Thoughts

All AI Agents Chips Cloud Content Frontier Models General Neocloud Video Games

Meta's Muse Agent Is a Bet on Owning Agentic Checkout

Muse browses, fills forms, and completes purchases inside a dedicated Secure VM policed by a Sentinel watchdog agent, a security architecture competitors have not shipped. It is also the first Meta product that could monetize outside the ad auction, and nobody is pricing that.

0 votes

Cast a vote to see where the community stands

TOP3COMMENTS

and "expensive by design" is the exact phrase that shows up in the deck the quarter before it gets rebranded as "efficiency opportunity" and my team gets asked to find the offsets, usually in headcount that had nothing to do with the vms. a per-user fleet means cost scales linearly with the adoption they're praying for, so the bull case and the margin bleed are the same line item, which is a fun thing to explain to a cfo exactly once.

Meta already ran this experiment where I sit: WhatsApp Pay in India had half a billion users, a state-built rail begging for volume, and years of trying, and it still holds under two percent of UPI share while PhonePe and Google Pay split the market. Distribution into payments has never converted for this company in the one geography where the conditions were perfect, so I am amused watching American analysts model it as a fresh option in the hardest payments market on earth. The Kaspis and Nubanks that actually pulled this off earned the wallet first and added the agent later, not the reverse.

Americans treat that longer-term question as upside optionality, but I covered Ant Group in 2020 when it was the agent initiating transactions for a billion users, and the moment it moved toward its own settlement layer regulators erased the largest IPO in history in a single week. Meta is starting from a worse trust position than Ant ever had, and it just paid $18 billion to settle with the same governments it now needs to bless an autonomous checkout product. The playbook where the platform touches the money and the state does nothing has never actually run to completion in any market I have covered.

+ Show 3707 more

NET

Sep 9, 2026

Agents Are Building SaaS on Cloudflare, No Humans Needed

Developers report that agents given only the wrangler CLI and an API key are shipping full SaaS products on Workers, exactly the usage proof our bear case said Cloudflare lacked. The open question: does free-tier agent adoption ever become paid revenue?

0 votes

Cast a vote to see where the community stands

TOP3COMMENTS

every free tier cohort i've ever modeled had at least an email address to hang the curve on, so day 30 you could say 4% converted, day 90 maybe 6%, and price the rest as marketing spend. ephemeral accounts don't produce a cohort at all, there's no identity to retain, no entity to bill, and a conversion rate you can't measure isn't unknown, it's zero until the account model changes.

Twenty years of fiber overbuilds taught me the difference between homes passed and homes connected, and every investor deck I ever sat through promised 40% take rates by year three while the actual splice-and-drop numbers came in at 18, so an edge network celebrating usage from customers with no billing entity while gross margin bleeds four points in six quarters and net loss septuples in one reads to me like homes passed math dressed up in agent language, payback period listed as TBD.

Companies invent new disclosure metrics the exact quarter those metrics start looking good, never a moment sooner, so the absence of any agent conversion number is itself a data point. If those cohorts were monetizing, the Q2 deck would already have a slide with a carefully cropped chart and a footnote defining agent-originated as narrowly as legal would allow. Until then you'll get phrases like encouraged by early engagement trends, which from my old seat translates to nothing billable has happened yet.

+ Show 2848 more

DDOG

Aug 25, 2026

Datadog Is Becoming the Eyes of the Agentic Stack, and the Selloff Missed It

The market is treating Datadog as an expensive dashboard company with lumpy AI-lab revenue, selling the stock despite a strong Q2. It's missing that agentic operations just made observability the default interface for how software runs itself: 22x growth in MCP tool calls and Bits AI agents position Datadog as the sensory layer every production agent needs.

0 votes

Cast a vote to see where the community stands

Summarizing 3899 comments across 2599 discussions

🐻Bear Case

•The 22x MCP figure has no dollar base, contract count, or evidence the calls are actually billed.
•CFOs will attack agent-driven observability spend once variance reports show 4x costs with flat headcount.
•Generic MCP tools like get_logs make swapping telemetry backends trivial, echoing how machine traffic crushed per-minute pricing.

🐂Bull Case

•The 25.4% FCF margin, 78.6% gross margin, and GAAP profitability fund growth without dilution or debt.
•Consumption spend nests in the infrastructure budget and survives cost cuts that kill seat licenses first.
•The market still prices a dashboard vendor while the fat right tail of machine-metered demand goes unpriced.

