OpenAI, SoftBank Partner with SB Energy for Stargate (2026)

TL;DR

OpenAI and SoftBank have partnered with SB Energy to supply renewable energy for the Stargate AI infrastructure project
SB Energy will develop dedicated power generation facilities to meet the enormous energy requirements of advanced AI training and inference
The partnership prioritizes clean energy sources including solar and wind to power what’s positioned as the largest AI infrastructure buildout in history
This addresses the critical bottleneck facing AI scaling: securing sufficient power capacity for next-generation data centers

What Happened

OpenAI and SoftBank Group announced a strategic partnership with SB Energy, SoftBank’s renewable energy subsidiary, to power the Stargate project—the $500 billion AI infrastructure initiative unveiled in January 2026. SB Energy will develop and operate dedicated renewable energy facilities to meet the massive electricity demands of Stargate’s planned data center network.

The partnership represents a vertical integration strategy for AI infrastructure, where the companies building the AI systems are now directly securing their own power generation rather than relying solely on existing utility grids. SB Energy brings experience developing over 5 GW of renewable projects across the United States, primarily utility-scale solar and wind installations.

While specific capacity numbers weren’t disclosed, industry analysts estimate that a project of Stargate’s scale could require 5-10 GW of dedicated power capacity—equivalent to the output of several large nuclear plants. The first facilities are expected to come online in phases starting in late 2027.

Why It Matters

Energy availability has become the primary constraint on AI development. Training models like GPT-5 or beyond requires data centers consuming hundreds of megawatts continuously for months. Inference at scale adds even more demand. The industry is already competing for limited power allocations in regions with existing data center hubs.

This partnership signals a shift in how AI companies approach infrastructure. Rather than waiting for utilities to expand capacity or competing for scarce grid connections, OpenAI and SoftBank are building their own power supply chain. This gives them control over timeline, cost, and energy source—critical factors when you’re betting half a trillion dollars on AI infrastructure.

The renewable focus matters for more than PR. Clean energy sources with predictable long-term costs provide hedge against fossil fuel price volatility. Solar and wind also face fewer regulatory hurdles in many jurisdictions compared to natural gas plants, potentially accelerating deployment timelines. However, the intermittency of renewables will likely require substantial battery storage or grid backup agreements.

Key Details

Partnership Structure:

SB Energy designated as primary renewable energy developer for Stargate project
Multi-year agreement covering initial phase of data center buildout
Mix of solar, wind, and energy storage facilities planned
Facilities will be located near planned data center sites to minimize transmission losses

Timeline and Capacity:

First renewable facilities targeted for 2027 operation
Full buildout expected to span 5-7 years
Estimated capacity in multi-gigawatt range (specific numbers undisclosed)
Phased deployment aligned with data center construction schedule

Strategic Implications:

Reduces dependency on existing utility infrastructure
Provides cost predictability through power purchase agreements
Positions Stargate as “carbon-neutral AI infrastructure”
May include co-location of generation and compute facilities

Implications

The Stargate-SB Energy partnership establishes a template that other AI infrastructure projects will likely follow. Microsoft, Google, and Amazon have already been securing dedicated power capacity for their AI data centers—Microsoft through nuclear restarts, Google through geothermal investments, Amazon through small modular reactor development.

This creates a new competitive dimension in AI: not just who builds the best models, but who secures the energy to train them. Companies with strong balance sheets and energy expertise gain advantage over pure-play AI startups that must compete for limited colocation capacity.

The renewable energy focus also pressures the rest of the industry. If Stargate achieves meaningful scale with clean power, it becomes harder for competitors to justify fossil fuel-heavy infrastructure. This could accelerate overall grid decarbonization as AI computing becomes a major electricity customer segment.

Our Take

The smart money is building power plants, not just data centers. This partnership confirms what energy analysts have been warning: AI scaling is now an energy problem more than an algorithmic one. OpenAI and SoftBank recognized this early and acted.

The renewable angle is strategically sound but faces real execution challenges. Solar and wind capacity factors mean you need 3-4x nameplate capacity versus a gas plant to deliver the same 24/7 power. Battery storage at this scale is expensive and unproven. SB Energy will likely need grid interconnection agreements for backup power during low renewable production periods—which brings them back to competing for transmission capacity they’re trying to avoid.

Watch how this partnership handles permitting timelines. Renewable projects face 2-4 year development cycles even in favorable regulatory environments. If Stargate’s data centers are ready before the power supply, they’ll be forced onto fossil fuel grid power temporarily—undercutting the clean energy positioning.

The bigger question: does dedicating this much energy to AI represent optimal resource allocation? A 10 GW AI infrastructure project consumes electricity equivalent to 8 million US homes. The value proposition assumes AI breakthroughs justify that opportunity cost. If AGI remains elusive while power costs soar, this becomes the most expensive infrastructure bet in tech history.

For now, the partnership solves a critical bottleneck. But energy constraints won’t disappear—they’ll just move to the next level of scale.