Anytime there is volatility in my photonics stocks.
I just think back to what Jensen told me.
The "silicon photonics technology capacity that we need is substantially higher than the world has today."
Don't think that's been solved in 3 months...
My take on investing:
Most people watch sports because they love the game.
They love the thrill, the tactics, the drama, the entertainment etc.
The score at the end matters, but it is not the only reason they watch.
Investing is the opposite for me.
I only care about the score.
Not because money is the whole game, but because the score to me is what lets the real game begin.
Living better.
Helping family.
Helping friends.
Backing people.
Building again.
Taking risk from a position of strength.
Not worrying about bills.
Not letting money quietly control every decision in your life.
That is the part I care about.
The market is just the arena.
The real game starts when the scoreboard gives you freedom.
$IREN is down 14% today and overall down the past month.
Here’s what happened in that same month:
✅ $3.4B $NVDA contract
✅ 5GW strategic partnership with NVIDIA
✅ $3.65B investment grade GPU financing - highest rated ever
✅ $3.0B convertible notes closed
✅ Mirantis acquired - software stack complete
✅ European expansion via Nostrum
✅ $4.4B ARR target announced
✅ First Australian campus - 800MW Bundey confirmed
✅ B. Riley price target raised to $96
✅ Macquarie Outperform - $90 target
The stock is down.
The company is unrecognisable.
Volatility is the price you pay for asymmetric returns.
NFA
Most can't see it yet but I am watching up close as IREN positions itself to make a run at being one of the largest companies on the planet. There are no guarantees of anything in life but IREN has all the right ingredients. Let's freaking go.
$IREN COO @DavidSh17762017 addresses the local impact of the IREN data center in Childress, including water usage.
He describes how @IREN_Ltd helps these rural towns and its residents, participate in the digital economy.
Courtesy of @ABC7Amarillo
Back on X.
It's been a while - the last few years have been heads-down with the @CoreWeave team. This week our engineering team brought up the industry's first NVIDIA Vera Rubin NVL72.
As good a moment as any to start sharing what I'm seeing from inside the AI buildout.
1/6
I've initiated a new position in $NUAI.
What New Era Energy & Digital has done over the last twelve months is one of the MOST aggressive and underappreciated pivots in small-cap infrastructur.
From a Permian Basin helium producer into a vertically integrated AI data center developer sitting on one of the most energy-rich pieces of land in North America.
Let me be precise, because the ticker name still misleads people.
$NUAI (New Era Energy & Digital) operates as a developer and operator of next-generation digital infrastructure and integrated power assets for advanced artificial intelligence hyperscalers.
The company was formerly known as New Era Helium and changed its name in August 2025, which tells you exactly when the pivot happened and how recent the re-rating opportunity is.
Headquartered in Midland, Texas, New Era sits in the heart of the Permian Basin, the most energy-dense real estate in the United States. That's not coincidence.
The single biggest constraint on AI infrastructure buildout right now isn't compute, isn't capital, and isn't land. It's STILL power. Reliable, large-scale, cost-effective power at the point of need.
$NUAI bet is that the Permian Basin, sitting on top of abundant natural gas with deregulated electricity markets, favorable zoning, and proximity to major transmission infrastructure, is the right answer to that constraint. And they're building the infrastructure to prove it.
The flagship asset is the Texas Critical Data Centers (TCDC) campus in Ector County, outside Odessa, Texas.
TCDC is master planned as a multi-phase development, with 438 acres owned (plus a 54-acre corridor pending), designed for phased expansion toward 1+ GW of total capacity, engineered specifically to support large-scale AI and HPC computing workloads for hyperscalers.
The structure of this campus is what makes it interesting. New Era isn't just building a building and plugging into the grid.
They're going vertical by owning the land, controlling the power generation, and delivering what they call powered shells and powered land leases: turnkey infrastructure that lets hyperscale tenants plug in immediately without building from zero.
On the power side: New Era announced a 450MW behind-the-meter generation plan at TCDC through a commercial arrangement with Thunderhead Energy Solutions and Turbine-X Energy, securing access to major generation equipment required for the project, designed to support a hyperscale anchor tenant's AI and High Performance Cloud workloads.
Behind-the-meter generation is the critical move:
Grid interconnection queues can run 3–5 years, building dedicated on-site power transforms TCDC from a data center that has to wait for the grid into a data center that simply doesn't need to.
Construction is set to start this year for a 2027/2028 launch, with the first 200MW phase powered via utility capacity and some 450MW of on-site gas capacity coming online after that.
The product itself is purpose-built. New Era is planning a modular deployment using its 25MW, 200,000 sq ft ATOM design, fabricated by RK Mission Critical, a liquid-cooled design that reportedly supports up to 2,500 racks and densities up to 135kW. That's a serious spec. 135kW per rack is the kind of density you need for the next generation of GPU clusters. This isn't being built for yesterday's workloads.