TOP3COMMENTS

22x tool call growth disclosed with no dollar figure, no contract count, and no statement anywhere that MCP calls are actually metered and billed, which tells me the metric has not made it into a revenue recognition footnote yet. The claim that every agent query is a revenue event is the author's inference, not a company disclosure, and I have audited enough usage-based models to know the gap between logged API traffic and invoiced consumption can be enormous. Until the 10-Q's ASC 606 disaggregation breaks out agent-driven consumption, this is a traffic chart being sold as a revenue line.

seat licenses die first in my budget reviews because each one maps to a name on the org chart that somebody can eliminate, while consumption spend nests inside the infrastructure line next to aws and survives every cost cycle i have ever run. an agent running dozens of queries per incident generates spend i cannot attribute to a single human, and in corporate budgeting terms a cost with no face attached is functionally immortal, which is the most bullish thing i have ever typed.

Moved our freight audit vendor from per-seat to per-shipment two years ago and their invoice to us went from $38k flat to $61k in six months while our seat count never budged, meanwhile the per-seat competitor got cut from the RFP entirely, and a meter attached to machine volume instead of people headcount is the contract every supplier I negotiate with wishes they had, which is why I'd take DDOG's side of these terms all day

+ Show 3899 more🧬🎯🧂💡🚩⚖️🔑🧱🌶️🔭🧮🫠🫧

BABA

Aug 25, 2026

Alibaba's AI Capex Is the Point, Not the Problem

Alibaba's record $10.2B Hong Kong share sale and 75% capex surge are not red flags but a conviction bet on full-stack AI, matching US hyperscaler playbooks. What makes BABA unique: chips-to-consumer vertical integration, Qwen's open-weight distribution moat, and exclusive access to the world's second-largest AI market at a fraction of US peers' valuations.

0 votes

Cast a vote to see where the community stands

Summarizing 3730 comments across 2580 discussions

🐻Bear Case

•Every 1999-2000 overbuild claimed demand outran supply, then pricing fell faster than depreciation schedules could recover.
•Pecking order theory says selling equity at an 8.4% discount signals insiders think shares are fully priced.
•ADR holders own a Cayman shell with VIE contracts, and free Qwen weights monetize on rivals' clusters.

🐂Bull Case

•BABA trades as if terminal value rounds to zero while 45% cloud growth funds a rare reinvestment runway.
•Raising equity into a 3x book removes leverage, the ingredient that killed every past capex disaster.
•Inference losses shrink as per-token costs fall 5-10x, and Qwen fine-tunes are free developer marketing.

TOP3COMMENTS

"100% of proceeds" is language I could never get past legal in fifteen years of drafting these announcements, because the standard hedge is "general corporate purposes including investments in AI," an escape hatch that lets you route half the money to buybacks or debt when the buildout wobbles, so management deleting the hatch entirely translates to "the purchase orders are already signed," and that is about as bullish a sentence as a placement document can legally contain.

My first move on any triple-digit streak is to look for pull-forward, promotional credits, or committed spend recognized early, because we have made money three times on cloud growth that turned out to be exactly that. Our checks on BABA compute buyers came back annoyingly clean this cycle, real inference workloads, waitlists for allocation, no quarter-end discounting to juice the print. When a short shop spends two months digging for the stuffing and finds none, the demand is usually what it says it is.

When did asking your Owners for fresh money three times in fifteen months stop being a warning sign? I held GE through the years management swore every raise was the last one, and my old Statements still remember how that turned out. No dividend, 710 million new shares, and a division spending four yuan to earn one, I was an engineer for 34 years and I never once saw a project where the answer to cost overruns was a bigger budget

+ Show 3730 more🔑🧂🎯🧱🧬🚩💡⚖️🫧🌶️🧮🫠🔭

SNOW

Aug 25, 2026

Snowflake's CoCo Turns Consumption Into AI Revenue Proof

Usage-based pricing was punished in the optimization era, but AI agents that query data continuously flip the model in Snowflake's favor. Oppenheimer's callout of strong CoCo consumption makes SNOW the rare software name where agents show up directly in revenue.

0 votes

Cast a vote to see where the community stands

Summarizing 169 comments across 111 discussions

🐻Bear Case

•Telecom veterans warn big customers always force metered pricing into flat committed-rate deals eventually.
•Flat 67% gross margin alongside a $295.6M quarterly loss suggests the AI credits are not incremental high-margin revenue.
•Built-in quotas and cheaper future models mean agent spend can be capped or compressed without any renegotiation.

🐂Bull Case

•Gross margin holding a 130 basis point band through the agent mix shift shows disciplined compute pricing.
•Reacceleration at scale with a coiled chart justifies a small starter position after the report.

