Bullpen.News From Edwin Dorsey
Get Daily Recap Feedback

NASDAQ

IREN Ltd

IREN

27 stories

$11B Market cap · 2026-10-21

@StableBread 1 click

Fajasy on X

Here’s what Boost Run $BRUN has already committed to pay, and what it still has to finance in H2 2026. $BRUN’s GPUs need space and power, which it rents in other companies’ data centers (colocation). The four leases signed since April already cover the space and power for the year-end deployments, and CEO Andrew Karos said its partnerships now reach 253 MW of power. $BRUN owes the rent on those sites whether or not the GPUs get financed. In 2027, the commitments $BRUN has already signed call for ~$221M of payments, including $103.5M on its GPU leases, $28.9M of rent on the sites in use as of June 30, ~$69M on the four new leases, and $20M for software. $BRUN’s $191.8M of pro forma cash and $93.6M of contracted prepayments, $285.4M together, cover the ~$221M of 2027 payments before counting any monthly billings from customers. But the same cash also pays $BRUN’s share of the H2 2026 build, so its monthly billings will have to cover part of the 2027 payments. To reach the $1.0-1.2B of GPU spending management expects for 2026, $BRUN has to deploy $663-863M of GPU servers in H2 2026, 2.3-3.0x its cash and contracted prepayments. If lessors fund 79.5% of the H2 2026 build, as they did in H1 2026, $BRUN needs $527-686M of new leases, 2.2-2.9x the $238.1M the company owed on finance leases as of June 30. The other $136-176M would be paid in cash. The $527-686M works out to $264-343M/quarter on average across Q3 and Q4, 1.3-1.6x the $208.2M of leases $BRUN took on in Q2. $BRUN’s lease debt would rise by less than the $527-686M, since the company keeps paying down its leases while adding new ones. During H1 2026, its finance lease debt rose $207.7M on $337.3M of new GPU equipment, or 61.6%, which for H2 2026 would imply ~$410-530M more debt. New leases will likely cost more than ~7% after the rise in rates. For context, at the 9.0% $IREN agreed to pay on its August GPU financing, the extra interest on $527-686M would be ~$6.5-8.5M/year, 1.6-2.1% of the $400M ARR target. And $BRUN isn’t slowing down. By June 30, the company had already placed orders well above the $578M first-year minimum in its April 2026 agreement with Dell $DELL, which also requires at least $215.5M/year of Dell equipment from April 2027 to April 2031 or a payment for the shortfall.
@convequity

Convequity on X

Neoclouds Now Trade as a Hedge Against AI Demand Growth Convequity’s latest ABBX review shows $CRWV, $NBIS and $IREN are inversely correlated with AI demand growth. Stronger demand widens the working-capital hole and keeps investors risk-off, even as order books expand. Insatiable demand is usually the reason to own the supplier. Here it does the opposite. Jensen Huang told CNBC at GTC Berlin that neocloud demand is “off the charts” and that working capital is the missing piece. Working capital is current assets minus current liabilities. A cash-conversion mismatch squeezes it even when demand is huge. Liabilities hit first. A neocloud pays $NVDA and networking suppliers in roughly 30–60 days. Asset-backed hardware loans add heavy near-term debt service. Assets lag. A signed customer contract sits off the balance sheet as backlog. It only becomes receivables or cash once the cluster is deployed, energized and billed — months later. Cash leaves before revenue arrives. A faster pipeline simply widens the short-term deficit. Latest ABBX figures NBIS: $8.5B gross debt vs $8.0B cash; guidance for >$9B customer prepayments this year. IREN: $7.6B debt vs $5.9B cash; $13.8B hardware and site commitments; 20% prepay on the Microsoft contract. CRWV: $35.6B debt vs $5.5B cash, plus $16.3B operating leases and $35.5B rent not yet commenced. Build-out relies far more on debt than customer prepayments. Even with those prepayments, NBIS and IREN keep raising large extra funding. CRWV faces the same hole with a far larger debt stock. Handing contracts back to resell at higher spot prices would ease CRWV’s cash hole only if customers and lenders agreed. Customers pay even if they do not use the capacity, CRWV must deliver it, and we found no public right to change the price or exit early. Break a customer contract and the hardware lenders can lock up CRWV’s cash. If that lockup holds, they take the cash to repay the loan. Borrowing later would cost more, and the switch only pays while spot stays high. If spot fell, CRWV would lose both the committed revenue and the contract value backing those loans. For investors to turn risk-on again, AI demand growth needs to slow enough for infrastructure supply to catch up.
@convequity 3 clicks

