Evan Ye

[Evan Ye @Milolo1028]

@flyingfishbug 花之哲学！

[Evan Ye @Milolo1028]

@StockSavvyShay No clue at all what are we seeing rn

[Evan Ye @Milolo1028]

@flyingfishbug 谢谢宝姐分享。投入的时间，效率，和处在的环境同样重要。

[Paulina宝姐 @flyingfishbug]

Two things in life are worth committing to for the long run.

1. Think in terms of compounding. Keep building.

Books you read, knowledge you gain, experience you accumulate, trust you earn, and good decisions you make may not seem like much in the short term. But give them time, and their value can grow exponentially.

2. Surround yourself with people who are genuine, intelligent, and positive.

Genuine people make trust easier and life less complicated. Intelligent people challenge your thinking and expand your horizons. Positive people help you see opportunities where others see only problems.

The people around us shape us more than we often realize. They influence the information we receive, the judgments we make, and ultimately, the direction our lives take.

Let time compound your efforts. Save your energy for people worth growing with.

⸻

人生有两件事，值得长期坚持。

第一，建立复利思维，持续积累。

读过的书、学到的知识、积累的经验、建立的信用、做过的正确决策，短期看可能都不起眼，但时间会放大它们的价值。

第二，选择与真诚、聪明、积极的人同行。

真诚的人，让你少一些防备，多一些信任；聪明的人，帮你打开认知的边界；积极的人，让你在问题之外看到机会。

人与人之间的差距，很多时候来自长期相处后的相互影响。你与什么样的人交往，决定了你接触什么样的信息、形成什么样的判断，也影响着你最终走向哪里。

把时间交给复利，把精力留给值得的人。

[Black Panther Capital @BlackPantherCap]

Let’s run it, shall we…

$IREN Napkin math:

• Market cap today: $15B
• Q3 FY26 / Mar 2026: $144.8M (transition quarter: AI Cloud up sharply sequential, mining still majority but declining)
• TTM revenue: $757M
• Year-end 2026 AI Cloud ARR target: >$4B (raised from $3.7B; 85% under contract)
• 2027 capacity/revenue potential: 1.2 GW gross AI Cloud → multi-billion run-rate trajectory
• Target market cap framing (on $4B+ ARR at scale): significant multiple expansion potential as revenue converts
• Near-term upside: execution on Horizon deliveries + backlog conversion

1/ $IREN is a vertically integrated data-center platform pivoting hard from Bitcoin mining into high-density AI Cloud infrastructure. It owns power, land, design/engineering/construction, and now GPU fleets. Key sites include Childress (TX), British Columbia, plus expanding North America / Europe / APAC pipeline. Recent moves (Mirantis acquisition for orchestration/monitoring, $NVDA Exemplar Cloud status) strengthen the full-stack offering for hyperscalers and AI labs.

2/ AI demand acceleration:

Massive contracted demand for liquid-cooled, high-power-density capacity. Flagship is the five-year, $9.7B $MSFT deal (Horizon 1–4 at Childress. Horizon 1 delivered & accepted Aug 2026 with NVIDIA GB300 NVL72 Exemplar status). NVIDIA deal $3.4B.

July 2026: additional $2.8B multi-year contracts with leading AI developers (Perplexity, Figure, Together, Fluidstack, Fireworks, etc.).

85% of the >$4B 2026 ARR target now under contract. Weighted average term 4 years. Recent deals include 45% customer prepayments on GPU CapEx. Secured power >4.5–5 GW.

Pipeline supports 480 MW gross AI Cloud capacity by end-2026 and 1.2 GW in 2027.

3/ Momentum:

Q1 FY26 revenue was a record $240.3M. Subsequent quarters reflect the deliberate mining-to-AI transition (mining hardware decommissioned ahead of GPU installs), with AI Cloud revenue roughly doubling sequential in the latest reported period to $33.6M. Horizon 1 acceptance is the first major conversion of contracted ARR into live capacity. Cash position has been strong (multiple billions earlier in the year; substantial financing packages including investment-grade GPU debt).

4/ Platform scaling:

Vertically integrated model (own power + build + operate) is the differentiator versus pure GPU lessors. Global team scaling; acquisitions add software/orchestration layer. Near-term focus: Horizon 2–4 deliveries later in 2026, additional GPU commissioning, further contract conversions across the 2026–2027 expansion program.