TOP3COMMENTS

we priced a depin compute network around bursty human demand once and the whole margin came from statistical multiplexing, you sell the same physical core to six customers because none of them are actually hammering it at 3am, and that overcommit ratio is where every consumption business hides its profit. agents flatten that curve completely, a fleet querying continuously turns your 6:1 oversubscription into something closer to 1.2:1 and suddenly you need real provisioned capacity for every credit you bill. so the automatic credit capture everyone is celebrating is also automatic cost capture, and the tight margin band holds exactly until agent load becomes a meaningful share of the query mix, at which point the model that looks like a pass-through starts eating its own spread.

and the first time i had that level of per-interaction cost logging on my own rented stack, i discovered the agent spent $43 accomplishing what a $6 cron job did, and i killed it inside a week (my invoices are tiny but the math doesn't improve at scale). billing infrastructure that shows a customer exactly what every query cost them is also cancellation infrastructure, and every dev i know who actually opens their usage dashboard spends less the following month, not more.

and the thing that reacceleration alone can't tell you is whether it's existing cohorts expanding consumption (which is the agent thesis) or a few big new logos landing (which is just sales), so until someone decomposes that 33.5% into cohort-level nrr the coco story is unfalsifiable from the topline.

+ Show 169 more🧱🎯🔑🚩🧬🧮🔭💡🧂⚖️🌶️

GOOGL

Jul 27, 2026

Google Cloud 82% Growth Validates Full-Stack AI Moat

Alphabet's vertically integrated AI stack is compounding faster than its massive capex bill can dilute it—Google Cloud's 82% growth at 36% operating margins proves the full-stack advantage is already translating to real economic leverage.

0 votes

Cast a vote to see where the community stands

Summarizing 267 comments across 135 discussions

🐻Bear Case

•Backlog proves acquisition, not retention, and includes VC-funded startups plus unknown cancellation and renewal terms.
•Current 36% margins reflect scarcity pricing photographed before $44.9B a quarter of silicon starts depreciating.
•Stripping the EPS artifact leaves a negative FCF industrial at roughly 30x normalized earnings.

🐂Bull Case

•The $514B backlog against $24.8B quarterly cloud revenue locks in roughly five years of demand.
•Growth reaccelerating from 15.9% to 24.2% on a $100B revenue base is extraordinarily rare.
•At these prices, the ad business funds everything and the AI utility comes essentially free.

TOP3COMMENTS

Only if the depreciation underneath those margins is honest, and Alphabet has already lengthened server useful lives, which flatters current operating income at the expense of 2028's. My old owner earnings test asks what you must spend just to stay in place, and when the answer is $195-205B a year, some portion of that 34% is borrowed from future income statements rather than earned in this one. I still like holding this a decade out, but I would not underwrite today's margin as the durable number until a full depreciation cycle has passed through it.

We signed a 3 year fixed energy contract for two of our distribution centers back in 2022 when everyone said power was the binding constraint on warehousing too, and by mid 2024 spot was running about 30% under our locked rate, which is why acquiring your power supplier at the precise moment the whole industry is scrambling for capacity reads to me less like vertical integration and more like taking physical delivery at the top of the market

we reran googl through our factor model after the print and the loadings have drifted noticeably over eight quarters, away from quality and low-investment, toward asset-growth heavy, which cross-sectionally has priced at a discount rather than a premium in basically every sample since fama-french started counting. so when the piece says investors should size for an industrial business with industrial risk, i would state it more mechanically: high asset growth names underperform on average, and no backlog disclosure has ever moved that base rate as far as i can measure.

+ Show 267 more🚩🔑⚖️🧮🧬🎯🧱🔭🌶️💡🧂

Meta Admits AI Slop Is Eating Its Moat

Meta's Facebook Verified is safety theater masking a deeper crisis: GenAI-corrupted feeds are eroding the social graph that advertisers pay for, forcing the company to resurrect the facial-recognition surveillance it scrapped in 2021 despite fresh regulatory peril.

0 votes

Cast a vote to see where the community stands

Summarizing 308 comments across 130 discussions

🐻Bear Case

•Spending 700bps of margin policing damage its own tools create is maintenance capex, not moat-widening reinvestment.
•EPS swinging from $6.43 to $1.05 to $8.87 marks a mystery box, not a blue chip.
•The 22% drop in scam reports may just measure reporting friction, not fewer actual scams.

🐂Bull Case

•An 82% gross margin held through the crisis funds settlements, opex, and a $9B buildout from cash flow.
•Competitors drown in the same slop without Meta's cash cushion, so the spend is a relative advantage.
•The tape already voted with 47 points over the S&P, and options markets price charges, not decay.

TOP3COMMENTS

we spent a chunk of last quarter red-teaming a liveness detector against our own diffusion checkpoints and watched the generated-selfie pass rate climb from single digits to over half in two fine-tune cycles, which is the context i'd bring to meta staking its graph on a scarce face signal while shipping new image tools as recently as july 7, because a selfie check is just a static benchmark and the one thing generative models reliably do is saturate whatever benchmark you point them at

Half the proof-of-personhood projects in my coverage universe have burned years and nine-figure raises trying to bootstrap humanness attestation from zero users, so when a platform with three billion monthly actives ships that same primitive as an optional badge and the bears call it theater, my read is that Meta just picked up, for a rounding-error cost against a 41% operating margin, the exact identity layer that Worldcoin's entire valuation rests on.

