Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.
Originally published by @awilkinson on X. Tech Twitter preserves the original source alongside this readable edition.
Before you dive in
• Discover why a former helium company pivoting to AI data centers could be the next 400% asymmetric bet on power infrastructure and hyperscaler demand.
Best for builders who want practical takeaways. 16 min read.
In January 2025, I called IREN a "Picasso at a garage sale.”
An AI data center company with insanely valuable assets trading dirt cheap because public investors had it in the wrong mental box.
Since then, the stock is up over 400%, and, I believe, still has a long run ahead of it.
Today, I think I found another garage sale Picasso…
But this one's covered in dust.
The frame is broken.
And there's a guy outside screaming that it's a fake.
Let me explain.
My friend @KashRamki has spent decades investing billions in the least sexy industry on earth: infrastructure.
Pipelines, power generation, and data centers.
It's a slow-moving, hyper-conservative world, with payback periods measured in decades.
In December 2024, Kash called me breathless.
He'd sold his personal stock portfolio—his entire net worth—and gone all-in on a crypto mining company called IREN.
I was surprised. It wasn’t like him.
“I feel like a toll-booth operator on some random highway who woke up one morning to find it's the only road to the world's biggest gold mine!”
He was excited. But I wasn't so sure.
I wanted nothing to do with what I initially thought was a crypto investment.
This was one of my most conservative investor friends. I worried he was getting swept up in the excitement.
But as he started walking me through the stock he had just put his life savings into, I began nodding along.
IREN was a Bitcoin mining company. Public investors had sorted it into the “bad” box because of the crypto label.
In reality, they owned massive, high-power datacenter sites near renewable energy plants.
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
Newsletter
We doomscroll, you upskill
Get the 10 tweets shaping how builders think today.
Kash saw something obvious that almost everyone else had missed: the same infrastructure that mines Bitcoin can power AI workloads. And hyperscalers like Microsoft were desperate for exactly the kind of shovel-ready, high-power capacity IREN had locked up.
A year and a half later, Kash and I have both made a small fortune on the trade.
I still own $IREN and am bullish.
But he has been hunting for the next one ever since…
Six months ago, he called me excited.
He told me about a company called New Era Energy ($NUAI).
My first reaction wasn't what he was hoping for.
I told him it looked sketchy.
Former helium company. Shaky management history. Short sellers circling. A lawsuit by the state of New Mexico.
But it was Kash, so I listened.
After I heard him out and dug in, I realized we might be staring at another Picasso.
Here’s why…
Let’s zoom out for a minute and talk about the AI power crisis.
There’s something most people don't understand:
The bottleneck for AI isn't chips.
It's electricity.
Microsoft, Google, Amazon, Oracle and Meta are spending obscene amounts on AI infrastructure.
They have the money. They have the chips.
What they don't have is power.
PJM Interconnection—the largest grid operator in America—has warned about looming reliability gaps as load forecasts explode. Goldman Sachs projects data center power demand will increase 165% by 2030.
The grid can't keep up. Permitting for new transmission lines takes years. Some projects get delayed a decade.
This is why hyperscalers are paying almost anything for “speed to power.”
The ability to turn on electricity fast, without waiting for the grid.
As a result, data centers aren't valued like normal businesses anymore.
They're valued by the megawatt.
Construction cost: $10–15 million per MW.
Acquisition multiples: 25–30x cash flow.
The prize: secured, deliverable megawatts on a realistic timeline.
But here's what most people miss.
You can't just build a data center anywhere. You need:
Land — hundreds of acres, ideally thousands
Power — cheap, reliable energy
Permits — environmental approvals, interconnection, local support
Water — for cooling
Fiber — high-speed connectivity
But getting all five in one place?
That takes years. Sometimes a decade.
The companies that already have these assets locked up are sitting on gold mines.
Which brings us to NUAI…
$NUAI—New Era Energy & Digital—was a helium company.
Yes, helium. The stuff in balloons 😂
They were a helium company with legacy New Mexico assets and natural-gas exposure.
Then they had a very big idea.
What if they stopped selling helium and started selling power to AI companies instead?
