Zag 🇨🇦

What’s wrong? I keep buying but the price keeps going down. $CIFR earnings Monday

Are we back? $CIFR @rftylerpage keep up the good work!

Hey $CIFR Did you finally wake up?

Overall $CIFR earnings were good. No bad surprises. First HPC lease revenue just hit ($5.9M from Black Pearl) and contracted backlog is now $16.6B across ~700 MW. Barber lake is starting rent. Whole AI data center is suffering. A really surprising amount of deleveraging in this profitable sector.

$CiFR barber lake renting‼️ ‼️ ‼️ $CIFR earnings call, what actually got said: • CEO: "we just delivered the RFS notice to Barber Lake for the first 51 MW of critical IT" — found out this morning. Closing line: "we are now collecting rent also at Barber Lake" • Black Pearl: 18 IT MW live, delivered early because the tenant asked for it • CFO: all three campuses "funded through completion," $724M unrestricted cash left at the parent • 3 of the 4 tightest-trading bonds among ~25 AI-infra high-yield deals are Cipher's. Credit market is voting with spreads • Colchis 1 GW: multiple top-tier tenants in diligence, final ERCOT classification December. CEO: "a real catalyst for an uplift in our valuation" • ~1 GW of on-site-gas data center leases in advanced talks, yields in line with grid sites • Barber Lake overrun put on the table: ~$409M change orders + $350–400M inflation. Cipher ate $359M for the 10-year extension, half the rest comes back as rent, yield-on-cost back to ~15% • On the water deal: "Together with Fluidstack and Anthropic." Colorado City is Barber Lake. Draw your own conclusion • Tyler, unprompted: "we are not a neocloud." Capital goes into buildings, not chips Takeaway: the bad news was fully quantified and the good news was buried in Q&A. Two campuses paying rent as of today, three builds funded, 1 GW lease on deck. Market still pricing the construction site.

$CIFR earnings out

$cifr We will find out tomorrow whether this was the best or worst buy

The ritual tonight $CIFR

Signs that $CIFR earnings will come out good: •No postponement. Date was set Oct 12, call is 8am ET Monday. Companies that are about to disappoint find reasons to move the call. •No pre-announcement 8-K. •$850M convert priced Oct 14, settled Oct 16, two weeks after quarter end, with an offering memo. Signs that earnings will be garbage: •The market has determined anything AI is worthless

Why I think $CIFR is relatively undervalued 1. What’s actually being sold this week The AI selloff is about capex and repricing: $AMZN’s cash burn hit a record, $GOOGL’s FCF went negative, $MSFT’s capex came in light. 2. The names that should get hurt are the neoclouds: $CRWV, $NBIS, $IREN. They borrow at 12-15%, buy GPUs that depreciate in 5 years, and rent them to 2-3 customers. 3. The names DO NOT apply to powershell companies ($CIFR, $WULF, $HUT, $APLD). They owns land and power. The tenant brings the chips. The lease runs 15-20 years. 4. $CIFR has ~$11.4B contracted revenue, 700 MW, 3 campuses: - Black Pearl: 300 MW, 15-yr, $5.5B, $AWS direct. First capacity delivered Aug, 2 - Barber Lake: 300 MW, 20-yr, ~$9B, Fluidstack with $GOOGL guaranteeing rent. Halls delivering Q4-Q1. - Stingray: $AWS, fully funded at a 6.0% coupon. 5. $CIFR value of pipeline is virtually 0. This is the ~3.3 GW of Texas pipeline across 11 sites. 6. $CIFR stumbles with - Barber Lake construction and overrun for budget $2.16B → ~$3B on tenant design changes, but gets lease extension for additional 10 years and +$5B contract value. - Oct 16: $850M zero-coupon convert to fund it. 7. After all of this, look at Market cap per contracted MW: Sep 30: $CIFR $9.4M · $WULF $8.8M · $HUT $11.4M Oct 30: $CIFR $4.4M · $WULF $7.0M · $HUT $7.4M Parity a month ago. 40% discount today. $CIFR back to $WULF’s multiple is ~$12-13/share. Consensus is $28. Cantor’s cut target is $19. This assume $CIFR can deliver as expected on Nov 2 call.

The AI beatings will continue until morale improves

$cifr Doing my part!

$MSFT capex came in at $45.2B against a guide above $50B, with the CFO saying some third-party capacity deliveries expected in the quarter were delayed; next quarter is guided back above $50B. $GOOGL spent $51.4B, kept 2026 at $195B to $205B and said 2027 rises significantly. $AMZN spent $75.9B, held 2026 at about $220B, said 2027 will be meaningfully more, and AWS backlog is $578B. Memory: Samsung's chip division made 106.5T won operating profit at a 72% margin, HBM4 sales tripled quarter on quarter to about 60% of HBM revenue, DRAM prices rose about 20% in the quarter, 2027 HBM is sold out, and it expects shortages through 2028. SK hynix posted 104.6T won revenue and 83.3T won operating profit, an 80% margin, with 14 long-term customers covering over half of 2027 to 2028 capacity. Samsung's net profit missed and SK hynix closed down 7.1% in Seoul on Friday. $MU Neoclouds: $IREN closed a $363M equipment financing on GPU servers backed by multi-year contracts, arranged by Wingspire with Eldridge and Silver Point as anchors. $NBIS reports Q3 on Nov 12, $CRWV on Nov 10. Microsoft is CoreWeave's largest customer, so the delivery delay comment landed on the whole group. Watchlist: B. Riley cut targets across the HPC group this morning, $HUT to $134 from $163, $WULF to $26 from $40, $APLD to $51 from $75, $CORZ to $31 from $42, and $COHR to $262 from $297. $CIFR reports Monday, $HUT Wednesday, $WULF next Friday. The $6.36B Binance hack overnight hit crypto-linked names as well. Photonics: Bloomberg reports $MRVL is in advanced talks to buy $LITE, which carries roughly an $83B market cap, with an announcement possible next week. ASE said CPO volume production starts this quarter and guided over $500M of CPO revenue in 2027. Innolight grew Q3 revenue 172% to RMB 27.8B while Eoptolink's revenue fell 13% quarter on quarter and its inventory rose 141% year on year; Chinese module stocks fell 8% to 20% on Friday. $COHR reports Nov 4. Nothing sounds too terrible by that’s not what my port says

Seems like all stocks dumped despite earnings. Should we all buy puts for companies with earnings next week?

Blood in overnight markets

$CIFR looking cheap relative to the backlog. Barber Lake just went 20-year / >$9B contracted (was $3.8B). Fluidstack + Google backstop for the first 10 years then direct AI lab. Black Pearl (AWS ~$5.5B / 15 yrs / ~300 MW) already collecting rent — 2 months early. Avg annualized NOI ~$793M starting late 2026. Why it screens better than the alternatives right now (indexed to market cap): CIFR ~$3.5B mcap is sitting on ~$16.6B contracted vs IREN ~$11.1B mcap on ~$13.8B and WULF ~$6.4B on the $19B Anthropic deal. That works out to ~4.7x contracted revenue per dollar of market cap for CIFR vs ~1.2x for IREN and ~3.0x for WULF. Near-term MW is even more skewed: 200 MW per $B mcap vs 43 and 63. Pure landlord model. Tenants bring the GPUs so less hardware risk and lower capex than the cloud guys. Longer 15-20 yr terms with real credit support. Texas power + 2.5 GW BTM gas optionality. 700 MW contracted HPC, ~5.3 GW total portfolio. Black Pearl already live.