5/ Financial path:

Gross margins historically strong on mining; AI Cloud expected to deliver higher, more predictable margins and operating leverage once capacity is online and utilization ramps. Prepayments + dedicated GPU financing cover the bulk of near-term CapEx, reducing dilution pressure over time. Company is still in heavy investment mode; profitability inflection expected as the $4B+ ARR base materializes (back-end weighted into late 2026 / 2027).

6/ Revenue build:

• AI Cloud (Microsoft Horizon, NVIDIA, multi-developer contracts) – primary long-term driver
• Residual Bitcoin mining (intentionally being displaced)
• Contributions from acquired capabilities and international sites
• Storage/ancillaries and higher-value managed services

Stock has pulled back from 2026 highs ($70s) amid transition volatility, dilution concerns from prior raises, and the lag between contracted ARR and GAAP revenue recognition.

Core thesis: one of the few pure-play, vertically integrated AI infrastructure names with multi-GW secured power, blue-chip contracts already converting (Horizon 1 live), and a clear path to multi-billion ARR.

AI infrastructure demand remains supply-constrained. $IREN is building and delivering the capacity at scale.

Risks: Execution/timing on remaining Horizon phases and broader 480 MW / 1.2 GW build-out, residual mining revenue air-pocket, capital intensity and any future dilution. Customer concentration ($MSFT large), competition for power/GPUs, $BTC price sensitivity in the transition window.

Catalysts: FY26 full-year results (Aug 27, 2026), Horizon 2–4 deliveries, additional contract announcements, ARR-to-revenue conversion visibility, further financing or prepayment updates, profitability milestones.

My personal 2027 price target range for $IREN:

Base case $80–$110
Bull case $130–$160+
Conservative/base-to-bull midpoint $100–$120

Key assumptions if they deliver:

• End-2026 AI Cloud ARR >$4B converts into meaningful GAAP revenue in 2027 (back-end weighted ramp means full-year 2027 revenue likely in the $2.5–$4B range, with some Street estimates clustering near $3B and higher scenarios if utilization and additional contracts land).

• 1.2 GW gross AI Cloud capacity largely online/commissioned by end-2027.

• Gross margins expand as high-margin AI Cloud displaces residual mining; operating leverage kicks in; prepayments + financing continue to cover most CapEx.

• Shares outstanding roughly 360–400M (current 357M; some further dilution possible but not extreme if prepayments reduce equity needs).

• Market assigns a growth multiple once the transition is proven (not pure Bitcoin-miner multiples).

Napkin valuation math:

• Current market cap $15B at $42. Apply forward EV/Sales or P/S of 8–12x on 2027 revenue of $3–4B (reasonable for a vertically integrated AI infrastructure name with multi-year contracted revenue once execution is de-risked; peers have traded in wider ranges depending on pure growth vs. profitability stage).

• $3B revenue × 10x = $30B enterprise value → equity value $28–30B → $75–85/share (base).

• $3.5–4B revenue × 10–12x = $35–48B → $90–130/share.

• Bull (strong conversion + more contracts + multiple expansion toward higher-quality data-center/AI infra comps): $130–160+.

Street 12-month consensus targets currently sit around $80–$83 (range roughly $41–$131, with several recent raises into the $90s). Longer-horizon models that bake in the 2027 capacity and ARR conversion push higher.

What has to go right

• Horizon 2–4 (and broader 480 MW → 1.2 GW) delivered on schedule with customer acceptance.

• ARR converts to recognized revenue without major utilization or pricing slippage.

• No large unexpected equity raises that dilute the share count heavily.

• Broader AI infrastructure demand remains strong and power/GPU supply constraints persist in $IREN favor.

TL:DR If the pipeline and ARR-to-revenue conversion largely deliver, a realistic 2027 zone centers around $90–$120 (roughly 2–3x from today’s $42), with upside to $130–160+ in a clean execution + multiple expansion scenario and downside if timelines slip.

Please note:

This is directional napkin math, not a formal model, execution remains the swing factor. This is not financial advice.