Here is what they have assembled since:
Texas Critical Data Centers (TCDC)
438 owned acres in Ector County, Texas
A further 54-acre corridor under definitive purchase agreement
More than 1 gigawatt of planned long-term capacity
A phased buildout beginning with roughly 200MW
100% ownership after acquiring the remaining interest for $70 million
Company guidance targets initial power delivery as early as the end of 2027
New Mexico AI Hub
An option to acquire roughly 3,500 acres in Lea County, New Mexico
A very early-stage concept targeting more than 7 gigawatts
Not yet owned, financed, permitted, contracted or under construction
The company’s own presentation puts the combined long-term vision at more than 8 gigawatts.
For context, IREN controls roughly 3GW and has recently been valued at around $12 billion.
NUAI trades for a few hundred million.
Now, obviously, this is not apples to apples.
IREN has operating infrastructure, secured power and real cash flow. NUAI has a Texas development site, a 200MW first phase and an option attached to a very early-stage New Mexico concept.
But that is why it is cheap.
IREN shows what validated megawatts can become. NUAI is a bet that it can get there.
The Texas site is the thesis. New Mexico is a giant, extremely speculative call option. If Texas works, the market may eventually have to decide what that additional 7GW is worth.
Ok, great. But how are they going to pay for all this?
NUAI probably won’t write one enormous cheque itself.
If this works, each campus will likely be financed as its own project, with a tenant, outside investors and lenders funding most of the construction.
That is the standard infrastructure playbook.
Here is the structure I am assuming:
The project-level company owns the land, power plant, contracts and data center. Outside investors and lenders provide most of the billions required to build it.
NUAI acts more like the developer, sponsor and GP.
In my model, NUAI contributes only around 1–3% of the project equity, earns recurring development or management fees, and receives roughly 10–15% carried interest above an agreed return hurdle.
That would let NUAI participate in enormous projects without funding the entire build from its own balance sheet.
To be clear: those are my assumptions, not signed terms.
NUAI has not disclosed the final fee, carry or ownership economics. Its eventual deal with Primary Digital, Stream or another capital partner could be better—or much worse.
The entire thesis depends on NUAI retaining enough of the upside to make the public-company shareholders rich.
NUAI has raised capital, including roughly $115M in its April equity offering, but shareholders have also been materially diluted. Macquarie announced a facility of up to $290M, of which only the first $20M tranche is committed and funded.
So the real question is brutally simple:
Can they sign a bankable tenant, close the financing and get Phase 1 turned on before the money—or patience—runs out?
Here’s another interesting detail about NUAI…
The Permian Basin accounted for 23% of all U.S. marketed natural-gas production in 2025.
It is one of the most energy-rich regions on Earth.
And because so much of its gas is produced alongside oil, pipeline constraints have periodically pushed local Waha gas prices below zero. At times, producers have literally paid someone to take the gas away.
NUAI is sitting on top of this.
Its plan is to build behind-the-meter generation: produce electricity on-site with local natural gas rather than wait years for new grid capacity.
No new long-distance transmission line.
Less dependence on an overloaded interconnection queue.
A direct path from stranded gas to AI compute.
That is the game now.
Speed to power wins.
Aside from not having to draw electricity from the grid, this also helps them in another way:
Data centers are becoming politically toxic.
Voters don't want to subsidize Big Tech's AI ambitions through higher electricity bills.
The rhetoric is shifting from “we should build more AI” to “fine—but don't make regular people pay for it.”
This is why bringing your own power matters politically.
They’re not competing with grandma’s air conditioning.
They’re not clogging an already hopeless interconnection queue.
And they’re a much harder scapegoat for rising electricity bills.
They can credibly say: “We bring our own power.”
That is a much better conversation to have with voters and regulators.
Let's talk about the land for a second.
Here is some dirty napkin math.
Powered data-center land can trade for roughly $300,000–$500,000 per acre.
NUAI owns 438 acres in Texas.
If—and this is a huge if—it turns those acres into genuine powered land, that implies roughly $131–$219 million of value.
Close the additional 54-acre corridor and the arithmetic becomes roughly $148–$246 million.
Then there is the option on 3,500 acres in New Mexico.