The most significant recent development at TCDC is the partnership announced in April 2026: New Era signed an LOI to form a joint venture with Stream Data Centers, one of the Tier-1 U.S. data center development platforms, along with an institutional capital partner with deep digital infrastructure and energy investment experience.
The LOI outlines a joint venture development structure in which New Era contributes its site control and local relationships, the Institutional Investor contributes equity capital and sources debt financing, and Stream provides the data center development and operating platform.
This structure matters. New Era is the land and power provider, the hard-to-replicate Permian Basin asset base.
Stream brings Tier-1 operational credibility and the institutional relationships to sign hyperscale tenants.
The institutional capital partner writes the check and arranges the debt.
Each brings what they're best at.
The GP/LP financing model being used here; New Era as sponsor at the project level, with institutional partners providing most capital at the SPV level, is exactly how serious infrastructure gets built.
It limits dilution to NUAI shareholders while allowing the project itself to be capitalized at the scale it needs.
If TCDC is the near-term execution story, the New Mexico campus is where this gets truly asymmetric.
New Era entered into an option-purchase agreement for approximately 3,500 acres in Lea County, New Mexico, for development of a 7 GW AI data center campus, the company's first fully-owned development separate from the TCDC joint venture.
The plan calls for more than 2 GW of natural gas-fired generation capacity, and 5 GW or more of nuclear power, to energize the 3,500-acre site, with the first power generation expected in 2028.
Let's sit with those numbers. Seven gigawatts. For context, the entire U.S. data center industry consumed roughly 200 GW of electricity annually as of 2024.
A 7 GW campus is not a data center. It's a city-scale computing ecosystem. And it's being positioned with a nuclear backbone, which, if executed, would make this one of the most energy-secure hyperscale sites in the world.
The campus will expand incrementally to its eventual 7 GW capacity, utilizing a mix of gas and nuclear generation, with New Era planning to offer both powered-shell halls and turnkey land leases to hyperscale and AI tenants.
The Lea County site was selected for its proximity to major gas transmission lines, existing power infrastructure, abundant water supply, a skilled local workforce, and high-speed fiber connectivity.
These aren't aspirational site criteria. These are the same criteria every hyperscaler uses to evaluate co-location. New Era picked this site because hyperscalers would pick this site.
Combined, TCDC and the New Mexico campus represent approximately 8 GW of total planned capacity, making New Era one of the largest planned AI infrastructure platforms in the United States by raw compute capacity, sitting inside a ~$570M market cap.
The reason this company has a chance at something real, and not just another development story that runs out of capital, is the vertical integration model.
New Era controls land, power generation, and the physical data center infrastructure under one roof. That means:
Power certainty. Behind-the-meter generation means they're not competing in interconnection queues or subject to grid-level curtailment risk. The power is theirs.
Speed to market. Tenants who co-locate with New Era aren't waiting for grid upgrades, they can deploy immediately once the facility is live.
For a hyperscaler trying to deploy 500MW in 18 months, the value proposition isn't just "land in Texas."
It's "land in Texas with dedicated gas-fired power already contracted, a Tier-1 development partner managing the build, and a modular design that starts at 25MW and scales to a gigawatt."
That's a completely different conversation.
What you want to see with a small-cap infrastructure play is whether serious institutional capital is taking it seriously. At NUAI, the signals are there.
Northland initiated coverage with an Outperform rating and an $11 price target, arguing that New Era "now has all of the right pieces in place" to deliver its first lease by fall and "offers a compelling investment opportunity."
Texas Capital assigned a Buy rating with an $8.60 price target.
I want to be direct about the risks here because this is not a mature revenue story, it's a development play, and development plays carry development risk.
Revenue today is minimal. New Era generated $885K in trailing revenue with an operating loss of $24.5M and negative operating cash flow of $11.7M.
The company is burning cash while it builds. That's expected for a pre-revenue infrastructure developer, but it means the investment thesis is entirely forward-looking.
If the hyperscale anchor tenant signing slips significantly, or if power procurement runs into structural issues, the timeline compresses the bull case.
The capital structure also warrants attention: fully diluted shares outstanding of 82 million versus 56 million total common shares as of March 2026. reflecting options, RSUs, warrants, and convertible debt that may increase the share count going forward.
Dilution is a real variable in any pre-revenue infrastructure company; it needs to be monitored.
The New Mexico play, particularly the 5+ GW nuclear component, is long-cycle. Nuclear permitting, even for small modular reactors, is a multi-year process. The initial 2 GW gas phase is executable in the 2028 timeframe; the nuclear buildout extends well beyond that. This is a decade-long asset, not a 2026 trade.
The current ratio of 1.57 suggests the balance sheet is manageable near-term, but execution risk is real.
Most of the capital flowing into AI infrastructure right now is chasing the same names: the hyperscalers themselves, the chip manufacturers, the data center REITs that are already priced to perfection. $NUAI is something different.
