Nowhere close to you both, but was an early investor in $PLTR back in 2022 / 23 when nonone cared. Invested in HBM ( $MU, $DRAM) Invested in photonics ($AIXA, $TSEM, $VECO, $LITE, $COHR, $CRDO) Invested in other semis too ($SANM, $KLIC) Very old investor in some of mag7 including Apple, $GOOGL, $AMZN Invested in minerals mines (Gold, Copper), now thinking of tungsten

A new position I’ve Invested in 🐊 Workiva ($WK). Here’s Why. While the market worries about AI replacing software, I’m looking for businesses where AI could strengthen mission-critical workflows. Workiva provides software for financial reporting, compliance, audit and sustainability reporting. These are areas where accuracy, traceability and control matter. 🐊 What I like ✅ Revenue grew 19% YoY to $255M in Q2 2026, with subscription revenue also up 19%. ✅ Free cash flow was $78M, versus $49M a year earlier. However, Q2 is seasonally strong; full-year FCF margin guidance is around 21%. ✅ Gross retention was 97% and net retention 111%. Encouraging numbers, although net retention also reflects pricing and expansion of existing solutions. 🟢 My thesis Workiva has recurring revenue, sticky enterprise customers and a position in complex reporting workflows. I believe AI could enhance its value rather than simply make it obsolete. 🚩 What could go wrong? GAAP operating margin was just 4.6%, versus 16.8% on a non-GAAP basis. Stock-based compensation and dilution deserve close scrutiny, particularly alongside buybacks. Competition, pricing pressure and valuation are other risks. A good business is not automatically a good investment at any price. I’ve invested because I see a potentially durable enterprise software business but future cash generation must justify the price I pay. The question: can Workiva turn its sticky customer relationships into durable per-share value in the AI era? I own shares in Workiva. Not financial advice.

Markets are down on AI fears and Broadcom CEO’s Statement on AI investing like casino gambling is being misinterpreted, I have a different read Broadcom CEO Hock Tan compared some AI investing to casino gambling. The market heard a warning about the AI boom. I think that misses the more important message. 🐊 What Hock Tan actually said “A lot of investors have no clue what they’re investing in, but set very extreme expectations only to find out that it’s almost like you are in a casino gambling.” But look at the context. Tan said capital, not memory or hardware, could become the biggest constraint on AI growth. He also acknowledged that leading AI labs are generating significant revenue. He is questioning investment discipline and funding, not declaring AI demand dead. 🚩 How the market reads it AI investment is getting excessive. Some companies may struggle to fund their ambitions, monetise their models or justify their valuations. Therefore, the entire AI trade deserves a rethink. That risk is real. But treating criticism of speculative investing as proof that AI infrastructure demand is collapsing is a different argument. 🟢 How I read it The AI opportunity and the AI investment bubble can coexist. Demand for computing, networking, memory and optical connectivity can keep growing while individual stocks become overpriced. The bottleneck could shift from chips and memory to capital, power, data-centre construction and, ultimately, customers’ returns on AI spending. That changes which companies benefit, and when. 🐊 What this means for investors like me I don’t want to buy a stock simply because AI demand is booming. I want to know who has real orders, who can convert them into free cash flow, who needs continued external financing, and how much future growth the share price already assumes. Tan’s comments aren’t a signal to blindly sell AI stocks. Nor are they a reason to ignore the risks. They’re a reminder to separate a great industry from a great investment. The question isn’t whether AI is real. It’s whether the returns will justify today’s expectations. Are investors confusing a warning about speculation with a warning about AI itself?

Micron $MU: Rosenblatt raised valuation to $1900 But To Me The real valuation debate is 2029, not 2028. As you all know I am invested in Micron and wider memory ETFs, but generally am always staying on the edge, trying to understand bear case as well and for me the key thing is when to sell. The good news is that, Micron delivered $54.2B in quarterly revenue, up 379% YoY, and guided next-quarter revenue to $61.5B. So the bull case is getting stronger: 🐊 HBM’s wafer requirements constrain industry supply growth. 🐊 10 additional Strategic Customer Agreements SCA improve demand visibility. 🐊 Management expects gross margins to rise again after Q1. 🐊 FY27 capex is now expected to exceed $50B, accelerating future capacity. Rosenblatt raised its target to $1,900, valuing MU at 10x FY28 estimated earnings. But that’s an analyst scenario, not intrinsic value. Here’s the tension I care about: Their earnings power is extraordinary. But Micron is investing heavily to expand capacity, and competitors won’t stand still. If customer commitments, AI demand and HBM complexity make this cycle structurally stronger, today’s earnings multiple could prove misleadingly low. If supply catches up and margins normalise, today’s record profits could be a dangerous valuation anchor. The question isn’t whether Micron can deliver FY28. It’s how much of today’s earnings power survives beyond it. Not financial advice. Past performance does not guarantee future results.

