Constraint Edge

Taxing AI tokens is like taxing the Mavericks every time they pass the ball. You might think it encourages fewer wasted passes. But really it just makes the offense worse. The goal is not fewer passes. The goal is more points. Same with AI. The goal is not fewer tokens. The goal is cheaper, better answers.

We should federally tax Tokens at the Provider level. Not a lot. Less than 50c per million tokens. It will accomplish 4 things (at least ) 1. It will push the big AI players to optimize tokenization, caching , routing and localization Which will 2. Reduce energy usage. Saving them in energy costs more than what they paid in tax and reducing strain created by the growth in energy consumption Which will 3. Generate maybe 10 billion dollars a year to start, but over the next ten years could grow 30x to 100x Which will 4. Create a source of funding to pay down the federal debt or deploy, in response to the things AI brings that we don’t expect or don’t like At some point the models will pass it on to customers. Of course. That’s ok. Customers will have the ability to choose between providers. Or to do everything using open source models locally. Thoughts ?

@windingwords Is the right way to think about $DGXX vs $BRUN: $DGXX = data center / infrastructure economics $BRUN = GPU cloud / compute economics Or is DGXX actually trying to become a full neocloud too?

Most people don’t realize how ambitious $DGXX is trying to be with NeoCloudz. They are leaning into the same services $NBIS and $BRUN are offering; CPU Nodes, Storage, and Kubernetes. The website is rough, vibe coded, and missing a favicon. But it launched only some days ago, there is time to improve. I think they will see what customers are requesting, then build for that, expanding capabilities as they go. They will, I hope, hire more talent to expand this side of the business. If they are able to offer add-ons enterprises want, it can take them from ~$10m/MW/y now to ~$18/MW/y.

@windingwords At $14M revenue/MW and $5M profit/MW, 20MW could support ~$100M Extend that to 2030: 50MW = ~$250M income 100MW = ~$500M income 150MW = ~$750M income

@ConstraintEdge this would be at the lower range of your estimates

If that takes 27MW, $BRUN is at $13.9m/MW/year, better than most are getting. Not insane, but good. If they were able to get better rates (since GPU market rates have gone up since sept 2025) then the number could be even better.

@windingwords I still have $5 profit per MW.

From their sept 2025 slides, $BRUN planned 20MW in Q4 2026, right after, they predict $275M ARR for Q4 2026. They have since raised that to $375M, a 36% increase. So I expect MWs to be higher as well. A 36% increase would be 27MW, but since pricing is likely higher (due to higher GPU rates), it might be less.

@TraderCombat @oguzerkan At $25M-32M per MW capex, 10GW = $250B-$320B gross capital need. 4GW $100B-$130B.

@ConstraintEdge @oguzerkan They're raising capital with profits, institutional investments (Meta, NVDA etc) and they have an at-the-market (ATM) option they haven't used yet. I think they will use it some time this coming 12 months to reach the CapEx needs, but bulls are spazzing at me for saying that

$NBIS is showing insane strength today. It doesn’t take a genius to see it can make another 15x from here over the next decade or so. Just look at how their 2026 contracted capacity target ramped up over time: Q2 2025: >1 GW Q3 2025: >2.5 GW Q4 2025: >3 GW Q1 2026: >4 GW This is 4x growth in a year. They have committed to deploy 5 GW by 2030. Even if we assume demand will stabilize over the next 5 years and deployed-to-contracted ratio will shrink from 1:4 to 1:2, we’ll be looking at +10 GW contracted capacity by 2030. Even if the contracted capacity ramp just stops there and it turns to fuly connected power by 2035, we are looking at $100 billion in revenue. Assuming 35% net margin (similar to AWS) and 25x exit multiple, we’ll have a $875 billion company. This is another 15x from here. We are still very early.

@Ren_aramb @babyfolio $BRUN baby NBIS?

When I see posts like this, it just pumps me up to believe in myself and go big or go home. $10M is my goal to quit corporate and go big on entrepreneurship. @babyfolio was one of the first people I followed on X when I first started posting about stocks back in October. His conviction for $NBIS taught me a lot. Truly, after 10 months of trading, my biggest gains by far have come from investing and holding, not trading. You inspire me, sir. I’m also a father trying to create generational wealth for my son. Give this man a follow if you want to learn the hardest part of trading. Growing your own conviction.