Convequity on X

$CRWV 5-year CDS hit a record ~1,195bp on 15 Oct and the whole neocloud group sold off with it. Our ABBX tracker suggests the market is pricing contagion into the wrong names. The balance sheet first. At 30 June CoreWeave had $35.6B of debt against $5.5B of cash, and its ~$33B market cap now sits below its debt. $4.4B falls due in the rest of 2026 and $6.2B in 2027, mostly loan amortisation; no bond matures before 2030. Its 9.625% notes sold at par in June now yield ~14.9%. Add $16.3B of lease liabilities and $35.5B of leases not yet started. CoreWeave's Q2 operating cash flow: Net loss: −$0.63B + Depreciation & amortisation: +$1.39B + Stock-based comp: +$0.17B + Other non-cash items: +$0.28B − Working capital (payables, receivables, prepaids): −$1.32B + Customer prepayments: +$0.79B = Operating cash flow: $0.68B The core business makes cash, but working capital ate it. Without the prepayments OCF was −$0.11B. That OCF is already after ~$0.44B of cash interest, equal to 65% of it, and capex ran 9.5x OCF. Funding is where the group splits. CoreWeave builds with borrowed money: GPU-backed loans, 8.5–9.75% bonds and vendor credit, with ~$9.8B of H1 equipment supplier-financed outside the cash-flow statement. Prepayments covered 12% of its Q2 capex. $NBIS builds with convertibles (~$9B this year) and prepayments that covered 21% of Q2 capex; at 30 June it held $8.5B of debt against $8.0B of cash. $IREN builds with equity, a $3B 1% convertible and prepayments that covered 87% of its latest quarter's capex; net debt is ~$1.7B and cash interest equalled ~1% of OCF. The exposure that matters is its landlords. CoreWeave is ~77% of $CORZ revenue, the tenant at Galaxy's Helios campus and leases 400MW from $APLD. Yet since 2 Oct Core Scientific is down 17%, while Microsoft-anchored IREN is down 34% and Nebius 30%. Our view: the risk is CoreWeave's debt-funded build, not the neocloud model. If CoreWeave stumbles, the hit lands on the landlords that depend on its rent. Nebius and IREN share the model, not the funding.
@StableBread 1 click

Fajasy on X

From June 1 to October 16, Boost Run $BRUN fell 77.3%, vs. 37.4-58.7% for CoreWeave $CRWV, Nebius $NBIS, and $IREN. That drop in AI infra stocks is tied to to rising bets on another Fed rate hike. Higher rates hurt GPU clouds in two ways, since they (1) raise the cost of the debt that pays for their GPUs and (2) lower what investors will pay today for profits that arrive years later. But all three peers kept or raised their 2026 targets: → $CRWV: Raised its 2026 revenue guidance to $12.4-13.2B on its Q2 call. → $IREN: Raised its year-end ARR target to more than $4B on July 20. → $NBIS: Reaffirmed its 2026 guidance with its Q2 results. The problem is that lenders are charging more now. CoreWeave’s 9.625% notes due 2032 yielded 14.74% on October 16, up from 13.19% on October 1, meaning CoreWeave would likely pay ~15%/year to borrow the same way today. So the selloff came from how investors now see the cost of funding the demand for GPU clouds, not from the demand itself.
@StableBread 11 clicks

Fajasy on X

I first wrote up an AI data center company that rents out $NVDA GPUs to AI startup companies on April 15, 2026, when it still traded as a SPAC at $10.70/share, just above its ~$10.60 trust value. The merger was a success, and the stock peaked at $39.20/share on June 1st. Now it trades below its SPAC price. That company is Boost Run $BRUN, which has fallen to $8.89/share, down 77.3% from its peak, even after it announced its largest disclosed contract on October 6 (worth $525.6M). The three larger listed GPU clouds have fallen from their 2026 highs as well, through the October 16 close: → CoreWeave $CRWV: Down 55.5% to $61.39/share. → $IREN: Down 60.2% to $26.99/share. → Nebius $NBIS: Down 42.3% to $165.47/share. So $BRUN fell with the group, but further than all three. I’d blame the drop from $39.20 on two causes that have nothing to do with $BRUN’s core operations: (1) higher rates and borrowing costs, and (2) $BRUN’s float growing ~3.1x, which was expected once its lockup ended and its warrants were called. Yet the business carries less risk than it did as a SPAC: → Contracts: $1.61B with disclosed terms and TCV over $2.6B, including $1.43B signed since my April write-up. → Customers: Thinking Machines Lab (the AI startup founded by former OpenAI CTO Mira Murati) and Cohere make up 61.9% of the $1.61B on three- to five-year terms, after three customers made up 75.9% of 2025 revenue, led by RunPod, a GPU cloud platform, at 44.5%. → Year-end target: If Thinking Machines’ GPUs are fully running by December 31 and the June order starts on November 30, the five disclosed contracts running by year-end cover $300.6M (75.1%) of the $400M ARR target. → Merger: Closed on May 8, 2026, with zero redemptions. → Balance sheet: The merger proceeds paid off its short-term bridge loans, the going-concern language is gone, and pro forma cash is $191.8M, up from the $108M I used in my June 3 update, right after the merger. But $BRUN still has to get the GPUs bought, financed, and running by December 31: → To reach the $1.0-1.2B of GPU spending management expects for 2026, $BRUN has to deploy $663-863M of GPU servers in H2 2026, 2.3-3.0x its cash and contracted prepayments. → CFO Erik Guckel said on the Q2 call that $BRUN is working on new ways to finance the build, but I found no new facility, lender, or term sheet in its filings or press releases from August 14 to October 16. → $BRUN’s named customers are private AI startups without investment-grade ratings, so its contracts are harder to borrow against. → If lessors funded 70% of the H2 2026 build instead of 79.5% (as in H1 2026), $BRUN would need $199-259M of its own cash for the build, more than its $191.8M of pro forma cash. → Both 10-Qs came late, after notices of late filing on May 26 and August 14, and $BRUN still reports three material weaknesses, the most serious kind of gap in a company’s financial reporting controls. → Founder and CEO Andrew Karos controls 87% of the vote. At $8.89/share, $BRUN’s EV is $989.7M. That’s 2.5x its $400M target for year-end ARR, down from 5.4x at the $22.93 price in my August 16 update. At 2.5x, $BRUN trades below all three peers, including $IREN at 3.2x. So did the thesis break, or just the stock? That’s what I answer in my new ~5k-word update on $BRUN, including my bear, base, and bull cases (Excel model included).
@aleabitoreddit