-BP

[憨厚的麦总 @Michael_Liu93]

最近这几天有不少小伙伴开始看AI，问我不是技术背景的该怎么去看AI这个市场，复杂的技术名词很多，然后各个细分赛道也很杂，怎么去做尽调？我就大概说一个框架吧，然后你在框架里研究就非常易懂了。

首先，现在大家去投资的这些AI股，投的商业本质是什么？
答：“美国一年6000-7500亿美元的AI基础设施的资本支出（AI Infra Capex），这个数字到2027年大概率会在1-1.2万亿左右，更简单点说，美国的公司现在每年砸6000-7500亿去买AI相关的基建设备”

主要是谁在投资这些AI基建呢？是美国的五大云厂商，Microsoft、Amazon、Google、Meta、Oracle（这些统称是Hyperscaler）

所以这一年将近1万亿美金的资本支出都砸在什么上了？主要是三大类：
1. IT设备 / 计算硬件（占比60-65%）
2. 电源与冷却（15-20%）
3. 建筑与土地（10-15%）

在IT设备 / 计算硬件里再拆分
1. GPU（65%）
2. 网络（10-15%）
3. 存储（8-10%）
4. 其他包括CPU、软件、HBM、机架（10-15%）

沿着这个思路你再去做行业图谱（Industry Mapping）就简单多了。

1. GPU
你投这个赛道基本只有英伟达，大头都被他一个人吃了（当然还有一些云厂商自己搞的自定义Asic，自己设计或深度定制的AI加速芯片，本质上是去替代掉一部分英伟达gpu的成本，比如Google (TPU)、Amazon (Trainium/Inferentia)、Microsoft (Maia)。

2. 网络
英伟达，博通，思科，HPE，最近火的Marvell（MRVL）就可以算在这里

3. 存储
海力士是绝对的第一，然后三星，美国的存储巨头就MU和SNDK（闪迪），溢价高，但市场份额没有海力士高。

4. 电源与冷却
做AI数据中心最大的三家，Vertiv（VRT），Schneider（SBGSY），Eaton（ETN），规模最大的是Schneider，但这个市场总体不算很集中，比较分散。

5. 建筑与土地（Building & Land）
完全分散的市场，这里环节就更多了，总包，涉及，工程，电气，开发，土地，等等，投AI的话这块儿没必要继续细看。

6. 其他类别
a. 做软件的，做cpu的，做机架，做光纤，做自动化等等，就不展开说了

然后围绕上面几大分类，又有做服务的，比如做集成的（就是把这些以上打包成一个整体服务器卖出去），Dell，联想，SMCI；然后英伟达的GPU的供应商又有一大堆，也不展开了。

结尾：框架就是这么个框架，本质上你现在投AI产业，你在买单的Thesis就是：
1. 美国在每年向AI产业的基建投近1万亿美元，并且未来五年每年至少30-50%的增长。
2. 抓的是AI资本支出中不可或缺环节中有足够强生态占位的玩家
3. 最终消费端，也就是openai，claude这些大模型的消费者，我们，会持续消费。

所以什么时候AI“泡沫”会见顶？
1. 这些AI巨头们融不到钱去继续投基建的时候
2. 我们这些消费者不再继续买单，大模型的营收开始下滑的时候

至少目前来看还早。

[Evan Ye @Milolo1028]

@flyingfishbug ✌️

[Julian Ciccale @jciccale]

1/ I am sharing here an extract of my report on $IREN earnings. Better late than never. There's a bonus with my PTs at the end of the thread

TL;DR
Story: Unchanged. Price action: down 25%, Action: hold.

$IREN reported Q2 FY2026 results that reflect a company deep in its transformation from BTC miner to AI Cloud platform. Revenue came in at $185M (down 23% QoQ), driven by declining BTC mining as GPUs replace ASICs, while AI Cloud rev more than doubled to $17M. The net loss of $(155.4)M is almost entirely noise, $219M in non-cash/one-time items from convertible note mechanics and a debt conversion inducement expense.

The real story is what happened outside the P&L: $3.6 billion in GPU financing secured at <6%, a new 1.6GW site in Oklahoma, and $2.3 billion of AI Cloud ARR now under contract. They're on track to deploy 140k GPUs by end of CY2026, targeting $3.4B in AI ARR.