@MrInvestor17 Isn't the backlog with Dell just a $1.44B purchase agreement to help finance customer contracts? Infrastructure access and financing support.

so $BRUN is 500m MC with 1.4B in backlog from Dell?

@stepnotonpets When do you think a deal occurs?

$BCARW 1.75 ....Things worth paying attention to.

If $BRUN can fill 150MW by 2029 and earn $4M profit/MW: 150MW × $4M = $600M profit After overhead, financing, and taxes, that could translate to $400M of net income.

@aleabitoreddit What are your thoughts on $BRUN - Boost Run?

Not sure if people realized this but unless a thesis completely breaks, companies like $NBIS can keep growing. Just look at $AMZN or $GOOGL over the past 15 years. If people "trim" it often triggers taxes. And a lot of corrections are typically less than those taxes paid. By the time a "50% crash happens", it's probably already compounded hundreds of even thousands of percent. If people need to pay expenses, once you hit 7-8-9 figures, you can always borrow against those assets and keep letting them appreciate. NFA, just personal opinion. You all do you, but it's highly, highly, dependent on the companies you pick. Can't do this with something trash like $IREN. But I do believe $NBIS is positioned to be the next hyperscaler.

$322310 - Its IR overlay tool is under qualification at two domestic HBM manufacturers for wafer-to-wafer hybrid bonding, measuring wafer alignment and bonding placement accuracy.

@BryzonX Revenue CAGR 25% - 2030 EPS $4?

I am long Clearfield, Inc. $CLFD Here’s my thesis: I’ve been researching a lot about the expanding importance of fiber optics in the new age data center as we move into the agentic ai era The recent deal between $NVDA & $GLW basically confirmed to me that high performance optical fiber is the foundation for AI infrastructure However... as more & more optical fiber start to be deployed data centers are hitting a physical "density wall" creating a spaghetti of cables that blocks airflow and traps heat, the #1 killer of GPU performance Because AI depends on light traveling through glass, physical management is now widely considered signal management In the current 800G environment, a single bend that is 1mm too sharp can leak enough light to stall an entire AI training cluster In traditional data centers, a rack might require a few dozen fiber connections but in AI clusters, they are pushing toward 10x more fiber per rack to support the massive speeds (800G to 1.6T) required for LLMs DC's can no longer expand capacity by just adding more racks. Instead, they are forced to pack more fiber into their existing footprint without causing a meltdown of tangled glass cables and trapped heat And the #1 thing DC's can't afford to have is MORE HEAT generated Here is why CLFD is uniquely positioned for this problem and why the interconnect layer where Clearfield specializes in will soon be a strategic priority for the entire AI industry For 15 years, Clearfield was a niche player protecting fiber in harsh outdoor environments Now, the AI Data Center is the harsh environment. The fiber density and heat levels inside a 2026 AI rack are so extreme that they require the rugged, modular protection Clearfield perfected for the outdoors Management realized this, which gave them an opportunity to enter the data center space In Jan 2026, they announced their new product “NOVA” which is their pivot to capture the market share of optical fiber density management The NOVA HD Panel is the only cassette based system that can fit 384 LC fiber ports into a 4U space (just 7 inches tall) While also guaranteeing a 0.2dB or less insertion loss (50% better than peers) Reducing insertion loss by 0.2dB can reduce a data center's overall power consumption by 10%, making Clearfield an infrastructure choice for an already power scarce environment For those who don't know, At 800G speeds, high insertion loss causes data errors, forcing AI systems to resend data, which wastes power and slows down training Here's the exciting part... Just a few days ago, the CEO spoke at Needham and she said that the data center community’s "biggest surprise" was recognizing Clearfield not as a newcomer, but as a superior expert in fiber termination and management that they now desperately need When NOVA was announced in Q1 26’ they didn’t expect any revenue at all until 2027 but in their recent Q2 26’ ER, orders already started flooding in Reporting a book to bill ratio of 1.3, backlog growing 39% for “adjacent markets” aka data centers at $31.6M On the call, management noted that data centers are skipping the typical multiyear testing phases because the current infrastructure is literally failing to support the new gen AI chips oh... and Clearfield is also a direct beneficiary of the move toward CPO A major challenge for CPO is that lasers are heat sensitive and fail often if they are buried inside a hot AI chip package The industry is moving toward ELS, placing the lasers at the front of the rack in a separate, hot swappable module Clearfield’s NOVA platform is designed specifically to act as the "standard socket" for these external lasers It provides the high density frame that allows technicians to pull out a failed laser cassette and snap in a new one in seconds without touching the delicate AI chip I am a buyer today because the signal from management is their confidence in their ability to pick up new customers given their expertise in the field + they had earnings already with a solid reaction which is a good signal for me I am anticipating their backlog to pickup at an even more rapid pace now fiber is being deployed at record pace This is an incredibly niche play, but I think their is plenty of upside of this company if they execute properly NFA. Please use for RESEARCHING PURPOSES ONLY.