Across the year, I’ve kept talking about photonics and I’m invested in some of these names. The more I dig, the more I think this is bigger than optical transceivers. AI is becoming a data-movement problem. Copper can’t scale forever, so light increasingly becomes the highway for moving data. 🐊 $LITE ~$1,085vs 52-week high $1,086. Lasers/optics; key AI-data-centre beneficiary. 🐊 $VIAV ~$46.57 vs $60.43 high. The “traffic inspector” testing optical networks. 🐊 $AAOI ~$115 vs $233.67 high. Vertically integrated lasers → optical modules; more volume leverage, more execution risk. 🐊 $COHR ~$335 vs $440 high. Broad photonics exposure: lasers, materials, components and transceivers. 🐊 $AXTI ~$86 vs $143.16 high. InP substrates — potentially a bottleneck, but China export-permit risk matters. 🐊 $IQE ~52p vs 73p high. Upstream compound-semiconductor materials. 🐊 $CRDO ~$228 vs $308.67 high. Copper + optical connectivity, increasingly moving deeper into silicon photonics. 🐊 $AEHR ~$107 vs $147.40 high. Wafer-level testing/burn-in, including silicon photonics. 🐊 $SIVE ~32.5 SEK vs 110 SEK high. InP lasers/photonics; high potential, but still a speculative execution story. The question I keep coming back to: Where is the bottleneck that AI demand cannot easily bypass? That may matter more than simply finding the company with the biggest headline growth. Prices are around Oct. 2, 2026 and can move quickly. Past performance does not guarantee future results. Not financial advice. stockcrock.substack.com/p/ph…

Photonics just became an even more interesting supply-chain story I’ve talked about photonics quite a bit this year, and I have relatively high exposure to the space. So when $LITE, $COHR and $AAOI started moving sharply, I obviously paid attention. But for me, the interesting part isn’t simply that the stocks are up. It’s where the value could actually move through the supply chain. 🐊 The latest catalyst is policy, but it’s not a done deal. Morgan Stanley’s Washington checks suggest the FCC is discussing restrictions on Chinese optical transceivers, potentially starting with 3.2T modules, with implementation potentially beginning as early as October. The important detail is that this appears to be a policy discussion, not a final rule. The FCC’s September equipment-authorisation decision did not include optical transceivers, after earlier reports had raised concerns about possible restrictions. There is also a separate bipartisan Senate bill targeting Chinese-made transceivers in sensitive federal national-security systems. That’s a procurement restriction, though not the same thing as a broad import ban. 🐊 But here’s the part I think the market needs to think about more carefully. If the reported 65% US-content framework is how the FCC ultimately approaches 3.2T, the key question isn’t necessarily: “Who replaces the Chinese transceiver makers?” It could be: “Who supplies the valuable components inside those Chinese transceivers?” If the test is based on the value of components supplied by US companies rather than where the module is physically assembled, a Chinese module maker could potentially remain the assembler while sourcing more of the DSP and laser content from US suppliers. That changes the thesis. The opportunity may not be US transceiver manufacturing replacing China. It may be US component suppliers capturing more of the economics underneath the module. 🐊 And that’s where I find the second-order effect more interesting. Think DSPs. Lasers. TIAs. Drivers. And separately, InP substrates. But I would keep the InP thesis separate: InP may be a critical capacity bottleneck, but it doesn’t drive the 65% BOM calculation. Morgan Stanley’s broader photonics work has also highlighted tight component capacity and the importance of InP supply. 🐊 The timing matters too. Even if policy action comes as early as October, that doesn’t mean a 2026 revenue explosion. The 3.2T ramp is still a future event. So I don’t want to chase a headline simply because photonics stocks are moving. For me, the real question is whether this policy becomes a catalyst for pricing power, qualification wins and incremental component content over the next few years. That’s a much more interesting investment question than simply asking which stock is up today. And it also changes how I look at $AAOI. I wouldn’t call AAOI a straightforward policy beneficiary based solely on the 65% framework. I’d look at it more through its capacity, qualification and ability to win share. The deeper thesis remains the same: Don’t just look at the company making the final product. Look one layer underneath. Because sometimes the biggest opportunity isn’t the company everyone is talking about. It’s the supplier whose component everyone suddenly needs. Not an investment advice.