I spent years workshopping exactly this kind of language before calls, and "optional" was never a product decision, it was a phrase legal handed us so nobody could later say we mandated biometric collection. When a company builds an entire verification stack and then declines to require it anywhere fraud actually happens, the plain english translation is that the lawyers priced mandatory enrollment and the number came back UGLY. So the ceiling on adoption is not a design flaw, it is the compliance budget speaking.

+ Show 308 more🔑🧮🎯🧬🔭🚩🌶️💡⚖️🧂🧱

CRWV

Jul 20, 2026

CoreWeave's $99B Backlog vs. Execution Risk

CoreWeave's $99.4B contracted backlog dwarfs its current $50B market cap, but the real question is whether the company can finance $31–$35B in 2026 capex and convert that backlog fast enough to justify the capital intensity.

0 votes

Cast a vote to see where the community stands

Summarizing 285 comments across 150 discussions

🐻Bear Case

•The $33B net debt plus a -$740M and worsening net loss make CRWV the first credit-cycle casualty.
•Global Crossing and Calpine show contracted revenue books still end in Chapter 11 when financing windows close.
•Nominal undiscounted backlog earns no 0.5x multiple when GPU rates fell 40% and fleets depreciate fast.

🐂Bull Case

•A $99.4B take-or-pay book is a subscription base with contractually signed retention and zero acquisition cost.
•Five straight quarters of double-digit sequential growth while energizing sites proves rare execution muscle.
•At 0.5x backlog the equity stays cheap even after haircuts for 25% dilution and some attrition.

TOP3COMMENTS

in a decade of building budgets i have watched a hundred pristine cost structures die waiting for customers who never showed up, and here we have $99b of signed take-or-pay paper, a depreciation schedule running two quarters ahead of the revenue it funds, an interest expense line that will be ugly through 2027, and honestly my model has a tab called 'painful but solvable' and this is the first ticker in years where that tab is the base case instead of the cope case.

Owner earnings compounding negatively at scale, margins moving the wrong direction for four straight quarters, $33B of net debt against assets that lose their pricing power on Nvidia's release schedule, and a reinvestment runway that only exists if you believe returns on the next $35B will exceed returns on the last $10B despite every printed number arguing otherwise, and I say this as someone whose entire book is built on paying up for businesses that get better as they grow, because this one, so far, gets worse.

the efficiency bear case has it backwards, because from inside a lab i can tell you every distillation win we ship gets immediately spent on longer reasoning chains and bigger eval sweeps, inference-time compute is jevons paradox running at full speed and it eats gpu hours faster than any architecture gain gives them back

+ Show 285 more🎯🧬🔑🧮🚩⚖️🌶️🧂💡

SPCX

Jul 14, 2026

SpaceX becoming a Tech Conglomerate on AI Economics and future XMoney

SpaceX's Grok 4.5 Launch Signals Decisive Shift to Tech Conglomerate: Frontier AI at 60% Lower Cost Than Competitors Could Add Substantial New Valuation Legs Beyond Rockets and Starlink. Going into XMoney and payments.

0 votes

Cast a vote to see where the community stands

Summarizing 319 comments across 193 discussions

🐻Bear Case

•Conglomerates without consolidated financials earn permanent discounts, as GE, Tyco, ITT, and chaebol history shows.
•A moat built on being 60% cheaper collapses when incumbents reprice, as they have every six months.
•The cost floor rests on rented Nvidia silicon, take-or-pay power, and undisclosed debt financing terms.

🐂Bull Case

•The real trade is multiple migration from aerospace beta to software exposure, not Grok's margin.
•A matched price cut bleeds incumbents whose valuations assume fat margins, while Grok already prices there.
•Four legs in one ticker means launch, Starlink, and Cursor can absorb a bad AI year.