Serenity on X

Feels like the public interpretation is all wrong… I’m pretty sure $MSFT backing out on NScale deal is bullish for Neoclouds like $NBIS or even $IREN with energized capacity with Microsoft? Pretend Microsoft just ate a bunch of Chipotle burittos: - They order toilet paper delivery scheduled Q1 2027 - NScale says: sorry 3-4+ year delay of the toilet paper (into 2030’s) because of UK grid constraints - Microsoft still urgently needs toilet paper so they will get it elsewhere The cancellation does send warnings about what happens if a Neocloud can’t execute on contracted demand. But that just makes the other toilet paper sellers more valuable since Microsoft will go elsewhere for supply.
@aleabitoreddit 2 clicks

Serenity on X

Sure... Random train of thoughts on markets now, optics like $LITE / memory + $IREN / Anthropic. On Memory: - Demand visibility on $MU / Samsung / SK Hynix extend into 2030-2031. Samsung has LTAs covering 70% of their capacity into 2031 (likely with SK Hynix type price floors)... and their operating profit is already absurd. There's a consensus that shortage in 2027 will be much more severe than 2026... Even $SNDK expends gross margins to exceed 80% entering 2030. You're going to hear Acer and other CEOs that want memory prices to go down cite bear cases like CXMT/YMTC, and Chinese players. But just looking at supply procurement that $NVDA is doing + hyperscalers, demand visibility is absurd for the next 5 years. And by that time, cumulative FCF coming in from these take/pay + pricing floor contracts might exceed some of their entire marketcaps so it feels like there's some baseline. Near-term is always volatile (IV dropped a ton across $EWY to $DRAM to $MU), so people with call options likely got nuked, along with underlying prices. But if Samsung is 3x 2027E, has 5 year LTAs with 70% of capacity committed from hyperscalers, and price floors, it's kinda hard to see these names suffer a sustained crash... For optics: - $LITE is literally sold out into 2029, starting discussions into 2030. With demand visibility lasting into 2031… with hyperscalers largely funding the capacity buildout. Their 2028 Greensboro fab expansion was ~$5B in UHP laser capacity revenue alone, yet it’s basically allocated toward $NVDA (50%+ with options for more) and co from channel checks. Existing supply chain bottlenecks are extremely severe as optical content/XPU skyrockets entering 2027-2028. So everything from DSPs/TIAs, photodiodes, lasers, InP/SOI substrates, wafers (epi, SOI, etc) are expected to go brrr. And what does that imply to anyone else with enormous amounts of capacity coming online like $AAOI + $SIVE + $MTSI capacity expansion in 2027? Or other players like Sumitomo / Furukawa? But again it's not quite the same situation as memory since the BOM % is extremely low in a cluster... If micron/samsung raised another 2000% hyperscalers would go bankrupt. But doing that for InP substrates seems possible. Or a 100% increase on CW laser content. I actually thought $LITE was overvalued at $100B+ MC, relative to capacity expansion on your smaller optical players to their respective MCs. So I chose to position accordingly (But lumentum does deserve premiums for demand visibility). For the companies leading the charge like $NVDA / $AVGO / $MRVL: - Nvidia has 70% forward revenue growth FY2028 which is supply constrained growth. - Broadcom 100% Y/Y for 2027/2028 which is also supply constrained growth. - Marvell FY2031 upgraded guidance to $70B-$90B from $46B analyst projections... They all seem to be very happy.... and read through on demand for next 2-3 years is enormous (also echoing what $INTC CEO said today on the panel with $AVGO CEO). So you have overwhelming growth on memory + photonics + from the leaders... But the nuance is not everything in the AI supply chain get the same economic benefit. HBM/Laser/CPU/MLCC and other players in the bottlenecks look to be printing money. And seem to be the largest beneficiaries of the current bottlenecks. Test players are also getting price hikes + capacity allocated by players like $NVDA, but economic impact seems a tad lower than other players since forward P/E for many I modeled were still in the 45x+ range even if you hiked prices like 10-30%. As for capex intensive players... there's some like $IREN ($25-30B for 2027) that despite a deal with Anthropic ended up dropping. I don't think markets seem to appreciate players spending a lot of money + ongoing financing (think they're like halfway through their $6B ATM). Just some hopium for $IREN investors, $NBIS also signed a deal with $META at the start of the year (but markets didn't care at all, since Nebius dropped to ~$70-80 at the lowest point?). But did 6 months later, as the time to revenue inflection gets closer and you're past the ATM/capex window, markets would likely care more. I don't think it's irrational but at the same time, the fundamentals have improved from Mirantis acquisition to others. I also think it was known for awhile it was either OpenAI/Anthropic from CEO comments earlier too? But either way to be a more risk-off environment where funds seem to be rotating to larger names as a flight to safety. $CRWD / $PLTR as just two example, but my take is that the valuations of these rotations are reaching clown levels. And that certain AI supply chain beneficiaries in comparison are undervalued now. I was thinking markets would just rotate to names that are extremely profitable like Samsung or names that have <2X P/E, so it was surprising to see $KOSPI and some legacy memory players get sold off today. Since all of this is colliding with uncertainty about sht macro environments and renewed sustainability concern of AI capex/commitments from frontier labs like OpenAI/Anthropic from due to that FT piece recently. I've always been sounding the alarm about the contractual backlog OpenAI has with $CBRS, $MSFT, $ORCL and others earlier in the year. But OpenAI did just raise $122B or so , so I don't quite think they're at risk of going down anytime soon. I also don't think overall capex can just exponentially go up, but even if it stalls / hovers around a certain point or even drops... Capex toward certain areas like optics will still explode as hyperscalers fight over allocations when CPO scale up comes about, but might drop in others where more capacity comes online like test or DRAM. So it's a lot more nuanced than capex down = everything down or vice versa. I also slightly agree with community agreements that open weight models wouldn't lessen demand on AI infrastructure (otherwise $NVDA wouldn't be funneling funds into Reflection), but all the contractual obligations that Anthropic has might get messy if their models/company get cannibalized. Either way, I'd expect some of these frontier lab valuations to likely compress but I do think Jensen knows what he's doing though to avoid contagion. But basically: - Extremely difficult market now, can't just throw money at any AI supply chain company like $GLW and expect it to go up. There's cause for concern of rate hikes mixed with financing per $AVGO CEO but not to the point you just blindly sell off everything. Also I think markets are stupid and ties baskets together so that leads to mispricing + individual buying opportunities with certain chokepoints too. If capacity + demand is visible and forward p/e compresses like $AAOI to single digits off hundreds or thousands of percent growth from hyperscalers off $471m transceiver revenues and hundreds of thousands of monthly ELSFP capacity. Or $SIVE coming online with 100M+ laser capacity + multiple foundries with ASP hikes on lasers. I wouldn't lump them into other AI supply chain names in IP like $RMBS or others. My take is that AI supply chain beneficiaries with closer inflection points, demand visibility, and forward P/E compressing due to economic leverage would likely outperform medium term despite current volatility. TLDR: - Memory strongest risk/adjusted return with LTAs + pricing floors extending to 2030's + buybacks. - My opinion is that optical supercycle for 2027-2028 with CPO/NPO is going to look absurd relative to capacity coming online of smaller suppliers. - Overwhelming fundamentals/growth colliding with sht macro + media scares - Markets are stupid and groups things in baskets and sells them all off together or buys overvalued names, creating short/long opportunities I'd give a prayer to people on short term options given underlying drop + IV decline, but there could be a reversal... who knows. Most valuable thing right now is time.
@convequity