Oppenheimer upgraded this nice looking stock to outperform with ~30% upside, but is this stock really that nice? Well, there was no pun intended, the name of ticker is actually $NICE, and $NICE is around $115–116, after a long de-rating driven largely by concerns that AI could disrupt its customer-experience software business. Oppenheimer just upgraded NICE to Outperform with a $150 target. That’s roughly 30%+ above the current price. But one analyst’s target isn’t the thesis. The numbers are. Q2 tells a mixed story. 🐊 Revenue was $782.3M, up 7.6% YoY, with cloud revenue growing 12.6%. But profitability deteriorated: 🐊 GAAP operating margin: 13.3% vs. 22.1% 🐊Non-GAAP operating margin: 25.3% vs. 30.2% 🐊 GAAP EPS: $1.40 vs. $2.96 🐊 Non-GAAP EPS: $2.70 vs. $3.01 So even the “clean” earnings number fell about 10%. And yet NICE guided to $11.06–$11.26 of FY26 non-GAAP EPS. At ~$115, that’s roughly 10x forward adjusted earnings. That looks cheap. But here’s the catch: The valuation assumes the earnings recover. That’s why I think the more useful question isn’t: “Is NICE cheap?” It clearly looks inexpensive on the company’s FY26 adjusted EPS guidance. The real question is: Why did margins fall, and do those pressures normalize? NICE is integrating Cognigy, the conversational/agentic AI platform it acquired, while also making targeted investments in its AI-native CXone platform. Management explicitly describes these as time-bound investments. That creates a very interesting setup. If these investments and integration costs roll off while cloud growth remains around the low-to-mid teens, the current earnings multiple could look unusually conservative. But if revenue growth stays around 7–8% and margins fail to recover, the low multiple may be justified. There is also evidence that AI isn’t simply cannibalising the business. AI ARR grew 66% YoY in Q1 2026, and NICE said AI was included in 100% of CXone enterprise deals. By Q2, AI ARR had reached $362M, representing 15% of cloud revenue. So my three checkpoints are becoming much more specific: 1. Cloud growth: Can NICE sustain 13–15% growth? 2. AI monetisation: Does AI ARR continue growing rapidly and translate into larger enterprise deals? 3. Margin recovery: Do operating expenses and integration-related investment normalise enough for earnings to recover? That’s the entire NICE debate for me. At ~$115, you’re not buying a clean growth story. You’re buying a potential margin-normalisation story at a low forward multiple while waiting to see whether AI becomes an accelerator rather than a disruptor. That’s much more interesting than simply saying “NICE is cheap.” I have opened a very miniscule “foot in the door” position and keep studying more and updating. Not financial advice.

MICRON $MU Q4 FY26 EARNINGS HIGHLIGHTS 🔥🔥🔥🔥🔥 🐊 Revenue: $54.23B vs. $51.07B est. 🟢 +379% YoY 🐊 Adj. EPS: $33.42 vs. $31.61 est. 🟢 🐊 Adj. Gross Margin: 87.0% vs. 86.1% est. 🟢 Q1 FY27 GUIDANCE 🐊 Revenue: $61.5B ± $1.5B vs. $57.02B est. 🟢 🐊 Adj. EPS: $38.15 ± $1.00 vs. $35.40 est. 🟢 🐊 Adj. Gross Margin: ~86.25% vs. 86.4% est. 🚩 CASH FLOW 🐊 Operating Cash Flow: $43.97B vs. $34.47B est. 🟢 BUSINESS UNIT REVENUE 🐊 Cloud Memory: $16.28B 🐊 Core Data Center: $18.00B 🐊 Mobile & Client: $13.11B 🐊 Automotive & Embedded: $6.82B OTHER Q4 METRICS 🐊 Adj. Operating Income: $44.64B vs. $42.41B est. 🟢 🐊 Adj. Net Income: $38.40B vs. $36.31B est. 🟢 🐊 GAAP EPS: $32.87 KEY UPDATE 🐊 Quarterly dividend: $0.15/share MANAGEMENT COMMENT ✅ “Micron delivered record fiscal 2026 results, and we expect an even stronger fiscal 2027.”