I am not concerned at all about the short term price action and holding my shares tight. If anything, I plan to add slowly as this, market wide, sell off resolves.

[Black Panther Capital @BlackPantherCap]

$RKLB is without doubt my current highest Space conviction for 2026 - 2027.

Here's why it can reach $205 and 3-4x:

The MOAT nobody’s talking about:

$RKLB isn’t just a launch company anymore.

They’re the ONLY vertically integrated space prime that controls STAGGERING 90% of their supply chain in-house.

Think about what that means:

• They manufacture reaction wheels, star trackers, solar arrays, AND launch the rocket

• They’re the “merchant supplier” to the entire industry.

• They capture value on missions their competitors launch.

It's IMPOSSIBLE to just “build a competitor” to a decade of flight heritage and dual launch sites.

The quiet margin expander: Space Systems

Space Systems revenue grew 48% Y/Y in Q3 2025. This segment is already majority of their $1.1B backlog.

But here’s the part everyone’s missing:

Solero solar margins are normalizing above 30%. The Flatellite satellite bus is entering mass production for SDA constellations.

The Mynaric laser comms acquisition locks them into the high-bandwidth satellite internet buildout.

By 2027, Space Systems could be 60%+ of revenue at 40% margins.

That’s not a launch company. That’s a defense prime with recurring revenue.

The $1.3 billion validation:

In December 2025: $RKLB wins $816M SDA Tracking Layer contract.

They’re now a LEAD PRIME alongside Lockheed Martin, Northrop Grumman, and L3Harris for the $175B Golden Dome missile defense program.

Their StarLite sensors and Phoenix payloads are being used by OTHER primes in the program. Total value capture: $1.3B+.

The U.S. government just told you who the winners are.

And $RKLB is one of them.

The real 10-bagger:

Tank rupture = uncertainty and noise. Every new rocket fails during qualification testing. That’s why you test.

Here’s what matters:

• Neutron moves $RKLB from the $300M small-sat niche into the $50M+ medium-lift constellation market.

• 13,000kg payloads vs. Electron’s 300kg.

• Archimedes engine production: 1 engine EVERY 11 days (already built).

• First flight: Mid-2026

If Neutron flies successfully, $RKLB breaks the small-launch ceiling forever.

They become the “Falcon 9 alternative” for operators who can’t wait 6 months for a SpaceX rideshare or refuse to fund their competitor.

Institutional ramp-up:

• Baillie Gifford: Added 5.7M shares ($400M)
• Vanguard: Added 5.6M shares ($391M)
• Total institutional ownership: UP 7.88%
• 986 funds now hold positions

These aren’t retail tourists. These are multi-decade horizon funds positioning for the infrastructure build.

The ALMIGHTY SpaceX IPO Catalyst:

SpaceX is valued at $1.25T after the xAI merger. Mid-2026 IPO likely.

When that happens, institutional capital looking for space exposure will have exactly ONE other option:

a flight-proven, vertically integrated U.S. space company with defense prime status.

$RKLB is the ONLY public benchmark.

If SpaceX trades at 100x sales, the market will easily justify 80-90x for $RKLB.

That’s the re-rating event.

Analyst consensus:
• 2025 Revenue: ~$600M
• 2027 Revenue: $1.29B (2x growth)
• 2027: First GAAP profitable year
• 2027: Positive Free Cash Flow

This is the year $RKLB stops being a “story stock” and becomes a cash-generating defense infrastructure play.

The numbers that matter:
> Current: $72, ~$39B market cap
> Q3 2025: $155M revenue (ATH)
> 41.9% non-GAAP gross margin

Q4 2025 Guidance:
> $170-180M revenue
> Backlog: $1.1B
> Liquidity: $1B+ (11 quarters of runway)

Flight Record:
> 81 Electron missions
> 100% success rate in 2025

The setup right now:

$RKLB pulled back from $99 to $66 recently.

The thesis didn’t change. The SDA contract didn’t change. The institutional buying didn’t change.

Only the price did.

