Starting today, Reflection AI is paying SpaceX $150 million per month for access to Nvidia GB300 chips at Colossus 2 in Memphis, Tennessee. The contract runs through 2029 — $6.3 billion if it runs to term — with a 90-day exit option after the first quarter.
Reflection has not shipped a frontier AI model. Its only public product, Asimov, is a code comprehension agent that has been on a waitlist since July 2025. And yet, as of March 2026, the company is valued at $25 billion, Nvidia has invested $800 million in it, and it just locked up GPU time at the same facility serving Anthropic and Google.
That’s the anomaly worth understanding. Not whether Reflection will win, but what the structure of this bet tells you about how AI infrastructure is being allocated right now.
Who Reflection AI Is
Reflection was founded in March 2024 by two former Google DeepMind researchers:
Ioannis Antonoglou is the company’s CTO. He was a founding engineer at DeepMind, and a core contributor to DQN (the first system to play Atari games at human level), AlphaGo, AlphaZero, and MuZero — the successive systems that defined what reinforcement learning could do at scale. He also contributed to Gemini post-training.
Misha Laskin is the CEO. He led reward model development — the RLHF systems that align large language models to human preferences — on the Gemini project at DeepMind. He did his postdoctoral work at UC Berkeley.
Their thesis: the Western open-source AI ecosystem lacks a frontier-scale model trained with serious reinforcement learning. DeepSeek proved what a well-resourced team could build without closed-lab restrictions. Reflection is betting it can do the same for the US market, with open weights released for enterprise self-hosting while keeping training data and pipelines proprietary.
The company has been deliberate about what it has not released. Asked about timelines in March 2026, the founders said: “You will need to wait for it.” No research papers. No model benchmarks. No training methodology disclosures.
The Deal Structure
The terms signed on June 22, 2026 and effective today:
- $150 million per month, starting July 1, 2026
- Through the end of 2029 (approximately 42 months)
- $6.3 billion total if the contract runs its full term
- Exit clause: either party can terminate with 90 days notice after the initial three months
- Hardware: Nvidia GB300 chips at SpaceX’s Colossus 2 data center in Memphis, Tennessee
For context, SpaceX now has commitments from at least three major AI tenants at Colossus:
| Tenant | Estimated deal size | Monthly rate |
|---|---|---|
| Anthropic | ~$45B | ~$1.25B/month |
| ~$30B | ~$920M/month | |
| Reflection AI | $6.3B | $150M/month |
| Cursor | Not disclosed | Not disclosed |
SpaceX’s AI compute revenue is now estimated at over $80 billion in committed contracts — exceeding its rocket launch business by a significant margin. Colossus has become a commercial compute platform, not just an internal training cluster for xAI.
The Nvidia Angle
Nvidia invested $800 million in Reflection AI in early 2026. It also manufactures the GB300 chips that SpaceX is deploying at Colossus 2 and leasing to Reflection.
This is a structurally interesting position. Nvidia profits from the chip sales to SpaceX, profits from its equity stake in Reflection, and benefits from Reflection’s research validating the GB300 architecture at frontier scale. Whatever happens to Reflection as a company, Nvidia has already captured value on two of the three sides of the trade.
This is not unusual for Nvidia — it has structured similar positions with several frontier labs — but it is worth noting when evaluating Reflection’s valuation. Some of the capital flow feeding that $25 billion number is circulating back through the same ecosystem.
What SpaceX Is Becoming
The Colossus data center started as xAI’s internal training facility for Grok. Within roughly 18 months, it has become the preferred large-scale compute destination for multiple frontier labs that are not xAI.
The pattern suggests something important about what SpaceX offers that hyperscalers don’t. The core advantages are integrated: SpaceX controls energy supply (Starlink power infrastructure), cooling systems, physical security, and network — a vertically integrated cluster that can be scaled without the procurement delays of traditional data center builds. Leasing a “slot” at Colossus is closer to leasing a supercomputer than renting cloud instances.
Some analysts have started using the term “neocloud” for this model: private compute infrastructure operating at cloud-equivalent scale, deployed at sovereign-grade security, accessible via long-term lease rather than metered API calls.
The Anthropic deal was the proof of concept. The Google deal confirmed it wasn’t a one-off. The Reflection deal shows that even startups — very well-funded ones — are choosing dedicated clusters over traditional cloud for frontier training runs.
What Reflection Is Actually Building
The company’s stated product goals:
A frontier open-weight general model: Trained with reinforcement learning at scale, designed around the architecture choices that made DeepSeek’s outputs surprising to Western observers — MoE design, heavy RL post-training, emphasis on reasoning chains. Weights will be released; training data will not.
Asimov: Their only shipped product. An agentic coding system that ingests a company’s codebase, emails, and Slack history to understand architecture context and generate relevant code. As of March 2026, it is in invite-only waitlist status.
Sovereign AI deployments: Enterprise and government customers who need to self-host a frontier model without dependency on US or Chinese closed-lab infrastructure. The South Korea data center deal ($6.8 billion, announced earlier in 2026) sits in this category.
The company has not shared benchmarks comparing its model to GPT or Claude. It has not published papers on training methodology. The $6.3B compute deal is not evidence of a shipped model — it is evidence of intent to train one at frontier scale.
Builder Implications
Three things worth tracking if you are building on or around frontier AI:
1. The open-weight frontier is getting serious compute
Until recently, the gap between open-weight models (LLaMA, DeepSeek) and closed-source frontier models (GPT-5, Claude Opus) was partly a compute gap. DeepSeek narrowed it with efficiency. Reflection is trying to close it with raw scale — the same scale Anthropic and Google are using.
If they ship a competitive open-weight model trained on $6B+ of compute, the calculus for self-hosting changes. A model competitive with Opus or GPT-5 that you can run inside your own infrastructure alters security posture, latency, and cost models for enterprise deployments.
2. Compute access is now a strategic resource, not a commodity
The Reflection deal starting today illustrates a supply constraint that is easy to underestimate. There are a small number of facilities globally with the power, cooling, and interconnect needed to train frontier models. Anthropic, Google, xAI, and now Reflection are locking up long-term capacity at them.
This means that if you are a mid-tier AI company needing GPU time at frontier scale in 2027 or 2028, you may be competing against contracts signed in 2025 and 2026. The capacity is not infinitely elastic.
3. The “no product shipped yet” valuation pattern is not accidental
Reflection is worth $25 billion without a flagship model. This is not irrationality — it is the market pricing the cost and difficulty of training frontier models correctly. The infrastructure required to enter the frontier tier is so large (compute contracts, energy, talent) that investors are buying into the option before the output exists, because waiting until the model ships means waiting until the moat is already built.
For builders, this has a practical implication: if you are waiting for a Reflection-style open-weight frontier model before building on top of open weights, you may be waiting a year or more. The better bet is to design your system to be model-agnostic at the API layer, so you can switch when the open-weight frontier catches up.
Reflection AI ships its first major product today — not a model, but a compute contract. Whether the model follows in 2026 or 2027 will determine whether this was the right call.
Correction policy: If Reflection releases model benchmarks or a public product between publication and your read date, treat the specifics here as a snapshot from July 1, 2026.