After reviewing SSI's confidential research, NVIDIA raised the stakes with its own contribution

Safe Superintelligence, or SSI for short, is finally no longer just a company rumored to be hidden in Silicon Valley.

On July 27, NVIDIA announced a long-term partnership with this laboratory founded by former OpenAI Chief Scientist Ilya Sutskever and has made an investment in it. The most eye-catching aspect of this arrangement is not the amount, but Vera Rubin. NVIDIA will allow SSI to use this next-generation, yet-to-be widely deployed platform, with the company stating that SSI's computing resources will be increased by an order of magnitude.

This company, founded two years ago, has no publicly available models, no products, and no demos. Co-founder and former CEO Daniel Gross left last year. Sutskever, who remained, took over as CEO, continuing to narrow SSI's goal down to achieving safe superintelligence.

This time, for the first time, he gave a sentence that can be seen as a progress bar for outsiders. According to the joint announcement from NVIDIA and SSI, Sutskever said: "We have research that is worthy of scaling up."

It is not a report card for models, nor is it a product release date. It is more like an admission ticket. In the two years that SSI has been operating, the first thing it did was to confirm whether a research path was worthy of large-scale computing power. Now, what NVIDIA is giving it is the machines needed for the next round of training.

Why has capital arrived before a product?

Superficially, SSI's funding story seems somewhat unusual. A lab that has not yet released a public product has had its valuation pushed up first. According to Reuters, SSI's valuation has risen from about $5 billion to around $32 billion in two rounds of public financing.

The gap in the picture explains why NVIDIA's involvement is not surprising. Between the two rounds of public financing, SSI's valuation has increased by about 6.4 times. The funds are not buying verified commercial revenue but rather an option. If Sutskever's identified path can continue to hold up on a larger scale, the early capital and computing power providers standing by will have a seat to move on to the next stage.

There is a discrepancy in the reported amount of NVIDIA's investment. Reuters states that the equity investment is $5 billion, while NVIDIA's official announcement only mentions a "large investment" without providing a specific number or disclosing SSI's current valuation. Therefore, the orange dots in the graphic represent the investment amount reported by the media and do not indicate a new valuation round.

SSI's capital density should not be misinterpreted as indicating that it already has an organization of the same scale. NVIDIA's announcement only states that they have been secretly advancing a new research direction for the past two years. In publicly available information, SSI has not put its models, products, and demos at the forefront. This silence is not a decorative lack of evidence but rather the true subject that this collaboration is meant to explain.

Sutskever: What Does "Worth Scaling Up" Really Mean

In the world of large-scale models, scaling up is not just about buying more GPUs. It means researchers believe that their current approach is worth testing with more data, longer training times, and a larger cluster. This judgment is costly and easily flawed. If the underlying path is not sound, no matter how much compute power is added, it will only amplify errors more quickly.

According to a joint announcement, SSI has been advancing a new research direction over the past two years. Sutskever stated that the team has "research worth scaling up." Nvidia, on the other hand, says it only entered into this collaboration after gaining rare access to SSI's confidential research. Putting these two statements together, you can see the sequence of events: first, the research direction received internal validation, and then the supply of compute power was brought to the forefront.

The "10" in the graphic is not a chip performance score. It represents the computing resources that SSI itself can access, compared to before the collaboration. Think of it as a lab moving from a single workstation to a full factory. The workstation is enough to determine if a path is viable, while the factory is used to answer whether it can be stably replicated on a larger scale.

This also draws a significant boundary. SSI can be understood as having completed the small-scale path validation and is preparing to enter the actual large-scale training phase. However, "worth scaling up" does not mean the model's capabilities have been publicly proven, nor can it be directly inferred as "product release imminent." In the collaboration announcement, SSI is still showcasing research direction to the public, not the finished product.

Danny Gross's departure has made this path more focused. Sutskever confirmed in July 2025 that Gross's last day in office was June 29, 2025. Today, SSI is led by Sutskever and co-founder Daniel Levy. The company's most easily recognizable asset has shifted from a founding team to Sutskever's judgment on the research direction.

Nvidia's Purchase Was Not Just an Order, But an Entry into the Next-Gen Platform

If this event is viewed simply as Nvidia selling GPUs, the most interesting part of the transaction would be missed. Nvidia not only provided equity investment but also secured access to Vera Rubin as part of a long-term collaboration. What Nvidia was aiming for was to embed itself in SSI's technical roadmap before the largest-scale training truly begins.

This preemptive relationship closely aligns with changes in Nvidia's own business. According to Nvidia's FY2026 results, data center business revenue has risen to $193.7 billion. This revenue group has expanded nearly 13 times over four fiscal years, indicating that the compute business is no longer just about a purchase order but about who can secure the default position for the next-gen platform before a research project becomes a large cluster.

There is a subtle time lag here that is easy to overlook. The significance of Vera Rubin to SSI is to provide about 10 times the computing power to an as-yet unpublished study. For NVIDIA, it means the early adoption of a new platform by customers before they even begin scaling up training. What was exchanged between the two parties was not an immediate income but a priority access to future training cycles.

SSI still hasn't given the answer to the outside world. What NVIDIA is betting on is whether the statement "worth scaling up" by Sutzkever can ultimately hold on a larger machine.