Convequity on X

The Information reports that Anthropic has an unannounced multiyear cloud contract with $IREN. Our view is that it is most likely a contract IREN has already announced and already counts. On 27 August, in its FY26 results, IREN said it had signed a new multiyear AI cloud contract with an unnamed "leading frontier AI lab". In the same release it put contracted annualised run-rate revenue (ARR) for its 2026 capacity at $4bn, against $1bn of ARR operating on 26 August, and said its 2026 capacity was largely sold out. The Information gives no size or term, and neither IREN nor Anthropic has confirmed the deal. If Anthropic is the August lab, the report names a customer but adds nothing to the $4bn. It would only be new revenue if IREN raised that figure, and IREN has not said it will.
@convequity 6 clicks

Convequity on X

OpenAI and Anthropic will rent most of the AI capacity they plan, but mainly from the big clouds rather than the listed neoclouds Convequity's AI Bubble Barometer (ABBX) tracks the data-centre capacity each company plans and how much of it it expects to rent from partners rather than own. 1. OpenAI plans 21.2 GW, of which 18.0 GW is to be supplied by partners. Anthropic plans 13.8 GW, all of it from partners. Together the two labs plan to rent 31.8 GW. 2. That is 2.8 times the 11.2 GW that the five hyperscalers, $MSFT, $AMZN, $GOOGL, $META and $ORCL, plan to rent from others. 3. The labs' named suppliers are mostly AWS, Oracle, Google, Azure and SB Energy. CoreWeave, a listed neocloud, and Nscale, a private one, appear on the lists, but the bulk of the 31.8 GW is not with listed neoclouds. 4. The seven listed neoclouds, $CRWV, $NBIS, $IREN, $APLD, $CORZ, $CIFR and $BTDR, plan to build 33.62 GW of their own. Much of that is gross site power rather than contracted IT capacity. 5. The labs are also behind most of the recent growth in AI capacity plans. Since 14 Jul, the total capacity that the 15 ABBX companies say they plan to have has risen by 22.19 GW, from 131.23 GW to 153.42 GW on 7 Oct. OpenAI and Anthropic raising their own plans accounts for 13.7 GW of that rise (62%). We think the labs' rental demand mainly supports $AMZN, $ORCL, $GOOGL and $MSFT. The neoclouds will need to fill their pipelines largely with other customers, including the hyperscalers' own 11.2 GW of planned rentals, so we would not read the labs' plans as demand for the neoclouds.
@convequity 4 clicks

Convequity on X

Neoclouds trade at 2.6x invested capital, less than half the hyperscalers' 6.1x The seven neoclouds in Convequity's AI Bubble Barometer (ABBX), $CRWV, $NBIS, $IREN, $APLD, $CORZ, $CIFR and $BTDR, lost 7.2% of their combined enterprise value this week, from $190.2B at the 2 Oct close to $176.5B at the 9 Oct close. Over the same week the five hyperscalers, $MSFT, $AMZN, $GOOGL, $META and $ORCL, gained 2.4%, from $13,316B to $13,637B. 1. Enterprise value divided by invested capital (EV/IC) at the 9 Oct close was 2.6x for the neoclouds and 6.1x for the hyperscalers. At the 2 Oct close it was 2.8x and 6.0x. 2. ABBX estimates the weighted average cost of capital (WACC) at 14.6% for the neoclouds and 9.7% for the hyperscalers. 3. At the 9 Oct close, neocloud share prices were 20% to 45% below each stock's best Friday close since 17 Jul: CIFR −44.5%, CORZ −33.3%, BTDR −25.8%, IREN −24.6%, APLD −24.1%, CRWV −22.0% and NBIS −20.4%. Among the hyperscalers, ORCL was 10.9% below its best, AMZN and META 4.4% and GOOGL 1.3%, while MSFT closed above its best. We think most of the discount reflects the neoclouds' higher cost of capital rather than a mispricing. We would not call 2.6x cheap until they report their July to September backlog and customer prepayment figures.
@convequity 1 click

