$CIFR just made Barber Lake a much more valuable long-term asset. The site was originally contracted for 10 years with Fluidstack. Cipher has now secured a second 10 year commitment from another leading AI lab, which starts after the first lease ends. That adds roughly $5.2B of contracted revenue and takes Barber Lake to more than $9B of total contracted revenue over 20 years. Cipher will absorb the first $359.3M of cost overruns above the original budget; after that, the tenant reimburses 50% through additional rent. This does not suddenly boost near term earnings, but it improves backlog and lowers long term re leasing risk.

Added a bit more $AAOI this week. On trailing numbers, $AAOI looks expensive, but the valuation can change quickly if management executes on the 800G/1.6T ramp. Management is targeting $471M/month of DC transceiver revenue by mid-2027. Annualized, that is about $5.65B. If we add CATV, AAOI could be approaching a roughly $6B ARR. At today's roughly $8.4B enterprise value, AAOI would trade at only about 1.4x ARR. The table below shows what different revenue multiples would imply. I don't think AAOI automatically deserves the premium multiples of $LITE or $CRDO today. Its gross margins are around 30%, way below those businesses. If revenue approaches management's targets while margins expand, even a 2–4x sales multiple would imply a very different valuation.

Please help me understand this latest conspiracy. $NBIS is top pick because SA got paid, and not because the company deserves to be there? Repeat the phrase above and try not to laugh.

My $CIFR valuation framework for the next 12 to 24 months: Contracted only: $9-11/share Base case: $22-26/share Bull case: $35-45/share The upside is substantial, but it depends on execution. The 5.3 GW headline portfolio has value only as Cipher converts those MW into real contracts and cash flows. I break down the assumptions, financing, risks and valuation behind each scenario in my full $CIFR deep dive below.

I read recently how BofA remains confident that the memory super-cycle can extend through 2028–2030, supported by high ASPs, strong demand and continued supply tightness. I’ve been following the memory/storage AI story closely for more than a year, and one thing is clear: the industry is still being valued through the lens of its historical cyclicality. That is visible in the multiples. Most DRAM makers still trade at roughly 5–7x forward earnings, while storage names trade slightly higher but still very low. Despite the massive earnings growth, the market is clearly still asking the same question: When does the music stop? - Do ASPs peak? - Does demand peak? - Does supply eventually catch up? Those are all good questions. We all know that memory has been cyclical since the beginning, and I’m not going to argue that cyclicality suddenly disappeared. My argument is different: the structure of this cycle is changing. SCAs/LTAs are giving suppliers greater visibility and potentially extending the duration of favorable pricing and utilization. Also, memory and storage are no longer just components attached to the AI story. They are becoming part of the underlying AI infrastructure itself. Training needs memory. Inference needs memory. AI agents need memory. Physical AI and robotics will need memory. Data centers need enormous amounts of storage. Virtually every layer of modern computing depends on memory and storage somewhere in the stack. Will ASPs eventually come down? Of course. Will supply eventually catch up with a large part of demand? Almost certainly. But that does not automatically mean the outcome looks like previous memory cycles. The market may continue treating memory as a cyclical industry, and fundamentally it still is one. The entire economy is cyclical. But cyclical does not necessarily mean the cycle has to look the same every time. My thesis is not that memory has stopped being cyclical. It is that AI may be creating a longer, stronger and structurally more valuable cycle than the market is currently pricing in. I was and still am very bullish long term on: Samsung ($SSNLF) $SKHY $MU $SNDK Kioxia ($KXIAY) $WDC $STX I wrote about my take on Micron (article below, also free version via link in my bio) where I presented my case on what will happen with lower ASPs and lower gross margin, and what Chinese competitors can do to take away market share from KR/US producers.