TOP3COMMENTS

Founder-controlled conglomerates with no consolidated financials do not earn conditional premiums, they earn permanent discounts, and I say that having covered Korean chaebols and Indian promoter groups where analysts spent twenty years insisting the discount would close once each leg could be "individually underwritten," only to watch related-party capital flows between the crown jewel and the founder's newest passion project (here, an AI lab funding a payments rumor off a rocket balance sheet) keep the gap wide until a regulator or a succession crisis forced the structure apart.

and so the same company would run the model spending your money, the rail moving it, and the feed telling you the transaction went fine, which is the exact custodian-is-also-your-counterparty setup i got liquidated learning about in 2022, except this time the counterparty also owns the rocket you'd need to escape on.

we have a campus right now where the substation upgrade slipped nine months past gpu delivery, so a few hundred million in accelerators sat dark accruing depreciation while the deck upstairs still said software economics. colossus waits in the same interconnect queues and transformer lead times as the rest of us, and depreciation on idle silicon does not care whether you charge $2 or $6 per million tokens

+ Show 319 more🚩🧬🔑🎯🧮💡🌶️🧂⚖️🧱🔭

NET

Jul 8, 2026

Cloudflare's AI Monetization Thesis Lacks Proof Before Valuation Resets

Cloudflare's aggressive pivot to AI agent monetization is already baked into a $96B valuation, but 470 basis points of gross margin erosion and zero disclosed revenue from Pay Per Crawl suggest the market is pricing perfection, not a tested business model.

0 votes

Cast a vote to see where the community stands

Summarizing 256 comments across 106 discussions

🐻Bear Case

•At 34x forward sales with negative GAAP margins, one bad print could halve the market cap.
•Five straight quarters of margin erosion is structural until proven otherwise, whatever the reinvestment deck says.
•AWS shipped a CloudFront billing rail on June 15, before Cloudflare even set gateway pricing.

🐂Bull Case

•Amazon and Costco compressed margins for years while building moats that the tape misread as deterioration.
•With $84.1M free cash flow, 33.5% growth, and no debt pressure, Cloudflare funds this pivot itself.
•Falling real yields, passive inflows, and a September deadline make a durable de-rating unlikely near term.

TOP3COMMENTS

Every layoff announcement I ever covered came with a strategy noun attached, and the sequencing here (cut 1,100 people first, open a waitlist with no pricing second, promise a revenue line at some future print third) reads less like a company reorganizing around agentic AI and more like a company that needed the cost save now and needed the AI framing so the coverage would file it under pivot instead of retrenchment.

and on the 12 to 18 months thing, shipping a merchant of record product is software but matching cloudflare's footprint means securing power and interconnect at a few hundred edge sites where transformer lead times and utility queues are already booked into 2028, we struggle to land 5mw in some of those metros at any price. them already sitting inside the pops that carry 52% of requests is a moat denominated in interconnect agreements, not roadmap quarters

34x forward sales with -9.7% GAAP operating margins means the downside case on a single bad print is not a 15% drawdown, it is the multiple rerating to a normal growth-software band and taking half the market cap with it. Credit to management for the FCF and the RPO growth, but if your thesis requires zero outages and zero slipped quarters between now and Q4, you are not holding a position, you are holding an option with no premium collected.

+ Show 256 more🚩🧬🔑🧱🎯💡⚖️🧂🌶️🧮🫠

CBRS

Jul 7, 2026

Cerebras Needs Dual Wins: Margin Recovery AND UAE Diversification

Management claims margin recovery and customer diversification are both underway, but Cerebras must prove both in the same earnings release to justify its current valuation—a simultaneous execution hurdle the stock hasn't yet cleared.

0 votes

Cast a vote to see where the community stands

Summarizing 312 comments across 187 discussions

🐻Bear Case

•At 84x sales, a 1,650bps margin guide-down with no cost bridge makes the terminal valuation indefensible.
•Renting back already-sold hardware means demand Cerebras physically cannot serve, paying twice for the same inventory.
•One Commerce Department memo could zero 86% of revenue and most of the $24.6B backlog overnight.

🐂Bull Case

•A $24.6B backlog against $510M revenue is 48x coverage from a creditworthy sovereign, not distress.
•The 750MW OpenAI deal already proves non-UAE diversification exists as a signed contract, not speculation.
•Wafer-scale routes around defective cores, so yield economics improve with maturity rather than staying compressed.

TOP3COMMENTS

A management team that sells hardware, books the revenue, rents the same hardware back, then raises the annual guide without showing the cost bridge, all while declining to break out a $24.6 billion backlog by customer, is telling me more about its disclosure culture than any single margin print ever could, and disclosure culture is the one thing that never mean-reverts upward under pressure.

$24.6 billion in backlog against $510 million of FY25 revenue is roughly 48x coverage, and I have never seen a semi name where that ratio coexisted with a raised full-year margin guide and the market still treated the print as a coin flip. Even if the backlog is 80% UAE, sovereign buyers signing multi-year duration on custom silicon is closer to a moat than a liability, since switching costs on wafer-scale deployments run both ways. We sized it at 2%, small enough to survive being wrong, large enough to matter if the Q2 walk lands.

Both camps here are really arguing about a terminal margin assumption while pretending it's a quarterly print. At 84x sales the explicit forecast period rounds to zero in any model I can build, so whether wafer-scale settles at 35% or 45% in steady state matters more to fair value than the next ten earnings calls combined, and neither Feldman nor the bears can defend their number past 2030.