Convequity on X

Despite investor doubts, $IREN is worth a small position. The shares have fallen about 17% from their 22 Sep close. The equipment IREN had ordered at 30 June is covered by the cash, committed GPU financing and prepayments it reported on 27 Aug. In the June 2026 quarter customers funded most of its capex. On signed contracts the shares are priced about where $CRWV is. The main risk is that IREN runs GPU clouds worse than its peers. 1. IREN guided FY27 capex (Jul 2026–Jun 2027) at $25–30B on its 27 Aug results call. Its enterprise value is $14.6B at the 5 Oct close. IREN's FY26 results show revenue of $0.7B, of which $0.58B was bitcoin mining. IREN expects to have shut mining down by the end of December 2026. 2. At 30 Jun 2026 IREN had $13.8B of commitments for GPUs and site work, almost all payable by 30 Jun 2027, per its annual report. Orders placed after 30 June are not in that figure. On 27 Aug IREN put its existing cash, committed GPU financing and customer prepayments at about $14B. That covers the 30 June commitments. 3. The other $11–16B of the capex guide is not yet funded. Management targets about $8B more in GPU financing and prepayments for new GPU capacity, and that funding comes with new customer contracts. IREN says it is in late-stage talks for a large share of 2027 capacity. It has not announced any signed contract for that new 2027 capacity. The rest would come from data-center debt, operating cash flow and corporate funding, which includes selling shares. By 14 Aug 2026 IREN had sold about $2.5B of shares under its $6B ATM programme, leaving about $3.5B at that date, per its annual report. It has not reported any sale since. 4. On signed contracts, IREN is priced close to CRWV. On our ABBX dashboard, enterprise value is 0.88x contract value for IREN, 0.74x for CRWV and 1.60x for $NBIS. The contract values are company-reported, and each company measures them differently. IREN reports about $16.6B, of which about $5.1B is cloud service contracts and about $11.4B is GPU lease contracts. CRWV reports a $104.2B revenue backlog. NBIS reports $37.5B of remaining performance obligations. On 27 Aug IREN said its 2026 capacity was largely sold out. It had $4B of contracted annualised revenue run-rate (ARR) for 2026 capacity, which it earns once that capacity is delivered and accepted. $1B of ARR was already operating. EV is about 3.7x the contracted $4B. 5. In the June 2026 quarter, IREN's customers funded most of its capex. On our ABBX dashboard, customer prepayments and deferred revenue covered 87% of IREN's cash capex in that quarter ($1.72B of $1.98B). The same measure was 21% at NBIS and 12% at CRWV. Working-capital inflows covered 90% of IREN's capex and 36% of NBIS's. CRWV's working capital was an outflow equal to 8% of its capex. More than half of IREN's customer prepayments came under the $MSFT contract. At 30 Jun 2026, $1.01B of IREN's $1.84B of deferred revenue sat in the subsidiary that holds the GPUs for the MSFT contract, per its annual report. IREN said on 27 Aug that $3.6B of GPU financing plus customer prepayments fund about 96% of the GPU capex for the MSFT contract. IREN also said its recent contracts carry prepayments of 45–55% of GPU capex. So in that quarter customer cash, not new debt or shares, paid for most of IREN's capex. At NBIS and CRWV it paid for a small share. 6. Operations are the weak point. Some people are questioning the quality of the GPU cloud service IREN delivers to end customers. We share the concern, and it is not new to us. We recorded it in our 1Q26 rebalancing meeting and again in 2Q26 and 3Q26. One of our biggest concerns on IREN is that the CEO treats software as easy, low-value work next to power, land and data centres. That includes the orchestration software that schedules customers' jobs and manages the GPUs. We expected this to slow IREN's growth, not to threaten the business, for three reasons. (A) In our view the bottleneck in AI infrastructure today is powered capacity, not the know-how to run GPU fleets. IREN said on 27 Aug that its recent 3-year contracts price above $20M of revenue per MW of computing capacity. It estimates about a 2-year payback on the GPUs and related equipment. It also said its 2026 capacity is largely sold out. (B) Hyperscalers and other bulk buyers of IREN's capacity can manage and optimise GPUs better than IREN can. What they need from IREN is powered GPU data centres delivered on time. MSFT's $9.7B contract is for GB300 GPUs in four 50MW buildings. MSFT accepted the first, Horizon 1, in August 2026. IREN targets Horizons 2–4 in the December quarter. (C) $NVDA is partnering with IREN and is trying to make running large GPU fleets more standard and simpler. In our view NVDA wants standard fleet-management software so that more companies can run its GPUs well, which makes its chips easier to buy and use. On 7 May 2026 NVDA signed a five-year, $3.4B managed GPU cloud contract with IREN that uses Mirantis orchestration software. The same day it announced a partnership to deploy up to 5GW on NVDA's DSX reference designs. DSX is NVDA's standard set of designs and software for building and running AI data centres. In its 31 May 2026 DSX release, NVDA says the DSX OS software handles tasks such as keeping GPUs healthy, scheduling jobs and running several customers on one cluster. The same release names IREN, CRWV and NBIS among the cloud partners deploying core parts of DSX. The risk in (C) is that IREN's CEO may be too optimistic about how capable NVDA will be at this and how soon. 7. IREN is buying software and hiring managers to run its cloud better. It bought Mirantis, a cloud software company with more than 1,500 enterprise customers, on 4 Aug 2026. IREN said at completion that it paid about 12.6M shares plus about $40M. In July 2026 it hired a product chief from $ORCL (its cloud unit), a data-center development chief from $GOOGL and CyrusOne, and a security chief from $NTNX. We found no hire of a head of cloud operations. We would change our view in three cases. First, if a Horizon delivery for MSFT is missed or rejected. Second, if the Q1 FY27 report arrives with no new contract for 2027 capacity announced. Third, if IREN's weak end-user experience turns out to come from its infrastructure, meaning power, cooling, networking, storage or the GPU hardware itself, and not just from its orchestration software. The third matters because bulk buyers can replace IREN's software with their own, but they depend on IREN for the site and the hardware.
@convequity 3 clicks

