NASDAQ
GOOGL
172 stories
$3.5T Market cap · 2026-11-06
Get GOOGL news in Claude, ChatGPT or your own code. See plans
I quite like $CIFR at $8 since their primary counterparties are $AMZN, $GOOGL + and an AI Lab. They need financing and I remember some projects ran slightly overbudged. But it’s quite different than others who need to buy GPUs like Neoclouds + they often get bunched into the rest
$AMZN, $GOOGL, and $MSFT play five roles for Anthropic at once: 1) Suppliers Anthropic has committed at least $269.4B to "regardless of actual usage." 2) Cloud marketplaces that sold ~48% of Anthropic's H1 2026 revenue. 3) Shareholders. 4) Customers. 5) Competitors. Some of Anthropic's suppliers also help fund its compute: → Anthropic can sell up to $20.0B and $15.0B more of notes or non-voting stock to $GOOGL and $AMZN, subject to certain conditions. → Anthropic can sell up to $42.0B of notes to $AVGO or a financial partner $AVGO designates, solely to pay the rent on its five-year chip lease. → $AMD agreed to buy up to $5.0B of Anthropic's non-voting stock in a July 2026 GPU supply deal, if certain conditions are met. So up to $82.0B of Anthropic's future funding can come from the same companies selling it compute.
Meanwhile, here's what Anthropic has signed up to spend on compute: → $GOOGL: $114.5B from April 2026 to July 2033. → $AMZN: $110.0B from May 2026 to April 2036. → $MSFT: $44.9B from November 2026 to May 2033. → Five-year AI chip lease: $125.2B of rent, up to $42B of which Anthropic can fund by selling convertible notes to $AVGO or a financial partner $AVGO designates. → $SPCX's xAI: Up to $120.5B into 2029 for $NVDA GPU compute, cancellable on 90 days' notice after early 2027. As of June 30, 2026, Anthropic had $265.6B of non-cancelable compute purchase commitments and $321.4B of signed leases that hadn't started yet, up from $54.6B and $50.2B six months earlier. Anthropic has signed ~$266B more since June 30, which by my rough math brings the total to ~$850B (the S-1's own total is still blank). Add an estimated $125-150B for TPU (Google's AI chip) compute in 2028-2031 that isn't in those figures.
Non-cash fair-value charges of $68.0B more than account for Anthropic's $66.3B net loss in Q2 2026. $59.6B came from revaluing convertible notes as Anthropic's valuation rose, and another $8.4B came from revaluing Anthropic's agreements to sell more notes and stock to $GOOGL, $AMZN, and $AVGO. Excluding those charges and stock comp, Q2 adjusted net income was $3.0B. Anthropic also generated $3.9B of operating cash flow in H1 2026, vs. $1.5B used in H1 2025. So the net loss reflects Anthropic's rising valuation, not its operations.
Oh look… more memory durability evidence for $MU and co. Following the news of $GOOGL, other CSPs such as Alibaba, Tencent, Bytedance, and are now negotiating memory LTAs with $SKHY and Samsung. Some customers are reportedly pushing for 10 year contracts as well, into 2035. “As U.S. companies move to secure memory supplies first, Chinese tech giants are joining the race.” Source: Maeil News Don't you think it's a bit strange for your hyperscalers (especially Chinese big tech) to seek LTAs 10 years out… If there's claims of CXMT + YMTC DRAM/NAND overflooding in 2028-2029?
For media enthusiasts, micro dramas are already a major force in Asia and on track to surpass the 2026 China movie box office, while just beginning to reach American audiences and analyst research notes. Are there any winners? Or do we just get a bunch of losers? $NFLX $DIS $GOOG $SKYD
~Half of Anthropic revenue is generated through CSP marketplaces like $AMZN bedrock, $GOOG Vertex, $MSFT Foundry. That is even more than I'd have thought.
Revenue per Employee at $MSFT $META $GOOG over time
The way we were... I've been wondering why there was such a rush to get the $ANT S-1 filed amidst a brutal sell-off in AI related names. Then I read the filing and it's painfully clear. $ANT has scaled at a mind-boggling clip up until this point. Yes the company did generate operating cash flow and profits in 2Q26, but that no longer is representative of the business model going forward. With $500B+ of compute cost commitments with $MSFT $GOOG $AMZN and $SPCX and a $161B equipment lease with $AMD, the entire cost structure of this business is now changing under the "Anthropic Direct" initiative. They're showing you that delivering inference was a structurally high margin business in 2Q26 (50%+ margin). Will it be going forward? Well they're sitting on something that looks like $85-100B in compute commitments ANNUALLY. As a point of reference they spent $5.5B in 2Q26 on compute. Even if they get towards $80B of ARR by year end (which could be optimistic), it's a big hurdle to get to profitability in 2027 unless ARR doubles (or more). There is a placeholder for October 2026 ARR in the S-1. I cannot think of another single sentence in a SEC filing that will have more of an impact on the fate of capital markets than that one. My guess is everyone in the VC and tech community will up their Claude spend this month for that very reason.
