How We'll Automate Discovery Itself
In the quiet, mist-covered ridges of East Tennessee, armed guards and miles of razor wire protect one of the government's top research facilities.
For more than 80 years, the Oak Ridge National Laboratory has been where the U.S. goes when the stakes are existential.
This is where the Manhattan Project learned how to split the atom.
It's where uranium enrichment moved from theory to industrial reality.
Today, the facility is humming with a "Manhattan Project" of today's age... a complete reset of America's AI industry.
At the heart of this effort sits Oak Ridge's crown jewel, Frontier, the world's most powerful supercomputer.
Frontier can perform more than a quintillion calculations per second – a number so large it borders on meaningless.
To put some context to it... if every human on Earth performed one calculation per second, nonstop, it would take more than four years to match what Frontier does in a single second.
The computer's storage system can hold 33 times the amount of data that's in the entire Library of Congress.
Frontier is a beast of a machine. But what's coming next out of Oak Ridge will dwarf it...
We're calling it the Genesis device.
The plan, set in motion by Executive Order No. 14363, isn't just to build a bigger supercomputer. It's to connect them all.
The U.S. Department of Energy ("DOE") has more than 400 supercomputers at 17 national laboratories across the country. To create the Genesis device, a specialized, ultra-high-speed fiber network will link all of them to Frontier.
From elite laboratories like Lawrence Livermore in California to Argonne in Illinois – Genesis will create a single, unified platform.
It's difficult to think about the potential power of Genesis. But imagine 17 of the smartest people on Earth suddenly being able to share thoughts instantaneously, working in unison.
To be clear, Genesis isn't just trying to create a supercool version of ChatGPT. It's a machine designed to automate discovery itself.
Consider the field of medicine... Traditionally, developing a new drug takes more than a decade and costs billions of dollars. Most drug companies fail. And 90% of drugs don't make it out of clinical trials.
The Genesis device will be able to simulate the interaction of billions of molecules in days. This has the potential to cure diseases that have plagued us for generations.
Along with biotechnology, Genesis will try to innovate several other industries... including nuclear fission and fusion, critical materials, quantum science, advanced manufacturing, and semiconductors.
The stated goals for Genesis are ambitious, to say the least... Take 10 years of AI progress and force it to happen in 10 months, or less.
Now, I've always been something of an AI skeptic.
The Genesis project has the possibility of beating China to artificial general intelligence, or AGI – a superintelligent AI that can function on its own...
It has the possibility of vastly improving our country's industries...
Will America's scientists ever pull that off? We don't know.
But here's what we do know... The White House is not wasting time. Executive Order No. 14363 set a very aggressive timeline, with specific deadlines that started last November. There's a 270-day deadline for initial operating capability.
Plus, the smartest people in the world are involved. These include OpenAI CEO Sam Altman and Nvidia (NVDA) CEO Jensen Huang. Combined with other tech giants and the brightest at the DOE, this is a rare moment when the federal government and private industry are pulling in the same direction.
This project is moving fast. And money is flowing.
The DOE made a $320 million initial investment for Genesis...
Companies involved in AI chips, models, and quantum computing – all the stuff of science fiction – stand to make a lot of money throughout the life of this project. This will be the most sophisticated AI network ever created.
But the biggest hurdle this project faces is something much more basic...
Genesis is creating a level of sustained, always-on computational demand. These machines won't sleep. And they can't afford outages.
A project of this magnitude requires a staggering amount of power. And, in general, generating enough power is the biggest hurdle to reaching an AI future.
Global AI power demand is projected to create severe energy constraints by 2027 or 2028. There simply might not be enough power to supply all of the world's new AI projects...
And that could put serious pressure on both the power grid and many of today's biggest tech stocks.
Last week, two of my colleagues – Marc Chaikin and Jeff Brown – went on camera to talk about this looming energy crisis...
They explained that it could trigger a transition from the first phase of the AI revolution to the second. It all has to do with a new innovation from an AI company that's going public soon.
Marc and Jeff say this is the single biggest opportunity for wealth creation that they've seen in their combined 90 years of investing.
Investors who know what's going on and position themselves strategically could see up to 12 times gains... Meanwhile, those who don't prepare could be stuck with massive losses.
Today is your last chance to hear about this hidden threat in AI and learn how you can take advantage. Click here for all the details.
What We're Reading...
- Something different: There was one way we'd agree to do The Devil Wears Prada 2, says Meryl Streep.
Here's to our health, wealth, and a great retirement,
Dr. David Eifrig and the Health & Wealth Bulletin Research Team
May 6, 2026