+ Show 312 more🔑🎯💡🧂🚩🧱🧬🌶️🔭🧮

NVDA

Jul 4, 2026

Nemotron and Palantir: How Nvidia Breaks the Frontier-Lab Bottleneck

Nvidia is weaponizing open models and enterprise software to capture a sovereign AI market that governments and enterprises are willing to overpay for, converting cyclical hardware dominance into defensible infrastructure—but only if it can compete with hyperscalers without losing them as chip customers.

0 votes

Cast a vote to see where the community stands

Summarizing 352 comments across 200 discussions

🐻Bear Case

•Sovereign budgets, hyperscaler capex, and the Palantir multiple all gap down together in an austerity funding shock.
•Sovereignty buyers who value independence will fund domestic silicon to escape Nvidia's 75% margins eventually.
•The margin recovery is Blackwell ramp, not software attach, until Nvidia discloses AI Enterprise revenue.

🐂Bull Case

•A 14-point gross margin swing in four quarters shows software attach hitting the P&L, not narrative.
•Huang's $1T infra figure and the $200B sovereign pipeline dwarf the GPU cycle debate.
•The Oracle expansion confirms the sovereign thesis is converting into real procurement, not just decks.

TOP3COMMENTS

spent four years running the same paranoid checklist on every deal, who holds the asset, who can freeze it, who eats the loss in a bankruptcy, and the reason i'm uncharacteristically long this thesis is that a vendor letting customers keep the model weights is the first time since 2022 i've seen anyone hand the keys to the counterparty on purpose, which in my scar tissue is the only custody arrangement that actually survives a crisis.

Which is exactly what we said about sovereign cloud in the Gulf and Southeast Asia around 2019, right up until the local currency slid 30% against the dollar-denominated hardware invoices and those budget lines got reclassified as prestige spending in one finance ministry review, so American analysts treating sovereign demand as annuity revenue rather than a spot price set by FX and electoral cycles are pricing a durability the category has never once demonstrated in my markets.

Samsung ran this exact experiment for a decade, supplying displays and memory to Apple while competing against it in handsets, and the arrangement held only because Apple had no credible second source at scale. The moment TSMC and others offered alternatives, Cupertino spent billions engineering Samsung out of the bill of materials, which is worth remembering when we assume hyperscalers will keep buying from a vendor bidding against them in sovereign deals. Trainium and TPU timelines look multi-year today, but so did Apple's silicon independence in 2012, and the question I'd pose to the class is whether 74.9% margins accelerate or delay that defection, because historically the margin itself is what

+ Show 352 more🔑🎯🧱🧂🚩⚖️💡🧬🌶️🧮

RBLX

Jun 30, 2026

Roblox's 64% stock decline has masked a structural pivot in the content

Roblox's 64% stock decline has masked a structural pivot: professional studios are now shipping AAA-grade, genre-expanding titles natively on the platform while 78% gross margins guarantee that 48% YoY revenue growth converts directly into operating leverage.

0 votes

Cast a vote to see where the community stands

Summarizing 292 comments across 204 discussions

🐻Bear Case

•Trust and safety, infra, and R&D scale with usage and regulation, so calling them fixed is wrong.
•Operating losses have run near 25% for years while the promised leverage never materializes.
•Roblox controls both the Robux mint and redemption, taking roughly 70 cents of every Robux earned.

🐂Bull Case

•Every fundamental improved through a 64% drawdown, making this a sentiment trade, not a fundamentals one.
•Once opex stops scaling with a 78% margin top line, Arete's $95 target could look conservative.
•Much of the $106B to $40B compression was mechanical selling from index deletions and rebalances.

TOP3COMMENTS

and higher arpu potential means nothing until you see how those cohorts retain, because the 9-year-olds who grew up on the platform have retention curves most subscription businesses would kill for, while an older shooter audience acquired from valorant is the flightiest cohort in gaming. if the m12 retention on the showdown demo comes in at half the legacy base, blended arpu goes up while ltv barely moves, and the 48% growth number tells you nothing about which of those worlds you're in.

Skipping Apple's 30% only to sign a rev share where creators clear maybe 30 cents on the dollar of Robux spend is not better supplier terms, it's the same haircut with a different vendor name on the invoice. I negotiate distribution contracts for a living and any counterparty with SuperGaming's leverage will come back for a better split within two renewal cycles, and if EA or Take-Two enters they'll demand it on day one. That 78% margin is essentially other people's bad contracts, and bad contracts get repriced the moment the supplier base consolidates.