My targets:

EOY 2026: $140 (SpaceX IPO halo effect + Neutron maiden flight)

EOY 2027: $205 (GAAP profitability + margin expansion)
The Risks (Because I’m Not An Idiot)

Concerns:
• SDA contract execution delays.
• Neutron maiden flight failure (likely for new rockets)
• SpaceX Starship unit economics collapse the medium-lift market.

These are real. But the $1B liquidity cushion + defense prime status = survivable downside.

The bottom line:

$RKLB at $72:
• $1.3B+ defense contracts (non-discretionary revenue)
• 90% vertical integration (margin protection)
• Space Systems segment quietly doing 48% Y/Y growth
• $1B+ liquidity (fully funded through Neutron)
• Institutional ownership ramping aggressively
• Only U.S. public space company with 81 orbital missions

Neutron flies mid-2026. SpaceX IPOs mid-2026.

2027 profitability hits.

You’re getting a defense prime with a rocket factory at a growth stock valuation BEFORE the re-rating event.

I’m buying EVERY pullback.

Note: This is NOT financial advice.

[Jake Wujastyk @Jake__Wujastyk]

If you’re an $ONDS investor, make sure to read this write up by @SharkChart

[Dulce @litigious_dulce]

Next week $IREN reports earnings. I am confident the the update will be impressive--particularly regarding CSP in Canada--and I will cover my predictions below. However, I will also include something that is more hope than expectation.

Obvious Predictions:
1. Bitcoin mining continues to be an afterthought and may not even be mentioned in the presentation.
2. The balance sheet will look pretty messy because of the convertible notes, upsetting some financial analysts.
3. IREN will announce a hyperscaler deal for Childress. The market suspects the tenant is $AMZN, and I have no reason to disagree.

Surprising Predictions:
4. CSP in Canada will exceed guidance. Due to the credibility from its $MSFT partnership, there should be much greater demand for IREN's CSP. Further, $/GPU-hr for every generation of $NVDA GPU has increased significantly. IREN will significantly raise guidance for CSP, both in terms of GPUs to be installed and ARR. I think this will have the greatest impact on stock price.
5. IREN will announce a host of new strategic partnerships, formed following the MSFT deal. For example, AMD may be a partner now given the dearth of NVDA Blackwells.

Hopeful Predictions:
6. Regarding the hyperscaler deal for Childress, I hope that it is still CSP but a JV this time. I still maintain that colo3o (a term I coined that at this point sounds like gibberish) is the most financially efficient arrangement because it lets each partner contribute its strongest assets while lowering overall capex and improving capital efficiency through creative structuring.

For example, Amazon can act as the OEM provider of Trainium accelerators, meaning the hardware comes at a significantly lower internal cost than market-priced GPUs, reducing the JV’s compute capex. IREN can contribute land, data center infrastructure, and contracted 2026 power, and can potentially lease the accelerators back into the project rather than requiring Amazon to fully fund and own the stack. Even if IREN does not directly lease the GPUs, the JV structure enables more flexible financing—hardware-backed leases, capacity prepayments, or revenue-sharing arrangements—that recycle capital across the ecosystem. In this way, the partnership can function almost like a circular economy, where compute, power, and infrastructure are jointly financed and continuously redeployed to maximize utilization and returns.

Accordingly, the MSFT deal can be viewed as a foot in the door for the CSP JV. Prior to MSFT, I doubt anyone would have wanted to do a JV with an unproven bitcoin miner, let alone give that miner the economic majority of 51% (which is required for a consolidated balance sheet). If IREN can succeed in doing a CSP JV with AMZN, then that establishes important precedent that clarifies its vision for the future, one where it is the dominant neocloud by far.

[Shay Boloor @StockSavvyShay]

A bullish setup for me is when the 9-day EMA is above the 21-day EMA & the stock is holding both the 50-day & 200-day SMA so both momentum + structure are moving in the same direction.

If the only level a stock is holding is the 200-day SMA then the long-term trend is still intact but the momentum has faded so I treat it as a hold.

When the price breaks below all three levels then that is when I consider the trend bearish because the short-term, intermediate & long-term signals are all pointing the same way

[Phil @lynch_legacy]

Stock of the Week: $KRKNF - Kraken Robotics.

First things first - logo is an A+. The story might be even better.

RELEASE THE KRAKEN ⬇️ 🧵⚓️

[mon @moninvestor]

I’ve been watching a growing trend on social media that investors should be careful with.