Convequity on X

Renting AI capacity protects Microsoft ($MSFT) if the AI buildout busts. It leaves Microsoft exposed if compute gets scarce. On Q2 2026 data, Microsoft's implicit and off-balance-sheet infrastructure commitments were 17 times its headline borrowings according to our AI Bubble Barometer. 1. Microsoft signs a multi-year deal to rent capacity. The partner, a specialist AI cloud such as $IREN or Nebius ($NBIS), typically raises the capital, builds the site, buys the chips and owns the assets. 2. Construction risk, the risk that the GPUs lose value, and most of the financing sit with that partner and its lenders. 3. Microsoft swaps capital spending today for rental payments later. Compare Alphabet ($GOOGL) at 10.2 times. Almost all of its off-book commitments are purchase orders, $811bn of them, for chips and equipment Google will own. We read Microsoft's choice as less than full confidence in the buildout, and Google's as the opposite. In our view the market underweights the risk that compute gets scarce. The partners own any capacity beyond their Microsoft contracts and can sell it to whoever pays more. Microsoft would then have to wait or pay up, as it owns fewer of the data centres and chips it uses. That means higher compute costs and tighter margins. On our reading, Microsoft wins if AI demand falls short, because its partners carry most of the asset risk. If compute gets scarce, Microsoft loses, because renting gives it less control of supply, and Google wins because it will own the chips it ordered.
@aleabitoreddit

Serenity on X

$AAOI has finished a $500m ATM In April.. $600m ATM in June. Then filed another $600m in August. It's out of my control if management wants to abuse this trash repeated ATM structure. Instead following $AXTI like LTA prepayments and using that to fund buildout. Or taking private placements like $NVDA + Nebius for $2B and using that to fund buildout. Or convertible notes at 40% premiums. It's really hard for the stock to break out of $100 or $150 or $200 if there's $600M of selling pressure at those levels (and maybe expectations of future ATMs). I had the same criticism with $IREN with their $6B ATM, which is why it's been rangebound for half a year. And I'll say the same thing with my own thesis names too. I see $AAOI operationally very bullish for 2027, with $471m/month transceiver revenue. 400k ELSFP units/month entering 2028. The revenue ramp is absurd and can compress to single digit forward P/E. But I find it hard to get excited over funding infinite ATMs during the buildout, when you're sacrificing opportunity cost with names currently profitable like $SNDK or Samsung. I think a lot of retail investors conflate stock performance with how a company is doing operationally. $AAOI is capitalizing on a bottleneck with high demand visibility, but stock has been going nowhere near-term with repeated ATMs. I do expect AOI to strongly outperform (especially in 2027) if they stop issuing these stupid ATMs.
@aleabitoreddit

Serenity on X

I would have a much more favorable view on $IREN after they finish their $6B ATM in terms of stock performance (I think they're like 41.5% through). I think Mirantis acqusition would materially shift a lot of perception of how they rank relative to $CRWV, $NBIS, Lambda, Tensorwave, and others in terms of cluster performance once they finish integration. I've always started software side of things is actually a moat (backing up what Coreweave CEO was saying), which was what $IREN lacked before. But M&A to get that gap fixed is probably the most bullish part of what they've done.
@aleabitoreddit

Serenity on X

Wow $NBIS relative outperformance is wild. Back in 2025 I wrote my Neocloud thesis on $IREN to $CRWV. But I made the decision months later to consolidate it all into Nebius... Despite the the hate from FinX back then. Very interesting to see Nebius up 6% even during sector drop… After Oracle was reportedly seeking payment protection if Project Jupiter misses its 2028 opening. With Oracle: "seeking to protect itself from potential cost increases related to a large DC ​project in New Mexico by sending a force majeure notice" per Reuters. Lot of new people like to piggyback after it’s up a lot, but there was a lot less people bullish back in 2025. Especially during the period with a ton of institutional short sellers back in December. I still think markets might be underestimating sum of parts growth, even if its core business grows like crazy 2027+ onwards.
@aleabitoreddit