If NUAI acquires that land and somehow turns the entire combined footprint into powered data-center campuses, the same crude arithmetic gets you to roughly $1.2–$2.0 billion.
Let me be crystal clear:
NUAI does not own 4,000 acres of powered land today.
The New Mexico acreage is under option. The power is not delivered. The tenants are not signed. The financing is not complete.
So this is not current liquidation value.
It is a rough map of what successful execution could create.
The Bet
The realistic way to think about NUAI is not “value it like a REIT.”
It is a development-stage sponsor whose value changes violently as each milestone gets checked off.
Phase 1 is currently planned at roughly 200MW. A binding lease and fully funded build would change the whole conversation.
Of course, that is also the catch.
The obvious time to buy is after the tenant and financing are signed.
The potentially lucrative time is before—when it still looks ridiculous.
That is why I think this is asymmetric.
More on the upside below. But first…
Let’s invert. How can this go wrong?
This is not a sure thing. But in my opinion, it’s an asymmetric bet.
As @mohnishpabrai famously says: “Heads, I win; tails, I don’t lose much.”
Let’s walk through some of the risks, and argue for and against.
“The management team is sketchy.”
Argument: Will Gray, the former CEO, has an uneven past in oil & gas. Short sellers have published critical pieces.
Counter: The company has materially changed its leadership. Will Gray is no longer CEO or a director. Charlie Nelson is now Chairman and CEO; Ted Warner is President and CFO; José Rodriguez is COO. NUAI has also added executives with backgrounds at EdgeConneX, AWS, CoreWeave, and Switch.
That improves the execution case, but the real test is whether this team can convert land and a development plan into a signed lease and delivered power.
“They're unproven.”
Argument: They've never built a data center. They're a helium company pivoting to AI infrastructure.
Counter: They're not trying to build it alone anymore. That's the point of the Primary Digital partnership.
“Where will they get the money?”
Argument: They need far more capital than a small public company can provide on its own, and shareholders have already experienced material dilution.
Counter: NUAI raised roughly $115M gross in April and has an initial $20M committed Macquarie tranche. The rest of the project capital must still come from a tenant-backed, definitive project-financing structure. The financing risk is real; it has not been solved.
Execution risk on power generation.
Argument: Gas turbine lead times are getting crazy—some projects are being pushed to late decade.
Counter argument: NUAI's plan leans on reciprocating engines (modular RICE generation), which have more available supply and deploy faster. Timeline risk is partially mitigated if the build is modular.
“They’re being sued by the State of New Mexico.”
This was the scariest headline.
And, in retrospect, probably the least important risk.
The stock dropped 41% after New Mexico sued NUAI and former CEO Will Gray over legacy oil-and-gas entities and environmental cleanup obligations. It sounded existential.
It wasn’t.
In July, a bankruptcy court approved a $1 million settlement. All five claims against NUAI are being dismissed with prejudice. The company admitted no wrongdoing.
To put that in context: the market briefly treated this like a threat to a multi-gigawatt data-center platform.
The actual company-level settlement was $1 million.
That’s noise.
Not nothing. Noise.
Three claims remain against Gray personally. And a separate securities class action is pending. If its allegations about permitting are ever proven, that would matter because permits sit at the center of the thesis.
But the original bear case—that the New Mexico lawsuit could swallow NUAI or kill the data-center pivot—didn’t happen.
The lawsuit was noise.
The risks that matter now are much simpler:
Can NUAI sign a real tenant, secure project financing, get power built and turn planned megawatts into energized ones?
That is what will make or break this investment.
I could be wrong here. This could fail due to permits, financing, litigation, counterparties, equipment lead times, commodity prices, dilution, or execution.
Ok, so those are the big risks.
But there is one thing that made me take the project much more seriously: Bill Stein and Primary Digital.
Bill co-founded Digital Realty Trust and served as its CEO from 2014 to 2022. Under his leadership, it became one of the largest data-center companies in the world.
In January, NUAI announced that Primary Digital—founded by Stein, David Ferdman, Peter Hopper and John Sheputis—would co-develop TCDC.
These are people who know how to finance campuses, structure leases, land serious tenants and turn plans into energized megawatts.
