analyst voices: grounded via Google Search at publish time.