Serenity on X

I'm not having a fun day since I got sick today. But the bottlenecks are having a good time: $AMD +9.13% | $INTC +14.16% | $ARM +14.33% CPUs go brrr $SIVE +10.32% | $AAOI +3.39% | AMS OSRAM +24.61% Lasers go brrr $EWY +4.08% | $DRAM +2.83% | $MU +2.23% Memory go brrr $NBIS +4.7% | $CRWV +5.03% | $IREN +1.73% Compute go brrr Probably most material developments were CPU:GPU ratio expectations + CPU demand (from new AI releases). As well as optical developments from ECOC 2026 (as seen with $MRVL PR + others) Still a lot to catch up on.
@convequity 2 clicks

Convequity on X

Convequity’s AI Bubble Barometer (ABBX) — Week 17 | 11 September 2026 Our weekly framework tracks valuations, returns, capex, financing risk and forward infrastructure economics across hyperscalers, neo-clouds and model labs. This week’s print is also an Oracle week. $ORCL reported FQ1 FY27 on 10 September: revenue and cloud growth accelerated, disclosed backlog (RPO) jumped to $664B, cash capex ran hot, FCF stayed negative but better than feared, and a $20B ATM equity sale was completed. Capacity delivery was the operational headline — 850 MW and >300k GPUs in the quarter. The ABBX read is below. As of 10 Sep session close (US ET), 13 selected companies (hyperscalers + neo-clouds): -EV / IC = 5.80x (↓ 0.06x from 5.86x on 4 Sep Friday close; ↓ 1.87x from 7.67x on 15 May series start) - All-in EV / IC (incl. private OpenAI & Anthropic) = 6.52x - ROIC / WACC = 3.07x in Q2'26 vs. 3.73x in Q1’26 - ROIC – WACC spread = +23.1% EV/IC tells you what the market is paying for every dollar of invested capital. ROIC/WACC tells you whether those dollars are still earning more than they cost. A classic bubble is when EV/IC keeps expanding while ROIC/WACC is rolling over. That is still not the setup. Returns remain more than 3× the cost of capital while EV/IC for the public builder set sits near 5.8x — well below the midsummer 7× prints and the 15 May series start. The multiple compression looks more financing-driven than fundamental: the return spread is still wide. Oracle is the live test of that financing channel — and on ABBX it sits halfway between the diversified hyperscalers and the neo-clouds. Start with backlog. ORCL’s $664B RPO is now in the same league as $MSFT (~$678B) and ahead of $GOOGL (~$520B) and $AMZN (~$496B). Neo-clouds are a different scale entirely: $CRWV ~$104B, $NBIS ~$40B, $IREN ~$17B. On disclosed backlog, Oracle is already peer to Microsoft — hyperscaler-scale contracted AI demand. The constraint is not finding counterparties; it is delivering power, GPUs and campuses against a backlog that already exceeds its AI enterprise value (EV/RPO ~0.7x on the company row). Valuation paints the halfway case. On Forward Valuation, Oracle’s AI EV is roughly mid-hundreds of billions against ~$163B/year of plan-implied revenue (~12.8 GW internal × ABBX blend (colo vs. GPU rental vs. mode/app derived revenue) — about 2.8x EV / plan-implied revenue and ~4.3x EV / implied gross profit. That sits well below Microsoft, Amazon and Google (roughly high-single-digit to low-double-digit EV/plan on the same framework) and Meta (~4.5x), but well above CoreWeave and IREN (sub-1x to low-1x on plan revenue). EV/IC tells a similar story: Oracle is cheaper than the big four diversified names and closer to the neo-cloud pack on capital multiples, while its backlog and quarterly capex now look hyperscaler-sized. The market is pricing Oracle more like a leveraged AI builder than like Azure/AWS/GCP — even though its contracted demand book looks like theirs. Capacity mix explains part of the gap. In AI Infrastructure Capacity Plans, Microsoft, Amazon, Meta and Google all show some external planned GW — capacity rented from specialists such as CoreWeave, Nebius and IREN. Oracle shows none. Its disclosed plan is treated as internal OCI build. Oracle is the landlord in this graph, not a tenant: it sells AI cloud to frontier labs and enterprises, and does not need a large rented-GPU line to hit the plan. That is control and a cleaner path to own-stack margin. It is also concentration. Build cost, leases and delivery sit on Oracle, not on a diversified supplier set. The Implied Revenue & Valuation Multiples table shows the same shape in the numbers. Oracle’s implied ABBX USD per planned GW is about USD 13 billion, below Microsoft at about USD 17 billion and well below Meta and Google in the mid-20s. That is not a claim that OCI is worse infrastructure. It is a revenue-density assumption. Oracle’s plan is weighted to large-contract AI infrastructure sold to hyperscale counterparties, closer to wholesale cloud economics. Meta and Google’s higher implied USD per GW embeds denser take on their own AI and platform stacks. CoreWeave can look high on a thin equity base. Oracle is the opposite: a huge RPO and more conservative revenue per gigawatt of disclosed plan. Financing is where Oracle looks more neo-cloud than Microsoft. Our Expanded Debt & Capex, split by disclosure vintage: From FQ1 FY27 (31 Aug, condensed balance sheet in the earnings release): borrowings $125.3B, cash $36.4B, IC $156.2B, plus operating-lease ROU/liability lines on the face of the BS. Still on the May / Q4 FY2026 full note lock (not in the FQ1 PR): on-books stack ~$167B (~1.3× borrowings), uncommenced DC leases ~$260B, purchase obligations ~$13B, all-in burden ~$441B (~3.4× borrowings). That roughly 3.4× all-in multiple sits near Amazon (about 4×) and above CoreWeave (about 2.5×) and IREN (about 2.9×), but far below Microsoft (about 17×), Google (about 10×) and Meta (about 9×). Those names carry large off-balance-sheet lease books relative to thinner headline debt. Oracle’s uncommenced lease line is still huge; the multiple looks moderate only because borrowings themselves are already large. The setup is notes debt, a completed ATM equity raise, heavy cash capex, negative FCF, and a large May commitment book that the next 10-Q still needs to update. Management says new contracts need no further equity raise. ABBX treats that as a claim to monitor, not a free pass. Oracle’s Q1'27 cash capex was about USD 28.5 billion — in the same league as Meta, below Microsoft, still below Amazon and Google. Against that sits about USD 664 billion of contracted future work, more than twenty times one quarter’s build spend, with management guiding that roughly half should become revenue inside three years. That is the highest backlog-to-build ratio in the group, and it is why demand on paper looks full. It does not mean this year’s cash bill is already paid. RPO is multi-year revenue, not cash this quarter. Capex is a repeating outflow — FY27 guidance is about USD 90–95 billion, or about USD 70 billion net after customer prepays — and Oracle has also signed about USD 260 billion of data-center leases (as of May 2026) that are not yet on the balance sheet and start as campuses come online. Rent is due once those leases begin, even if customer cash is later. That is why the stock can de-rate on funding even with a full book. The high ratio argues for utilisation if delivery lands. Free cash flow and the lease stack are the separate question of who funds the build before the backlog turns into cash. Other hyperscalers can look weaker on the same ratio because more of their spend serves their own ads and cloud stacks, which never appear as Oracle-style external backlog. Delivery is catching up with that book. Oracle has put about 850 MW and more than 300,000 GPUs into service, with GPU utilisation at 97.9% and renewals at a premium — capacity is being used when it lands. At the Abilene, Texas campus, six of eight buildings are live, and customer acceptance of new clusters has shortened to about 24 hours: delivered capacity is signed and usable in a day, not weeks. The earlier caution still holds: lab tests of how many tokens a cluster actually produces, and real traffic samples through gateways that serve open-weight models, are still thin. Campus delivery looking full is not the same as those checks. Direction still matters. Across the complex, Friday’s snapshot still looks like a build-out, not a hangover. For the 13 names, aggregate AI enterprise value is about USD 10.7 trillion. AI EV over disclosed backlog is in the mid-single digits. AI EV over plan-implied revenue is in the high-single digits on the broader cuts. Planned capacity is still well above 100 GW. Those multiples look attractive if the book converts: a few years of contracted work, and under 10 times the revenue the build plan implies — not a 20-times software multiple on a hope. Financing risk is the other side of the same picture. One debate is a high EV versus invested capital. Another is a high EV versus off-balance-sheet commitments. Oracle now forces both into one name: hyperscaler-scale backlog, a neo-cloud-like equity valuation, and a commitment stack you can inspect in ABBX. Verdict: no broad AI bubble signal. Week 17 of our ongoing weekly series. Updates every Friday. Full interactive dashboard (Fundamentals, Forward Valuation, Expanded Debt & Capex, Capacity) available at Convequity. #AIBubbleBarometer #AIBubble
@convequity