It's a Permian Basin energy company that realized it was sitting on the exact ingredients that AI infrastructure desperately needs, cheap gas, plenty of land, favorable regulation, and decided to build the platform itself rather than just sell the electricity.
That's a founder-level insight executed at real speed:
The name changed in August 2025, and by early 2026 the company had 100% ownership of TCDC, a Tier-1 development partner in Stream, a 3,500-acre New Mexico option with a 7 GW plan, and institutional capital in the structure.
That execution velocity in a 12-month window is unusual. And the market cap is still small enough that most institutional funds can't touch it yet.
The anchor tenant announcement is the unlock. When a hyperscaler signs at TCDC, this stops being a development story and starts being a revenue story.
That's when the valuation conversation changes materially.
I'm in early. I'm watching the milestones closely. And I think the Permian Basin is about to become one of the defining AI infrastructure geographies of the next decade.
This is not financial advice. This reflects my personal analysis and position as part of my portfolio. Do your own research. Know your risk.
-BP
The AI bottleneck continues to move and seep into other names that I have mentioned.
So let's start with the basics.
The race was bottlenecked by compute.
Then by memory.
The durable one, the one nobody can print, is power and somewhere to put it.
$IREN just reminded everyone who owns that link (hint, they do).
This morning IREN signed a transmission connection agreement for an 800MW data center campus in Bundey, South Australia.
First announced Australian project, one of the largest in APAC to date, about 78 miles northeast of Adelaide. Four 330kV feeder exits, enough to support the full 800MW without network upgrades.
Submarine fiber into Singapore, Indonesia, South Korea, and Japan. Energization from 2028.
This pushes IREN's power pipeline north of 5GW.
That is on top of the Nvidia deal to stand up to 5GW of AI infrastructure, the $4.4B ARR Blackwell target at Childress, 50,000+ Blackwell Ultra GPUs being validated with BE Networks on Nvidia DSX Air, and a $3.65B investment-grade GPU financing facility closed June 1.
Canaccord went to $79 from $70 today. Cantor sits at $99. This is my largest position and I am not trimming into a power land grab.
Now widen the lens, because the whole stack moved:
→ $MU: HBM4 36GB 12H now in volume production for Nvidia's Vera Rubin, north of 2.8 TB/s, 20%+ better power efficiency than HBM3E. Entire 2026 HBM supply sold out, management can fill roughly 60% of demand. Memory is not a commodity anymore, it is an oligopoly with pricing power.
→ $AAOI: The high-beta way to play the same theme. $324M plus of 800G and 1.6T orders, doubling Sugar Land capex, Mediacom DOCSIS 4.0 win.
→ $POET: Optical interposer, photonic interconnect. Still pre-scale and carrying a securities class action with a June 29 deadline.
Chips, then memory, then optics, then power.
Every layer repriced because demand is structural, not a headline.
Hyperscalers are guiding to $725B plus of 2026 capex, with roughly 75% of it spent on infra, data centers, and power capacity.
The scarce resource is not silicon. It is energized real estate.
IREN keeps buying it.
I have a theory.
$AAOI CEO Dr. Thompson Lin wants to become a billionaire.
It's not often the path to a 10 figure net worth is this clear. PhD Lin is not stupid.
$AAOI is his baby that he founded in 1997.
He started it out of a University of Houston lab and built it in Sugar Land, Texas.
Now you're telling me after all these years, he's going to pass up the chance to make it the Intel of Optics?
Why do you think they're raising so much with ATMs?
Onshore laser fabs aren't cheap, but once built they are a major natsec advantage. To put this in context, they were discussing CHIPS act funding to achieve their goals (very ambitious).
Lin holds 2.26M shares, worth ~$430M at $191.
At ~$443 a share that stake crosses $1B.
That is a 2.3x from here at ~$35B market cap, a number I feel is very feasible given their ramp to a run rate of over $400M/mo or ~$5B in revenue run rate in 2027.
The incentives look aligned here for this cracked scientist CEO, even if he made his first informative sell in years a few weeks back (after strictly buying it before). Give him some respect, he made a 10x+ on buying his own stock with $1M of his own money.
For someone worth $400M+ selling $26M is like you liquidating a little in your brokerage.
But it is very critical you understand the stock so you don't get shaken out.
PE ratios and estimates you read are probably wrong given the ramp.
IREN has announced a planned 800MW data center campus in Bundey, South Australia.
This marks IREN’s first announced Australian data center project and one of the largest in the Asia-Pacific region announced to date.
Learn more: https://t.co/3bOYCUG3pk
Oh great I love the idea that after a decade of founders, engineers, and investors lighting billions on fire to build AI, the plan is: ‘thanks, we’ll be taking 50% of that now.’ Peak startup risk, Soviet exit. 👍🏽🤣
@Agrippa_Inv It's just crazy how undervalued IREN is right now. It's just maths. The numbers get insane when you extrapolate that data across its total power portfolio.