Micron Earnings Expectated in hours: Red Flags 🚩 or Green lights 🟢? My Micron position even after DCA-ing is on the way up, about 2x I've held $MU for a long time and been talking about it here for quite a while. With Earnings in few hours what are the things what will be red flags for me 🚩, and what would be gree signals I will be looking for 🟢? What is expected - Basics 🐊Micron reports FQ4 after the US close on 30 Sept. 🐊The guide is $50B ±$1B revenue, ~86% GM, $31 EPS. 🐊The Street sits at ~$50.4–51.2B and ~$31.1–31.6. 🐊A beat is the base case and largely priced. 🐊Options imply roughly an 8–10% move either way. The detail most will miss: 🐊FQ4 has 14 weeks, not 13. That extra week is worth ~$3.5B. 🐊FQ1 FY27 is back to 13 weeks, so a guide that looks flat on the headline is still ~8% growth per week. Peers already set the rule. 🐊SK Hynix printed a 76% operating margin and sold off on a consensus miss. 🐊 Sandisk beat with an 84.6% GM and fell on a guide ~2.5% light. 🐊Memory trades on the guide, not the print. 1. FQ1 guide 🟢 Revenue guided at or above FQ4's headline, which is real growth on one fewer week. 🚩 Below ~$47B (rounded up) . That means shrinking per week, not just a calendar effect. 2. Gross margin 🟢 ~86% delivered and FQ1 GM holding despite the lost week. The floor is doing its job. 🚩 FQ1 GM guided down meaningfully. The CQ2'26 price ceiling is starting to bind. 3. SCAs 🟢 More SCA volume committed, take-or-pay economics intact, and evidence that newer HBM generations can command premiums above existing-product ceilings. 🚩 No incremental commitments, weaker volume visibility, or evidence that customers are pushing back on pricing/terms. 4. FY27 capex 🟢 Disciplined, around ~$51B or below, weighted to HBM and construction. 🚩 Well above that. It reads as a real supply response for 2028. 5. Buybacks 🟢 A sized framework ahead of the CHIPS restriction ending 9 Dec. 🚩 Only "over time" language. That's mild, since management is conservative. 6. Consumer and China 🟢 PC and phone softness framed as mix shift, with the Apple business intact. 🚩 Unit declines accelerating, CXMT share gains cited, or Apple sourcing shifting. Supply chain backdrop 🐊TrendForce still has DRAM and NAND contract prices rising in 4Q26, led by AI servers and enterprise SSD. 🐊PC and phone buyers are at their limit, and LTAs are capping the pace. 🐊That's a plateau with support: not a crash, not a moonshot. Bear vs Bull 🐊Bear: CXMT keeps gaining share, consumer demand is being destroyed at these prices, and Burry swapped his short for June ~$500 puts, a bet on a 2027 turn rather than this print. 🐊 Bull: floor pricing management says sits above any prior cycle's peak margins, five-year take-or-pay contracts, and 2027 HBM pricing still negotiating up. My view The question isn't whether this quarter is good. It's whether Micron can show the cycle flattening into shipment growth rather than price growth. That's what earns a rerate, or doesn't. Not advice. I hold the stock.