@windingwords Are they both selling the same thing?

@ConstraintEdge I don't know, but that would be $18.75M/MW/year, high, but not impossible with enterprise customers (not hyperscalers) + value added services? $DGXX did a 2 year deal with SubQ for ~$9.8M/MW/year but their cloud is newer, basic features.

Can someone confirm that $BRUN is projecting $375M ARR on 20MW?

$BRUN math is interesting if they can price 15% below $NBIS on H200-equivalent GPU cloud pricing. Nebius H200: $3.50/hr on-demand, $2.30/hr committed. 15% discount = $2.98/hr and $1.96/hr. BRUN’s 2026 plan implies ~950 GPUs/MW. At 85% utilization, that equals $14M to $21M revenue/MW. If cash cost is $10M to $12M/MW, cash profit could be $4M to $10M/MW. BRUN has not fully disclosed its GPU mix. The upside depends on actual GPU class and utilization.

@aleabitoreddit The buildout is 50B per GW with chips that are depreciating.

As I said before $IREN is basically dogsht compared to $NBIS. $NVDA didn’t give $IREN funding yet. So IREN needs to figure out how to buy enough GPUs to monetize 5GW capacity through their 6B ATM and other means. It’s an endless dilution machine just because they secured power. I call $IREN holders 0 IQ because they just buy in it to get diluted without understanding nuances of financing. Nvidia actually gave $NBIS funds. While Nvidia got a free no-risk purchase agreement for allowing $IREN to use their logo. $IREN is basically a marketing company at this point, while the other Neoclouds actually allow equity appreciation.

@StockSavvyShay 30x 27 earnings?

MICHAEL BURRY STARTED A NEW POSITION IN $MELI

@Million_Sancet HBM moves to 12-high and eventually hybrid bonding. If Auros qualifies at Samsung/SK Hynix, this becomes a tiny Korean toll-booth on HBM yield.

This company is $322310 Auros Yes, gentlemen, not everything is about Europe and the US Enough with $SIVE $LPKF $IQE … Asia is on fire I like Japanese stocks but they are more established Taiwan is also non stop but I had some issues with IBKR and their currency TWD I prefer to avoid them, just a personal thing So South Korea is now in my sights Take a good look at this massive company $322310 (Auros) is a key but little known piece of the Korean semiconductor supply chain Its MC is barely 210-269M Tiny compared to its peers My thesis is simple, pay attention It is based on a transition from a decade long development phase to an imminent high volume manufacturing (HVM) stage The company is betting on expanding into two business lines In which it seeks to break established monopolies Metrology for HBM (HBM4/HBM4e/HBM5) Specializing in infrared (IR) metrology for the hybrid bonding process They currently compete in a sector historically dominated by $KLAC It is in the qualification process with $005930 (Samsung) and $000660 (SK Hynix) A strong volume increase is expected in the second half (H2) of 2026 Driven by the deployment of 12-layer HBM Thin Film Measurement: It is in the validation phase with major domestic Korean manufacturers The goal is to start large scale supply during the current year If this transition works and is successful, the valuation could exceed 630M Basically a 3x Considering it trades at a massive discount compared to peers like $6315 (Towa) Recent Trendforce reports indicate that hybrid bonding validation for 12 layers has been completed Which accelerates adoption timelines Success depends entirely on passing technical qualifications Any delay in the move to mass manufacturing would affect the investment thesis This is something to keep in mind when you want to allocate to this position