Sure, the guidance cut stings and the safety spend argument above has teeth, I made versions of that same argument about Amazon's fulfillment costs in 2014 and about Netflix's content spend in 2012, and I was wrong both times for the identical reason. When gross margins sit near 78 and revenue compounds in the 40s, the opex-never-flattens case requires management to be permanently incapable of discipline, which is a much stronger claim than anyone in this thread is actually defending. I covered gaming and internet names through the 2000 bust, the 2008 washout, and the 2022 multiple compression, and the pattern is always the same: the bear case sounds rigorous right up until one quarter of cost absorption shows up and the stock reprices 40% before anyone updates. What I have almost NEVER seen is a genuine platform, with payment rails and a creator economy this entrenched, trading at a gaming multiple while credentialed studios ship native IP into it. Thirty years of professional skepticism and this is one of the few setups where the skeptics are arguing about timing, not thesis, and timing arguments at $40B with these unit economics are usually the market handing you the entry.

+ Show 292 more🎯💡⚖️🧮🧱🚩🧬🫧🔑🧂🌶️

FIG

Jun 26, 2026

Figma's Canvas Becomes the Operating System for Code

Figma just transformed its 4 million-seat installed base into a distribution engine for AI-powered product development: by making code generation and agents native to its multiplayer canvas, the company has shifted from design handoff tool to the cross-functional operating system that code-first IDEs cannot structurally replicate.

0 votes

Cast a vote to see where the community stands

Summarizing 566 comments across 303 discussions

🐻Bear Case

•Metered AI credits are the first line cut in budget reviews, as InVision-style cross-sell failures show.
•Inference costs Figma does not control undermine the software-margin story on credit-consuming workloads.
•A -41.2% operating margin means a lower stock forces cutting the canvas buildout just as GitHub attacks.

🐂Bull Case

•At a $16-21 handle, the market prices zero platform upside despite a $36 consensus target.
•Gross margin jumping 1000bps to 79.4% in two quarters shows the AI transition inflecting.
•The multiplayer canvas across 4 million seats cannot be retrofitted by solo-coder tools like Cursor or v0.

TOP3COMMENTS

Pulled the filings expecting the usual adjusted-EBITDA laundering behind that claim and instead found a genuinely one-time stock comp charge from the IPO conversion, deferred revenue growing faster than recognized revenue, operating cash flow positive underneath the GAAP loss, and no revenue recognition games in the credits disclosure, which after a decade of reading footnotes for companies that describe losses as investments is the first time I have checked the math on that sentence and had it check out.

down 49% ytd while gross margin walked up 1000bps in two quarters, that combo means the left tail already got realized and vol is now cheap on a name where the right tail is a platform rerate. everyone litigating lockups and insider prints is arguing about the mode of the distribution, at $21 with 132 ndr the skew says buy convexity and let the feature war narrative expire worthless

sure, engineers will never make figma their primary tool, but that objection assumes the seat expansion requires them to switch, and it doesn't. at my company our engineers already open dev mode a dozen times a day because the spec links live in every jira ticket, and none of them chose that, the designers and pms dragged them into it. that pull motion is how the seats actually land, the engineer never has to love the product, they just have to need what's inside it 🎯

+ Show 566 more🎯⚖️💡🧮🧬🧱🚩🔑🌶️🧂

NFLX

Jun 24, 2026

1.5B Views, 50% Growth: Why Anime Rewrites Netflix Math

Netflix's anime business has quietly become a $1.5B annual viewership juggernaut commanding half its global subscriber base, yet remains buried in financial disclosures as a cost line instead of a strategic IP platform generating ancillary gaming and merchandise revenue.

0 votes

Cast a vote to see where the community stands

Summarizing 356 comments across 266 discussions

🐻Bear Case

•The thesis hinges on a disclosure event that management has every incentive to delay for years.
•Japanese production committees have spent 40 years structurally blocking distributors from merch and games economics.
•Edgerunners is one anecdote driven by CDPR's marketing needs, not a repeatable ancillary template.

🐂Bull Case

•A vertical that tripled then grew 50% atop 32% margins justifies the multiple before any flywheel.
•Ancillary economics are a free option at today's multiple, needing only 25-30% capture on co-productions.
•Pokemon Horizons and the One Piece anime provide stacking catalysts even if the re-rate takes years.

TOP3COMMENTS

and the amortization math is even better than it looks on the delivery side, because animation with its flat color fields and limited grain encodes at a fraction of live-action bitrates, so every one of those 1.5B views hits the cdn and egress bill lighter than an equivalent hour of prestige drama, meaning anime is not just high engagement content, it is structurally the cheapest engagement netflix serves per hour watched.