Macros Today are acting like a bad Croc Starting with Bonds 🐊 The 10-year Treasury yield just hit 5.21% its highest level since mid-2007. 🐊 The 30-year is running near 5.48%, the highest since 2004. My fear is that when the long end moves further than the 10-year, it's usually a signal about long-term inflation and fiscal risk, not just a garden-variety rate move. This is a bearish stretch for bonds, plainly. Yields up means prices down, and three things are pushing in the same direction at once: 🐊 oil spiking toward $107 on renewed US-Iran tensions (inflationary), 🐊 one-year inflation expectations at a stubborn 4.6%, and 🐊 markets now pricing roughly 70% odds of another Fed hike at the October 28 meeting, 🐊 on top of the Fed's first hike in three years back on September 16. Layer on heavy Treasury issuance to fund the deficit, and you've got a genuine higher-for-longer, more-supply story not a one-headline blip. Gold’s reaction Gold's reaction is the more interesting tell. It's down nearly 4% in the same stretch during a geopolitical flare-up, when gold is supposed to catch a bid. I'd resist the temptation to read that as "capital rotating into Treasuries" that's a bigger claim than the data supports. The more defensible read: rising real and nominal yields increase the opportunity cost of holding a zero-yield asset, and that's currently outweighing gold's usual geopolitical risk premium. Impact on AI stocks Where this bites hardest is long-duration growth equities, AI names included but the mechanism matters. Higher yields raise the discount rate applied to future earnings, full stop; that's a valuation math problem, not a verdict on AI demand. The real question for AI stocks isn't whether demand holds up it's whether earnings can grow fast enough to outrun both a higher discount rate and the enormous capital intensity required to produce those earnings. That's a different, harder question than "is the AI trade over." What to prepare for: Continued chop in rate-sensitive tech and semis until yields find a ceiling; oil and Iran headlines as the near-term catalyst to watch; gold not reliably doing its "safe haven" job while real yields stay elevated; and a bias toward short-duration fixed income, pricing-power businesses, and energy exposure over long-duration growth names still leaning on future earnings that haven't shown up yet.

My three best positions in the trading portfolio: $MU, $TSEM, $SITM. Why: each owns a chokepoint in the AI data center build that the market was slow to price, with 🐊 $MU riding HBM-driven memory tightness, 🐊 $TSEM as the silicon photonics foundry behind 800G/1.6T optics, and 🐊 $SITM supplying the precision MEMS timing every faster link now needs. Not a financial advice.

Is $CGNX becoming more than a cyclical machine-vision company? The turnaround is real. Margins are recovering. AI could expand its customer base. And the $500M RealSense acquisition adds an intriguing Physical AI option. But at ~$59, how much of that future is already priced in? I break down the thesis, valuation, probabilities and my staged approach. stockcrock.substack.com/p/co…

A hidden beauty that I called out earlier has Doubled for me in less than a year with +127% YTD gains So I have written about this position earlier as well, in my detailed posts, and has doubled for me. In Seth Klarman’s book he says that many investors dont practice what they preach, and tells a story of how sardines for selling are so bad that sardine seller would not eat themselves. But I mostly do invest in things that I recommend. So this postition is $SMTC (Semtech), and stock is up today despite a sea or red. And the question I'm asking myself now isn't simply: "Is Semtech a good company?" It's: "Is the business improving fast enough to justify the valuation the market is now assigning it?" In other words should I sell now and reap the benefits? If I look at fundamentals ✅ Semtech's Q2 FY27 revenue was $341.9M, up 32.7% YoY. ✅ Data-center/infrastructure revenue rose 69%, with 800G strength and the early ramp of 1.6T FiberEdge and CopperEdge. ✅ Signal Integrity revenue jumped 64%. ✅ Industrial revenue grew 25%, helped by LoRa and IoT hardware. And it's not just trailing numbers: ✅ Q3 adjusted EPS guidance sits at $1.02–$1.08, and ✅ Zacks notes six full-year estimates were raised (none lowered) over the past two months, lifting consensus EPS from $2.66 to $3.48. 🐊 So what makes me bullish? This isn't simply an "AI stock" riding sentiment, the growth is already showing up across data-center and industrial revenue lines. If 800G → 1.6T → 3.2T continues to drive higher content per system, the connectivity opportunity could be larger than the market previously expected. 🐊 However, there is always a however, The risk isn't that Semtech lacks a growth story. It's that the stock may now be pricing in too much of it. Shares ran from $120.91 on Aug. 24 to $166.85 on Sept. 24 alone. Estimates are rising, but the question is whether they're rising fast enough to keep pace with how quickly the stock has rerated. I don't want to confuse a great business with a great investment at any price. Zacks recently upgraded SMTC to Strong Buy, but I'd treat that as a momentum/estimate-revision confirmation, not the core thesis. What matters now isn't my return from cost. It's my expected return from today's price. Not "sell because I'm up 100%." Not "hold because the story is working." That's the debate I'm having with my own position. The honest tension is this: a 48x multiple already assumes years of continued strong growth. If Semtech keeps compounding revenue in the 25-30%+ range and margins keep expanding, the stock can still work from here even after this move, because earnings will grow into the multiple. If growth decelerates toward more normal semiconductor-cycle rates, or if the 800G/1.6T ramp proves lumpier than expected, a stock at this multiple has very little room for disappointment and could give back a lot of the gain fast that's typical of high-multiple semis in a down quarter. I will “hold” it for today, but Mull over it and keep a close eye. If anyone has a diff view, would love to hear that. Disclaimer: This is not financial advice. I am not a financial advisor. This reflects my own research and portfolio, not a recommendation to buy or sell. Please do your own research before making investment decisions.