The way we were... I've been wondering why there was such a rush to get the $ANT S-1 filed amidst a brutal sell-off in AI related names. Then I read the filing and it's painfully clear. $ANT has scaled at a mind-boggling clip up until this point. Yes the company did generate operating cash flow and profits in 2Q26, but that not longer is representative of the business model going forward. With $500B+ of compute cost commitments with $MSFT $GOOG $AMZN and $SPCX and a $161B equipment lease with $AMD, the entire cost structure of this business is now changing under the "Anthropic Direct" initiative. They're showing you that delivering inference was a structurally high margin business in 2Q26 (50%+ margin). Will it be going forward? Well they're sitting on something that looks like $85-100B in compute commitments ANNUALLY. As a point of reference they spent $5.5B in 2Q26 on compute. Even if they get towards $80B of ARR by year end (which could be optimistic), it's a big hurdle to get to profitability in 2027 unless ARR doubles (or more). There is a placeholder for October 2026 ARR in the S-1. I cannot think of another single sentence in a SEC filing that will have more of an impact on the fate of capital markets than that one. My guess is everyone in the VC and tech community will up their Claude spend this month for that very reason.
@enriqueabeyta @themarketear What about recent and announced share issuance :) ... $goog, $meta, anthropic IPO, spaceX cursor acquisition etc
CEWE sold its low-margin printing unit and agreed to buy Kodak Moments for €88M, expanding into North America.
Alphabet's Gemini 4 Argon aims to undercut rival AI models using proprietary TPUs, while Android, Siri and Google Cloud distribution…
Dustin Quasney 3 clicks
Alphabet, Amazon, Meta and Microsoft plan nearly $800 billion of AI infrastructure spending, testing whether incremental returns can stay…
$NVDA Jensen Huang: Surplus of memory supply is unlikely by 2029. The CNBC question referenced a Korean newspaper opinion on $KOSPI / Samsung / SK Hynix that predicted "new supply will leave the memory sector a red ocean in 2029." "One of the things that's really great about memory is that you can always use more of it", Jensen said, citing LPDDR usage in DCs. And responded "I don't think so" in terms of possibilities of oversupply in 2029. Just my personal opinion, Nvidia probably has the most visibility in the entire industry. Especially around supply procurement and emerging capacity, given they're the ones leading all the architectural shifts. On durability, you have Samsung already allocated majority of capacity into 2031 from SED and hyperscalers like $GOOGL are already reportedly securing memory LTAs into 2035 from KeyBanc. Yet there's endless "it's all over" sentiment on memory now. I'd probably trust Jensen's opinion here.
AI names are backed by actual earnings and contractual demand. Memory names are under <3x fwd P/E, with LTAs extending 5-10 years from your richest companies in the world like $GOOGL. Optical names are compressing to <10x fwd P/E despite 500-1000%+ growth next year Back in the day companies like Cisco/Yahoo were in 100x-600x+ earnings range.
It's fascinating to me that the best capital allocators in the world ( $AMZN, $GOOGL, $TCEHY, $MSFT) with clear views of returns and use cases across many industries are saying that AI Capex is their best use of capital, yet everyone is still skeptical.
@firstadopter $GOOG bull now Tae?
$LITE CEO interview on demand, counterparty + financing risk: - "CFO of $GOOGL came on and reported capex going up Y/Y, we expect hyperscalers for report an increase in capex in earnings. Based on what we see, we see their spend going up." - "The purchase agreements we have, to a large degree, are noncancellable." - The difference between Lumentum and companies like $AVGO, "we are not as dependent on AI datacenter buildout going forward" There's a conversion between electrical products to optical products" creating structural demand. Q: When do you see able to see yourself having more inventory? (catching up to surge in demand) - When we look at demand, there is "no end in sight" TLDR: Financing risk and demand aren't equally distributed across AI supply chains. But markets are just selling everything off together. As for photonics in specific, I said this before, but increasing optical content per XPU + CPO/NPO adoption keeps going up so we'll likely see sector spend go up, even if aggregate capex goes down. There's the usual "no end to demand in sight" commentary, but $LITE does have capacity commitments into 2029 now... and are negotiating for 2030. (with new information that most of which are likely hyperscalers + non-cancellable) So doesn't quite sound like it's all over for the AI trade, especially photonics.
Who had Hyperscalers accelerating to fastest growth rates since Covid at the same time Anthropic massively decelerates on their bingo card for 3Q26? I think very few 3-6mo ago, but here we are! $MSFT $GOOG $AMZN
Just some words of emotional support for your SK Hynix, $SNDK, $MU fanclub. $DELL CEO on memory, suggested: "It's a 4 year problem because there's just not enough supply" Factories take about 4 years to build, meaning new production would emerge 2029 or 2030. and that semiconductor shortages could worsen in 2027 stating: "I do think it gets worse next year in terms of demand and supply". I think it's strange we're getting unprecedented forward visibility with memory... as $GOOGL and richest hyperscalers seek LTAs into 2035. And a major sector crash as well over durability/financing concerns.