Experienced operators putting their names and time behind TCDC matters—but it is not a guarantee.
NUAI later announced a non-binding LOI with Stream Data Centers and an institutional-capital partner. That is encouraging, but it is not a signed hyperscaler lease, a definitive JV or fully committed construction financing.
The catalyst is brutally clear:
A binding anchor tenant, with real financing and a credible construction plan behind it.
So why aren’t institutions piling in yet?
Because today this is almost impossible for a large fund to own in meaningful size.
The stock is small and illiquid. The diligence file begins with “helium company pivots to AI,” then adds management baggage, litigation and no signed tenant.
If a portfolio manager is wrong, they look like an idiot who chased a meme.
If they are right, the position may be too small to move the fund.
That is where weird bargains hide.
Too strange for the big funds.
Too small to matter if it works.
Career-ending if it blows up.
A binding hyperscaler tenant changes that overnight.
Suddenly, it stops looking like a meme and starts looking investable.
Are you too late?
The stock currently trades around $4–$4.50. With roughly 101 million basic shares outstanding, that implies an equity value of roughly $400–$460 million. Fully diluted, it is higher.
I know.
Putting exact price targets on a pre-lease, pre-construction project is borderline fan fiction.
But you need a map, so here is mine:
Bear case: $0–$1.50 per share.
No tenant. More dilution. The data-center thesis dies and the stock gets valued like a failed development project.
First real validation: $30–$50 per share.
NUAI signs a binding anchor tenant for Phase 1, closes credible financing and proves the 200MW build can actually happen. At today’s basic share count, that is roughly a $3–$5 billion equity value.
Base case: $60–$90 per share.
Phase 1 gets built, the path toward the full Texas campus becomes bankable, and the market starts valuing NUAI as a legitimate data-center developer with fee-and-carry economics. That is roughly a $6–$9 billion equity value before further dilution.
Bull case: $150+ per share.
Texas works, the sponsor model proves repeatable, and the market begins putting real value on the New Mexico option and whatever campuses come next. At today’s basic share count, that is a $15 billion-plus outcome.
These are not analyst targets.
They are my rough probability map, based on today’s basic share count and the assumption that NUAI does not vaporize shareholders with another giant raise.
More dilution means lower per-share outcomes.
And every one of these scenarios depends on the same sequence:
First you get the anchor.
Then you get the capital.
Then you get energized megawatts.
Then you get the multiple.
If Primary Digital steps away, the tenant never materializes or financing keeps slipping without a credible explanation, the thesis is broken.
If that risk makes your stomach turn, fair.
This one isn’t for you.
I hate paying for ideas.
I’ve historically been most successful as an investor when I simply buy something that is already up and running and hold it.
But every once in a while, there’s an asymmetric bet that is so attractive it’s worth making.
NUAI, like IREN, feels like one of a small number of toll roads to AI superintelligence—a road that every one of the world’s largest technology companies needs access to at the same time.
IREN looked sketchy when I bought it. “Bitcoin miner pivoting to AI” sounded like a desperate rebrand.
NUAI looks even sketchier. “Helium company pivoting to AI data centers” sounds insane.
But the assets are real.
The Texas land is owned. The TCDC buyout closed. Primary Digital is involved. Serious people are working on the tenant, power, permits and financing.
But the pieces that turn a promising site into an actual data center remain unproven: the binding tenant, final partner structure, fully committed construction financing, permits and delivery timeline.
Hyperscaler demand is undeniable.
And the market is still pricing this like a potential zero. Every development-stage project can be a dud, but my money is on this being a Picasso.
Sometimes the Picasso at the garage sale has a broken frame and is covered in dust.
But it’s still a bloody Picasso!
That's why I’m buying it, fixing the frame, getting it cleaned and taking it to the appraiser.
—
Important disclosure: I (and entities I control) beneficially own shares of IREN and NUAI. I wrote this after establishing these positions. I may buy more or sell at any time without updating this post. This is not investment advice or a solicitation to buy/sell securities. This is a high-risk, speculative situation and you can lose all your investment. Forward-looking statements and scenarios in this post are speculative and may not occur. Do your own research and read company filings.