Two Major Developments - Muse & Jev … Two Different Tracks for AI - A Tale of Two Cities My AI posisitions are in deep green, and seems market is excited about a new development, but these developments remind me of my first few books from childhood ie A Tale of Two Cities by Charles Dickens Muse by $META 🐊Meta's new AI agent, launched Sept 8, books travel, manages your calendar, fills out forms, shops on your behalf. 🐊It hit 448,000 daily active users in 10 days (ChatGPT took 49 days to get there) and became the #1 free app on iOS. 🐊Meta stock jumped ~11% in a day, and the rally spread straight into chip stocks $ARM +17%, $Intel +12%, $AMD +10% because agents like this run continuously in the background, and that's a heavier, stickier compute load than a chatbot ever was. Jev This is a very different kind of model, from TypeSafe AI, built by Diogo Almeida (one of the researchers behind ChatGPT's RLHF). 🐊Jev doesn't generate text at all, it outputs calibrated probabilities for decisions. 🐊No hallucinations by design, 5-18x faster than chatbot-style models on classification tasks, and 10-20x cheaper. 🐊 It's built for the unglamorous, high-volume decisions software has to make constantly sorting, routing, flagging not for talking to humans. The impact of each, separately: 🐊Muse is a demand story it's proof that consumer AI is about to need more compute, not less, which is why it lit up the big hardware names. 🐊Jev is a cost story it's proof that a huge category of AI work (automated decisions, not conversations) is about to get radically cheaper, which is why smaller edge and automation-chip names (Texas Instruments, Analog Devices, ON Semi, STMicro) are the ones getting attention there. The bigger story is that AI is splitting into two distinct tracks. 🐊One track is "more compute, more capability" agents doing rich, open-ended tasks, which keeps justifying the massive capex buildout. 🐊The other is "same capability, near-zero cost" narrow, high-volume decisions getting commoditized to the point they can run on cheap hardware everywhere, including robotics and edge devices. Those aren't competing narratives they're the two halves of AI becoming actual infrastructure instead of a single hot product category. For us investors, that split matters more than either headline on its own. Treating "AI stocks" as one basket stops working here: 🐊Muse-type news favors the compute-heavy names (Meta, the GPU/chip supply chain feeding agents), 🐊while Jev-type news favors a completely different set of beneficiaries (analog/edge semis, automation infrastructure) that don't show up on most people's AI-exposure lists. Both stories are also still early Meta hasn't shown Muse revenue yet, and Jev's accuracy at scale is unproven so I would think that the smarter read is "figure out which of these two tracks a company's business actually sits on" before assuming AI-related news is good or bad for it.

One of my positions spiked +30% today in one day, and I realised some gains $NUAI, New Era Energy & Digital, spiked over 30% premarket on a Vistra power deal, then part of it faded most of the day, though I have taken the gain and closed the position, rrealising some gains. 🐊 Background: The company develops power-backed AI data center infrastructure on a 493-acre Permian Basin site with long-term potential toward 1.4 gigawatts, though only 207 megawatts is actually contracted today. It rebranded from New Era Helium in August 2025 and has since added a Chief Development Officer and General Counsel, though it remains a very early-stage developer. 🐊 The catalyst is a 20-year PPA for 200 to 207 MW from Vistra's Odessa gas plant, however, please note the power contract is not the same thing as a customer. New Era's actual data center tenant is still a non-binding letter of intent with Stream Data Centers, so revenue is several steps away. Why I sold? 🐊 The PPA itself is not free money. New Era must post a $116 million letter of credit plus up to $82.8 million in additional security, with conditions running through December 2027, so "firm power" comes with real near-term capital obligations. 🐊 Fundamentals still lag the story badly. Q2 revenue was $36,497 against a $20.4 million net loss, cash sits around $69.8 million, and management's own $25 to 30 million 12-month spending estimate could rise materially once a binding tenant deal is signed. 🐊 At roughly 106.5 million shares and today's price, equity value is around $625 million against essentially no revenue, so the real question isn't the day's move, it's how much of the future data center pipeline is already priced in. Look this was and is a high-risk, high-volatility name for risk capital only, and for some time I was under water, good to be back in green and I dont see a lot of upside based on fundamentals. Not personalized financial advice.

