Earlier this week, the Trump Administration lifted the curtain on the Genesis Mission, a grandiose project ostentatiously labelled as "Manhattan Project 2.0."

At its core, the initiative aims to create an “integrated platform” aimed at “doubling the productivity and impact of American research and innovation within a decade.” In less grandiose terms, it’s a Department of Energy-led project aimed at creating high-fidelity training data for AI, with some of the biggest names in tech set to transform vast troves of U.S. scientific research data into usable data for these models.

That’s through the American Science Cloud (AmSC), a national cloud platform that provides access to scientific data for U.S. researchers. A lot of that effort will be taking huge swathes of information currently bound to tape systems and digitizing it so it can be used to feed those models.

According to the DOE’s Genesis Mission homepage, a total of 52 organizations are in line to take part in some capacity for the project, including Nvidia, OpenAI, Cisco, HPE, Anthropic, and AMD.

There’s also going to be participation from three of the big four hyperscalers (AWS, Google, and Microsoft - no Oracle at the time of writing).

Nestled among the expected participants is Cornelis Networks, which is set to provide networking technologies to connect data across participating labs.

Cornelis CEO Lisa Spelman sat down with SDxCentral to make sense of the Genesis Mission and how this long-term project aims to accelerate scientific discoveries using AI.

The unsexy truth: Modernizing tape libraries

For all the AI moonshot rhetoric, much of Genesis comes down to a more prosaic challenge: making decades of scientific data actually usable.

“There's so much data that [DOE labs] have generated that is so useful and impactful, and it's hard to use,” Spelman explained. “It's on a tape, but you can almost call it like it's behind a wall, even for the scientists within the DOE.”

In the former Intel executive’s view, the vision of the Genesis Mission goes beyond just upgrading hardware. She noted: “It's one thing to update the infrastructure, but you also have to add in the AI tools and capabilities that make it more accessible.

"It's not just, if you build a better infrastructure, then everything magically becomes easier. You have to do the infrastructure and the user tools so that you get more for it.”

For its anticipated role in the project, Cornelis will provide high-performance Ethernet networking tech to connect data across DOE labs.

“We want to own the role of connecting all the data across all the labs and making it really accessible with modern AI tools,” Spelman said.

Despite only being a five-year-old company, Cornelis has worked extensively with the DOE. Its CN5000 tech, for example, is set to feature in Lynx, an upcoming supercomputing cluster that’ll be used to perform continual assessments and certification on the U.S. nuclear stockpile as part of the Advanced Simulation and Computing Program (ASC).

“We've partnered with DOE since our founding on both research and development programs as well as deployments in multiple of the national labs,” the CEO explained. “Their interest in us is that we provide high-performance networking technology for their workloads, and we do it within an open ecosystem, so they can pick their OEM of choice for processors or GPUs.”

The contract situation

While more than 50 names are attached to the Genesis Mission, the reality is a bit more vague. No firm budgets have been set. Contracts aren’t set. And President Trump’s executive order outlines milestones every 60-180 days, but much of the definitional work is still ahead.

The Cornelis CEO told SDxCentral that, upon unveiling of the project, “the public's getting an early look ... It's not a situation where all these contracts have been set.”

The executive order sets out an ambitious timeline. Within 90 days, the DOE must identify available computing resources. It is unclear at this time whether the undertaking will require building new clusters, though it specifically specifies leveraging “resources available through industry partners.”

Within 120 days, initial datasets must be cataloged, with a plan in place to ingest datasets from federally funded research, as well as other agencies, academic institutions, and approved private-sector partners. Within 270 days, the department should demonstrate an initial operating capability of the platform for at least one “national challenge.”

But, central to the order is the explicit statement that all activities are “subject to available appropriations” – phrasing that appears a total of six times throughout the document.

The order does not, however, explicitly say there will be financial contracts. Given the inherent value of the dataset, there is potential that the AI model developers could get access to access the data for free.

For example, both Anthropic and OpenAI offers ChatGPT access to U.S. federal agencies for just $1. Gaining access to such data could prove a boon for those firms, though no details on data usage have been outlined. SDxCentral has contacted the DOE for comment.

“This is not something where all the work has been done, and then it's being announced as a finalized product,” Spelman explained. “The public and the world are being brought in at the genesis of the definition and the concept, and that is where the reach out happened ahead of the announcement about our willingness to participate."

Spelman described the work of the Genesis Mission as an “industry-plus-government type of effort to define the best solutions for the variety of different use cases and workloads.”

When asked about investment amounts and timelines, Spelman was candid: “As far as I know, they weren't really announcing at the time investment amounts, and I think that will start to come through as the U.S. government goes through appropriations.”

That uncertainty takes on added significance given recent events. The Genesis Mission was announced shortly after the longest government shutdown in U.S. history. But it’s still operating without a full-year, long-term budget plan in place, as the bipartisan agreement was only for a temporary funding measure, or a continuing resolution (CR).

Without a long-term budget, that could potentially impact the ability of participants like Cornelis to match the ambition behind Genesis: to create a unified pool of scientific training data to power advanced AI.

“This is an undertaking and a vision that requires a lot of us in the industry to come together and put our best foot forward,” Spelman added.

Why the Genesis Mission actually matters

For all the questions about budgets and timelines, Spelman argues that the unglamorous work of modernizing scientific infrastructure could have a genuine impact.

“When you go visit the DOE and walk around those buildings, you see how much the science is serving the U.S. and the world,” she said. “I think that sometimes it's a bit of an unfortunate situation that more people don't have access or understanding to what these agencies do that really does benefit the U.S., and how important it's been all over the world for government-plus-industry partnership to solve the grand challenges.”

While the naming convention of the project conveys a grandiose scale, in reality, it’s all about data. Platforms like AmSC will leverage AI models and tools built off the data modernized by Genesis participants. And that’s information that could genuinely accelerate discovery in climate science, energy, materials research, and national security applications.

"The thought of bringing that out and making it more accessible to the scientists of the world is a really exciting part, and we want to role of connecting all the data across all the labs and making it really accessible with modern AI tools,” Spelman said.

Comparing Genesis to the Manhattan Project may be overblown. But the work itself, connecting decades of siloed scientific research, modernizing infrastructure, and making data accessible, isn’t trivial nor unimportant.

Whether it delivers on its lofty promises will depend less on the branding and more on whether Congress appropriates the funds to turn modernization plans into reality.

For now, though, the “Manhattan Project 2.0” remains a vision waiting for a budget.