People claim they called a stock at one dollar or two dollars, but when you look for receipts, they are rarely there. Most of the time, the proof is an old post where around ten different stocks were listed, each with a claim of ten times potential.

If you throw enough darts at the board, something is going to hit. That does not tell you much about judgment or process.

In 2025, I had five stocks in my long-term portfolio. I only add a stock to that portfolio if I believe it has ten times potential at the time of entry. Two of those stocks have already reached that target. IREN and HIMS both hit my ten times price target from where I first bought them. ASTS and ONDS are still progressing toward that level based on my original entries, which were posted on X when they were made. Anyone can go back and verify that.

The fifth stock was RR. I still believe RR has meaningful upside, but the execution risk is higher than the others. My conviction there has declined relative to the rest, though I still think the upside case exists. I expect it to take more time and require clearer execution.

All five stocks were added under the same upside framework. Two have already delivered. Two are still working through their trajectories. One remains more uncertain but not written off.

Today, I remain invested in IREN, HIMS, and ONDS. I sold ASTS solely because I run a very concentrated portfolio with limited capital. Position sizing matters. If I had more capital, I would still be holding ASTS and likely RR as well. RR would probably sit outside my core long-term allocation, but I would still own it.

With limited capital, every position has to earn its place. My capital stays where upside and conviction are strongest at this point in time.
Right now, that is IREN, HIMS, and ONDS.

Investors should check for documented decision-making.

> Pointing to winners after the fact is easy.
> Showing entries, staying invested through volatility, and adjusting conviction over time is the real work.

[Ray Myers @TheRayMyers]

"2026 is an inflection point for AI Revolution," Dan Ives

He predicts that tech stocks will climb 20% to 25% in 2026 as the AI revolution moves into a phase of broad monetization!

Dan emphasizes that this growth will extend beyond big tech to include "physical AI" leaders like $NVDA and $TSLA, particularly through advancements in humanoid robotics and autonomous systems that will be showcased at CES.

The acquisition spree is not over, as companies spend to win in this "AI Arms Race". He previously said that $NBS is a clear acquisition target. Would $AMZN $MSFT or $GOOGL or someone else shel out $40-50B for $NBIS?

Furthermore, the US currently holds a lead over China in the AI race thanks to $MSFT, $PLTR, and $NVDA. However, Chinese players like BYD, Alibaba, and Tencent are narrowing the gap through robotics and EVs.

He identifies Microsoft as a primary winner in the enterprise space with significant unpriced monetization potential, while also noting that Apple is finally entering the consumer AI revolution in a meaningful way.

Lastly, despite some investor nervousness, Ives maintains a highly bullish outlook, stating that the AI "party" is only in its early stages and that massive capital expenditures of $600B by big tech will continue to drive the market forward!

[Jim Liu @jiahanjimliu]

$IREN 2026, 2027 Financial Model with Feedback from Frans, Dulce

Model: https://t.co/cNsypVJ1or

1. @FransBakker9812 suggested a more accurate model for 2026 based a reasonable build out rate of 200MW IT of Hyperscaler IaaS for SW1 and 200MW IT for H5-8. H9-10 is not included in 2026 price target and SW1 has only 200MW IT towards revenue in 2026. P/E of 40.
2. @litigious_dulce suggested a model for 2027 based on SW1 build out rate accelerating in 2027 and IREN fills them with Vera Rubins. To model Vera Rubin I did 1.5x to both the unit price and hourly rate. Vera Rubin also takes more power so we can only fit 140/200 the amount. 140/200 corresponds to rack power ratio of Vera Rubin to GB300. Dilution is this model is 60% on top of all convertible shares counted as outstanding shares. Vera Rubins are expensive but as IREN grows in market cap, it takes less dilution % to get same amount of funding. P/E of 30.

Feel free to modify the variables for yourself. Default price targets:
1. Canada Only: $51.60
2. Canada + Horizon 1-4: $78.21
3. Frans 2026 Canada + Horizon 1-8 + SW1 200MW IT Hyperscaler IaaS: $157.60
4. Dulce 2027 Canada + Horizon 1-10 + SW1 933MW IT Hyperscaler IaaS: $298.45