My Micron $MU and $DRAM positions are up again today, and the move looks driven by real supply-and-demand fundamentals rather than pure sentiment, arriving right before its fiscal Q4 earnings report. 🐊 AI demand optimism is lifting memory names broadly, and South Korean semiconductor exports hit record highs in the first twenty days of September, a strong read-through for the whole memory chain that Micron and SK Hynix sit in. 🐊 DRAM and NAND remain supply-constrained, which is giving chipmakers genuine pricing power into next quarter instead of just a hopeful narrative. 🐊 Micron reports earnings on September 30, and Wall Street has been highlighting favorable valuation multiples relative to the rest of the semiconductor space, with several banks reiterating buy ratings and raising price targets ahead of the print. 🐊 Traders are also speculating about a possible share repurchase announcement, which would layer a capital-return catalyst on top of the fundamental case. 🐊 Technically, it is a sector-wide rerating rather than a Micron-only story. Fundamentals, earnings positioning, and technical momentum are all pointing the same direction today.

Is the AI build-out slowing? Look at token pricing, not headlines. Since February, frontier pricing has moved up while mid-tier pricing has compressed. 🐊GPT-5.5 launched at $5/$30 per M tokens, double GPT-5.4. 🐊Anthropic added a new top tier at $10/$50, 2x Opus, 🐊while Sonnet 5 arrived 33% cheaper than Sonnet 4.6. Two conclusions follow. 1Demand isn't the weak link. Customers are paying up for frontier capability even as everything below it deflates. That points to a build-out paced by power, grid, permits, packaging and HBM, not by a shortage of buyers.
 2The model layer is bifurcating. Pricing power exists at the frontier, but the frontier keeps moving. Below it, capability is being commoditised.
 Where I'd hold: 🐊Bottlenecks: HBM/DRAM, advanced packaging, power/grid equipment, cooling. 🐊Inference volume: cheaper tokens mean more tokens. Falling model prices are a tailwind for silicon demand. 🐊Frontier leaders able to monetise a genuine capability lead. Where I'd be careful: 🐊Mid-tier model providers and API wrappers selling capability that keeps getting cheaper. 🐊Businesses whose margins assume today's token prices persist. 🐊Capacity plays without a structural bottleneck. Scarcity pays; surplus doesn't. The risk however is that if frontier progress slows, diffusion slows too, weakening the inference growth underpinning today's capex. Supply binds today, but demand elasticity decides whether it keeps binding. I would still own the choke points, and not the commodities.

Micron $MU thy beauty Many dont understand Micron, and misread the genuine need in the market, and despite its ups and downs Micron (MU) closed near $1,016 on Friday (Sep 18), up about 4% on the day after a choppier stretch that had briefly taken it down mid-single-digits on the week, as positioning builds into the September 30 fiscal Q4 print. Shares remain up several hundred percent over the past year on the AI-driven HBM/DRAM supply crunch, and its funny desp Analyst targets have been drifting higher into the print. 🐊 Consensus across roughly 45-49 analysts sits near $1,513 (Strong Buy), though dispersion is wide street-high $2,200, street-low $361. 🐊 Recent reiterations: TD Cowen's Krish Sankar at Buy/$1,600, and RBC's Srini Pajjuri at Outperform/$1,500 (Sept 18) both effectively flagging that a lot of the Strategic Customer Agreement (SCA) bull case is already in the price. Earnings will land on September 30 (fiscal Q4/full-year FY26). Micron's own guidance from the June print calls for 🐊record revenue of $50B +/-$1B, 🐊 ~86% gross margin, and 🐊 non-GAAP EPS of $31 +/-$1 (GAAP EPS ~$30.73) already well above the pre-guide Street numbers, so the real bar into the print is that guide, not stale consensus. 🐊 Options are pricing roughly an 11% swing either way post-report. Things I will be watching out: 🐊How much of the "structurally different this cycle" SCA thesis is already baked into a $1,500+ average target; 🐊Samsung's HBM ramp and CXMT's expanding China NAND/DRAM push on competitive share; 🐊margin-normalization chatter tied to Micron's own new capacity landing post-2027; and 🐊a still-live Netlist patent appeal as a smaller legal overhang. Note this is not a financial advice.