Even if the AI model Nubeva Technologies $NBVAF (TSXV: $NBVA) is training to read malware binaries (compiled programs) works, six risks could keep it from paying off: → Labs closing the gap: OpenAI’s GPT-6 Astra nearly doubled GPT-5.6-sol’s score on SRE-Bench, a Columbia University-led test of AI models on reading binaries, within a month of SRE-Bench’s release, and Anthropic says its Claude Mythos Preview model can already reconstruct plausible source code from stripped binaries. The better the labs’ own models get at reading hard binaries, the less they need to pay for outside labels. → Big tech building in-house: Microsoft’s $MSFT Project Ire reverse engineers software files on its own to decide whether they’re malicious. Ire caught only about a quarter of the malware among nearly 4,000 hard-to-classify files in a test Microsoft described in August 2025. But in June 2026, Microsoft said Ire flagged a new malware sample that most major security products, including CrowdStrike’s $CRWD and SentinelOne’s $S, still missed. Google $GOOGL already owns VirusTotal and Mandiant, so the likelier customers are labs that don’t own a malware-scanning site or an incident-response firm already, like OpenAI and Anthropic. → Paid like a contractor: Data labeling companies (firms that turn experts’ knowledge into training data for AI labs) pay 60-70% of what they bill to their contractors, and Scale AI is already recruiting its own malware analysts. So a data labeling company could pay Nubeva by the hour or task, or hire analysts itself, instead of licensing the model or buying the company. → Selling labels teaches the buyer: Every binary Nubeva labels helps the buyer’s model learn the same skill, the reason CEO Randy Chou gave for not patenting how Nubeva reads binaries. So each sale could shrink the edge Nubeva is selling. → Cash and compute: Chou gave no cost for running “hundreds and thousands” of AI agents on a private model. For reference, Nubeva spent $714,772 on R&D in FY2026, equal to 28.7% of its $2.49M in cash as of April 30, 2026. → Missed targets: Nubeva’s AI decryptor builder missed four targets starting in August 2023, so “days and weeks” (Chou’s timeline for the model) is a target, not a result just yet. So even with a working model, Nubeva still has to (1) sign a contract before the labs can read the hardest binaries themselves, (2) get paid for its model and engineers instead of by the hour or task, and (3) fund the training without draining its cash.
AI labs and data labeling companies (firms that turn experts’ knowledge into training data for the labs) could make my $12-19M sale range for Nubeva Technologies $NBVAF (TSXV: $NBVA) look conservative. My previous write-up’s sale comps were five cybersecurity companies buying small startups that detect, stop, or recover from ransomware, and I valued a Nubeva sale at $12-19M, the low end of those deals. AI labs and data labeling companies would widen the set of buyers beyond cybersecurity companies, but I found none that has bought a ransomware recovery company like Nubeva. The closest deals were for a stake in a ransomware recovery company, malware data, and specialist teams: → Microsoft $MSFT: Made an equity investment in Veeam, a data protection and ransomware recovery company, in February 2025, as part of a partnership to build AI-based threat detection and other Microsoft AI services into Veeam’s products. → Google $GOOGL: Owns VirusTotal, a malware-scanning site it bought in 2012, and Mandiant, the incident-response firm it bought for $5.4B in 2022. Its experimental Sec-Gemini model, introduced in April 2025, draws on Mandiant’s threat intelligence data. → OpenAI: Agreed to buy Promptfoo in March 2026 for an undisclosed price. Promptfoo, which tests AI systems for security holes, was valued at $86M after its July 2025 funding round, per PitchBook. → Mercor (a data labeling company): Agreed to buy Deeptune, which builds practice environments for AI agents, in July 2026, months after Andreessen Horowitz led Deeptune’s $43M Series A. The two 2026 deals didn’t disclose prices, and Mandiant and Veeam are far larger businesses than Nubeva, so I can’t put a new number on a sale. Still, my $12-19M range is probably conservative now, since it only counted cybersecurity companies as buyers, and an AI lab or data labeling company would be paying for Nubeva’s engineers and model rather than its ~$500K of FY2026 revenue. But an AI buyer wouldn’t get patents on what it’s paying for. Nubeva’s patents would cover how it recovers files and how it detects hidden encryption, but it won’t patent its binary analysis (figuring out what malware programs do) at all, and CEO Randy Chou explained why: “Anything you patent, that means you have to show how you did it. If you show how you did it, well, somebody can train an LLM to go do it.” So the model is only worth what Nubeva can prove it does, through a signed contract, a named partner, or test results.
I mean it's probably more important to look at the actions over words... I'm sure $GOOGL trying to secure supply until 2035, would spark other hyperscalers to follow suit as well. They probably wouldn't be going 9 years out if they thought there would be overflooding in supply from China in 2 years.
Letters that discuss GOOGL.
Companies the stories above are also about, besides GOOGL.