I read recently how BofA remains confident that the memory super-cycle can extend through 2028–2030, supported by high ASPs, strong demand and continued supply tightness. I’ve been following the memory/storage AI story closely for more than a year, and one thing is clear: the industry is still being valued through the lens of its historical cyclicality. That is visible in the multiples. Most DRAM makers still trade at roughly 5–7x forward earnings, while storage names trade slightly higher but still very low. Despite the massive earnings growth, the market is clearly still asking the same question: When does the music stop? - Do ASPs peak? - Does demand peak? - Does supply eventually catch up? Those are all good questions. We all know that memory has been cyclical since the beginning, and I’m not going to argue that cyclicality suddenly disappeared. My argument is different: the structure of this cycle is changing. SCAs/LTAs are giving suppliers greater visibility and potentially extending the duration of favorable pricing and utilization. Also, memory and storage are no longer just components attached to the AI story. They are becoming part of the underlying AI infrastructure itself. Training needs memory. Inference needs memory. AI agents need memory. Physical AI and robotics will need memory. Data centers need enormous amounts of storage. Virtually every layer of modern computing depends on memory and storage somewhere in the stack. Will ASPs eventually come down? Of course. Will supply eventually catch up with a large part of demand? Almost certainly. But that does not automatically mean the outcome looks like previous memory cycles. The market may continue treating memory as a cyclical industry, and fundamentally it still is one. The entire economy is cyclical. But cyclical does not necessarily mean the cycle has to look the same every time. My thesis is not that memory has stopped being cyclical. It is that AI may be creating a longer, stronger and structurally more valuable cycle than the market is currently pricing in. I was and still am very bullish long term on: Samsung ($SSNLF) $SKHY $MU $SNDK Kioxia ($KXIAY) $WDC $STX I wrote about my take on Micron (article below, also free version via link in my bio) where I presented my case on what will happen with lower ASPs and lower gross margin, and what Chinese competitors can do to take away market share from KR/US producers.

To celebrate 20k 🎉 I'm offering 50% discount on annual subscription. I covered $CRDO $AAOI $MU $CIFR $WDC $PPIH $ACFN $PSIX $SNDK $OCC $KXIAY and soon $NBIS $APP and many more. Link in the comment. The offer expires in a week.

$CIFR is one of my favorite AI infrastructure stories right now: a former $BTC miner turning scarce power and land into long term data centre assets. The opportunity here is large, but so are the execution and valuation questions. I went through the business model, contracted portfolio, financing structure, industry setup, risks, and what the current share price is already assuming. Due to popular demand, I added the article on X. Enjoy, it's free! If you like it, please like/share.

Was talking to a friend last night about how I dislike $ORCL’s financials. I think the business itself is solid, but I told him I had no interest in buying the stock. Today, after the news and sell-off, I started a position. I might be right or wrong, but that’s not really the point. The point is, we should be flexible. New information came in, the price changed, and so did the risk/reward. Today I also sold $META, a company I believe in much more long term, to buy $ORCL, a company I rate lower fundamentally, because I currently see more upside in $ORCL over the short to medium term. Having conviction doesn’t mean becoming attached to an opinion. When the facts, price, or risk/reward change, your view should be allowed to change too. NFA

To celebrate 20k 🎉 I'm offering 50% discount on annual subscription. I covered $CRDO $AAOI $MU $CIFR $WDC $PPIH $ACFN $PSIX $SNDK $OCC $KXIAY and soon $NBIS $APP and many more. Link in the comment. The offer expires in a week.

I don’t like making things public here, but this guy has been taking shots at me nonstop. He keeps complaining that I charge for my Substack while he posts everything for free. Guaranteed value, right? Meanwhile, he’s lost most of his portfolio on a small cap, while I’m up 70% since the end of May and 314% YTD. Pick your winner. Choose wisely who you follow.

$CIFR is one of my favorite AI infrastructure stories right now: a former $BTC miner turning scarce power and land into long term data centre assets. The opportunity here is large, but so are the execution and valuation questions. I went through the business model, contracted portfolio, financing structure, industry setup, risks, and what the current share price is already assuming. Due to popular demand, I added the article on X. Enjoy, it's free! If you like it, please like/share.

$ORCL’s sell-off could be a buying opportunity and not because the risks disappeared, but simply because the news sound worse than than the near-term earnings impact they might cause. At roughly $135, $ORCL trades at: 17.6× adjusted run-rate P/E 16.7× FY2027 P/E 12.3× FY2028 consensus P/E The New Mexico force majeure notice reportedly seeks payment protection if Project Jupiter is delayed (not an exit). Oracle says the project remains on schedule and previously said it would not affect FY2027 guidance. If this stays an issue related to this project only, easing fears could cause a rebound without any other catalysts. What I don't like about $ORCL: ~$125B of debt, negative quarterly free cash flow and a recent $20B equity raise. IMO, a good buy at today's price if you don't mind the debt/risk.

$CIFR is one of the more interesting AI infrastructure stories right now: a former Bitcoin miner turning scarce power and land into long term data centre assets. The opportunity here is large, but so are the execution and valuation questions. I went through the business model, contracted portfolio, financing structure, industry setup, risks, and what the current share price is already assuming. Enjoy the read! (no paywall) Link in bio.

Speaking about bigger accounts throwing insults towards @SemiAnalysis_, I can say only one thing: When an argument is lost, insults and personal attacks become the default tool of the defeated. $IREN $SHAZ $WYFI $NBIS