SpaceX inks compute deal with Reflection AI, an open source AI lab

June 22, 2026 · DigitalEdu

SpaceX locked in a $150M/month compute deal with open-source lab Reflection AI, granting instant access to Nvidia GB300 chips at the Colossus 2 data center near Memphis. Running July 2026 through 2029, the agreement underscores how frontier AI labs now secure massive infrastructure without building their own.

The Deal Structure and Hardware Specifications

Under the terms announced, Reflection AI will pay $150 million a month beginning July 1, 2026 through 2029 for immediate access to Nvidia's latest GB300 AI chips and supporting hardware across SpaceX's Colossus 2 data center near Memphis, Tennessee. The GB300 series represents Nvidia's most advanced generation of data center GPUs, designed specifically for large-scale AI training and inference workloads. By locking in capacity at Colossus 2, Reflection AI bypasses the extensive waitlists that currently plague the high-end accelerator market, where lead times for equivalent hardware often stretch 12 to 18 months. The four-year commitment provides Reflection AI with predictable, dedicated capacity—a critical advantage for an open-source lab planning long-term research roadmaps and model release cycles.

SpaceX's Pivot to AI Infrastructure Landlord

SpaceX's $150 million monthly compute deal with Reflection AI highlights how rapidly AI infrastructure demand is scaling, with the startup securing immediate access to Nvidia's latest GB300 chips across SpaceX's Colossus 2 data center through 2029. While SpaceX is primarily known for launch services and the Starlink broadband constellation, the company has been quietly building out significant data center capacity to support its own internal AI workloads, including autonomous flight systems and satellite network optimization. The Colossus 2 facility near Memphis represents a strategic expansion of that footprint. By leasing excess capacity to third parties like Reflection AI, SpaceX transforms a capital-intensive cost center into a recurring revenue stream, effectively operating as a specialized AI cloud provider. This model mirrors the early history of Amazon Web Services, where internal infrastructure was externalized to become a dominant business line.

Reflection AI and the Open-Source Compute Challenge

Reflection AI operates as an open-source AI lab, a category that historically struggles to access compute at the scale enjoyed by closed-source counterparts like OpenAI or Anthropic. Closed labs typically raise billions in venture funding backed by exclusive cloud partnerships—Microsoft for OpenAI, Amazon and Google for Anthropic. Open-source labs, by contrast, often rely on distributed compute, grants, or smaller dedicated clusters. This deal marks a departure: a dedicated, enterprise-grade allocation of cutting-edge hardware for an open-source entity. It suggests a maturing funding ecosystem where investors and infrastructure providers are willing to back open development models with the same caliber of resources previously reserved for proprietary efforts. The arrangement could accelerate the release of competitive open-weight models, broadening access to advanced AI capabilities for researchers and commercial developers alike.

Implications for Small Business AI Adoption

This partnership underscores a broader trend: businesses across sectors are investing heavily in specialized AI hardware to power automation initiatives. For business owners, the key takeaway is that AI automation—using software to handle repetitive tasks like lead follow-up, scheduling, and customer support—can free your team to focus on higher-value work, with most DigitalEdu clients seeing measurable results within the first 30 days of deployment. The infrastructure deals happening at the frontier level—GB300 clusters, multi-year contracts, nine-figure monthly spend—eventually trickle down as lower-cost inference capacity, managed services, and accessible model APIs. Small businesses do not need to procure their own accelerators; they benefit from the oversupply and price competition that massive build-outs like Colossus 2 create. As frontier labs consume compute for training, the resulting models and the inference infrastructure built to serve them become the tools that small businesses deploy for chatbots, workflow automation, and data processing.

Market Context and Competitive Dynamics

The agreement arrives amid a broader scramble for AI compute. Major cloud providers—Microsoft Azure, Google Cloud, AWS—are expanding capacity aggressively, while a new tier of "neocloud" specialists like CoreWeave, Lambda, and Crusoe have raised billions to build GPU-centric data centers. SpaceX's entry adds a unique player: a vertically integrated aerospace manufacturer with proprietary power and networking expertise, now monetizing data center assets. The Memphis location is notable; the region has attracted multiple large-scale AI projects due to favorable power availability, industrial zoning, and proximity to major fiber routes. For Reflection AI, the deal de-risks its compute roadmap. For SpaceX, it validates a new revenue vector. For the market, it signals that non-traditional infrastructure owners are becoming credible alternatives to the hyperscalers, potentially increasing negotiating leverage for AI companies of all sizes.

As compute capacity expands and the cost of intelligence declines, the barrier to entry for AI-powered automation continues to drop. Small businesses that adopt these tools early—automating customer support, lead qualification, appointment scheduling, and data entry—position themselves to capture efficiency gains while competitors rely on manual processes. The SpaceX-Reflection AI deal is a leading indicator: the infrastructure foundation for the next wave of accessible AI is being poured right now.

Frequently Asked Questions