Opened a position in Deckers Brands $DECK today. The thesis is simple: this isn't a "one hit wonder" story. HOKA has become one of the fastest-growing names in performance footwear, and UGG has proven it can stay culturally relevant well beyond its original niche two brands, two different customer bases, both still taking share. What I like most is the balance sheet discipline behind the growth this isn't a company scaling revenue while bleeding margin to do it. In a market where investors are getting pickier about paying up for "story" stocks, that combination of brand momentum plus profitability is what keeps me interested. I'm treating this as a longer-term build, not a trade on this week's headlines adding on strength, not chasing. Not financial advice, just tracking my own thinking publicly. Curious if others are watching DECK here too.

I correctly predicted and have been saying that FED would increase rate, and not to my surprise, and honestly not to surprise of many, they did increase the rate by 25 bps. The Federal Open Market Committee voted unanimously 12-0 to raise the federal funds rate target to 3.75% to 4%, citing strong job growth, economic expansion, resilient spending, and persistent inflation linked to geopolitical tensions. But as I said earlier the more imp thing is Whar they say about future, and Officials aim to speed the return to their 2% inflation target, with projections showing one more 25-basis-point hike by end-2026 and rates above 4% through 2027. Now the most imp thing to measure in market reaction is 2yr vs 10 yr yield movement. Some of it was already expected as markets were choppy and dropped earlier but markets slipped as investors adjusted to the higher-for-longer outlook, marking a shift from earlier rate-cut hopes. Stay tuned I will share my read on 2yr vs 10yr market reaction and next steps for me as an investor later.

Everyone in the market have held their breath and are anxiously waiting for few things A) if Fed increases by 25bp - betting markets say 85% probability they would B) More imp, than hike what do they say about future C) More imp than both above how does market react in form of 10 yr / 30 yr bond yields D) More imp than all above, read market reaction closely, including reacton of 2 yr vs 10yr yields. And understand how market percieves Fed’s credibility and capability and broader macro environment. Given all of above, if I put my self in shoes of Fed chairman, what decision would I have taken? That would help me unpack the probability of (a) above a rate hike ( for b, c and d above refer to my detailed post over weekend). So as Fed Chair, here is how I read the table before tomorrow's decision: 🐊 Inflation is still hot, but its the composition matters more than the headline. The recent surge is concentrated in oil and a war-driven supply shock, not broad-based demand and no rate hike has ever produced a single barrel of crude. 🐊 The 10-year yield near 5% is not purely a growth signal. It reflects an inflation risk premium, heavy Treasury supply, and a rising term premium. A dovish surprise here steepens the curve further and pressures the dollar. ( Read my post for details) 🐊 Markets have priced roughly an 85% chance of a hike. As an incoming chair, my scarcest asset is credibility. Refusing to deliver what is fully expected, in my very first meeting, risks un-anchoring inflation expectations at the worst moment. 🐊 Political pressure to hold rates low is loud and real. But a central bank that visibly bends to it forfeits the independence that makes every future commitment believable. 🐊 Financial stability is a genuine constraint. Heavily indebted hyperscalers and leveraged balance sheets are exposed to higher rates, and I cannot ignore the plumbing. 🐊 The honest tension: tightening into an oil-and-war shock risks compounding the weakness the bond market may already be sensing. Conclusion: I would deliver a 25bp hike. This is less about the oil shock which I will explicitly look through and more about anchoring expectations and protecting a new chair's credibility. I would pair it with clear guidance that we distinguish supply-driven inflation from demand, and stand ready to pause the moment expectations stabilise. Doing nothing now, with a hike fully priced, would be the loudest dovish signal I could send and the bond market would punish it hardest. Beyond Conclusion I have said this earlier as well, and will say it again, as investor my role is not to be a prophet or economist, but to look at probabilities and be prepared to deal with difficult hand. The probabilities of rate hike are clear i) 85% rate goes up 25 bps ii) 15% it does not But 15% not going up will have a higher impact, vs the vice versa, albeit posititive impact that would pump the markets. But if (i) happens above then that needs to be coupled with bond market reaction as decribed in my detailed post. But again, am not FEd chairman, and this is not a financial advice.