Convequity on X

The AI build is moves more aggressively off the balance sheet Convequity’s debt scan across $MSFT $AMZN $GOOGL and $META: what these companies borrow in public is rising slowly. What they have promised to spend later — leases not yet started, equipment orders, contracted compute — is rising much faster. In 2Q26 that gap is the widest in four quarters. Off-book commitments are running toward +316% year-on-year. On-book debt is barely moving. If you only watch headline borrowings, you miss the cycle. MSFT is the clearest case. On-book debt +15% YoY. Off-book commitments +138%. That is why its all-in financing load is about 17x reported borrowings. Microsoft is under pressure from two sides: shareholders who want more AI spend, and shareholders who want less capex and faster returns. The compromise is to slow the capex it puts on its own books and lock in capacity from $IREN, $NBIS and other neo-clouds instead. How that works: IREN or NBIS borrows, builds the site and owns the chips. Microsoft signs a multi-year contract to pay for the capacity. Microsoft does not put the building on its balance sheet. It takes a future bill — rent and service payments that show up later as operating cost. Asset risk moves to the neo-cloud. Payment risk stays with Microsoft. GOOGL looks similar on a multiple and is doing something different. All-in commitments are 10.2x borrowings: $100bn of headline debt, $121bn including leases, $902bn off-book. $811bn of that off-book pile is purchase orders for its own stack — TPUs and the kit around them — not rented neo-cloud sites. On-book +190% YoY. Off-book +836%. That multiple is Google buying the factory, not renting it. META is closer to Google than to Microsoft: on-book +127%, off-book +680%. Some of Meta’s orders are reserved compute from CoreWeave and Nebius, which Meta will expense as it uses the capacity. Most of the pile is still Meta committing to build and buy for itself. AMZN is the most balanced of the four (+59% on-book / +80% off-book). Both lines are rising together. If the cycle breaks, Microsoft is less stuck with buildings and chips it owns. It is more stuck with bills it already signed. Google is more exposed because more of the capital is already spoken for on its own account. If compute gets scarcer, the extra megawatts IREN and NBIS have not yet sold can go to whoever pays more. Microsoft then has to wait or pay up. Because it owns less of the physical stack, it has less spare capacity it can simply switch on — which means higher compute costs and tighter margins. Powerful cycle. Not a broad bubble. The heat is in the promises, not the 10-Q debt line. Full AI Bubble Barometer available at Convequity.

Tickers on this page

Companies the stories above are also about, besides IREN.