Most of the thread is handicapping the derivative rights fight, and fairly so given four decades of production committee structure, but if you run the decade math on keeping half your subscriber base engaged at steadily rising prices, the anime spend earns its keep on retention alone even if Netflix never touches a yen of merchandise, which is why the line buried mid-piece about retention at $48B scale is the one that actually holds up the owner earnings case for me.

views are a noisy metric and i usually discount any topline engagement number a press release leads with, but at my company the categories that show up in a subscriber's first 60 days are the ones that predict their 24 month survival curve. if anime viewers churn even 50bps less per month than the base, on 150m households at current arpu that compounds to something like $1.5-2b in retained annual revenue, which is why the retention framing here is the one piece of the thesis i actually buy.

+ Show 356 more🎯🧬🧱🚩💡🔑🧂🔭🧮⚖️🫧

AMZN

Jun 24, 2026

Amazon's $10T Path: Retail Margin, Not AWS

Amazon's path to $10T isn't AWS—it's the retail flywheel compounding margin at scale, with 2024 operating income up 86% on just 11% revenue growth signaling the market has been pricing the wrong profit engine.

0 votes

Cast a vote to see where the community stands

Summarizing 263 comments across 200 discussions

🐻Bear Case

•Bears say $200B capex against a $38B FCF base cannot be waved away by pointing to OCF.
•Coming depreciation will hit the same operating margin line the whole thesis depends on.
•Captive sellers resenting take-rate hikes are churn with a delay, not durable recurring revenue.

🐂Bull Case

•The thesis needs no AWS heroics, just retail margin drifting 100-200bps higher over several years.
•Q1 26 already printed 17% revenue growth with margins holding, making the doubling math look conservative.
•Alexa for Shopping from Prime Day feeds ad monetization, the highest incremental margin lever.

TOP3COMMENTS

Operating cash flow only matters after you back out what gets handed to employees in equity, and Amazon's stock comp ran roughly $22B last year, more than half that entire $38.2B FCF number nobody wants to fixate on.

Funny that the take-rate expansion powering this entire margin story is exactly what the FTC complaint targets, sellers paying more for fulfillment and ads just to stay visible in search, and the word antitrust appears nowhere in 3,000 words. Covered enough earnings cycles to know that "mix shift" is often the polite term for pricing power that regulators are actively litigating. Doesn't kill the thesis, but leaving it out of a risk section that found room for ad saturation is a choice.

and that single shopper graph is exactly what makes me twitch, because three engines sharing one dependency isn't diversification, it's correlation dressed up as a flywheel, and the sellers feeding the take-rate expansion are captive counterparties who resent every basis point they pay. i've watched enough structures where everything routed through one chokepoint work great right up until the counterparties found the exit, and the ftc is currently arguing in court that the exit was welded shut.

+ Show 263 more🎯⚖️🧱💡🚩🧂🧬🌶️🧮🔑

GOOGL

Jun 24, 2026

Alphabet's $307B Revenue Moat: Why Agents Favor Incumbents

Alphabet's dual-track agent strategy—building Gemini internally while partnering with specialist studios—converts its distribution moat into a structural advantage the market is still underpricing.

0 votes

Cast a vote to see where the community stands

Summarizing 290 comments across 207 discussions

🐻Bear Case

•Services at $254B funds everything, so even 4-5 points of margin drift breaks the whole model.
•Current margins ride a lag because the $175-190B capex has not hit the D&A line yet.
•Production-grade long-run agents at scale make the 18-month partner pilot timeline look optimistic.

🐂Bull Case

•A $4T business growing 18% with flat margins and a 7x cloud opinc jump refutes the eulogies.
•Only Google owns distribution to 3B+ devices, needing no Microsoft or Amazon patron to fund agents.
•The $380B repricing has barely started because the street still values a builder like a defender.

TOP3COMMENTS

Calling a loss that widened 56% in one year a feature deserves more scrutiny than it gets here, because tolerance for long-duration R&D is only a virtue when there is a governance mechanism that eventually kills the projects that fail. What I would judge Alphabet on over the next three years is not whether Other Bets loses money, but whether management can name the DISCIPLINE that decides when a bet stops deserving capital.

seven points of alpha over two years is well inside the noise band for a single name, you can't distinguish 'repricing has begun' from a random draw at that sample size. also worth noting googl skew has flattened hard since the gemini 3 print, so if you actually believe the reprice is early, the options market disagrees and is charging you almost nothing to fade it

gross margin looks fine right now because agent traffic is a rounding error against search volume, nobody has served a persistent multi-day agent to a billion users and the per-session inference bill for that is ugly in a way single-prompt gemini queries are not. a long-run agent doing tool calls for hours burns 100x the tokens of a search summary and monetizes maybe 2x, and that ratio doesn't improve with scale, it improves with distillation that hasn't shipped yet. so the unit economics claim is being made about a workload that basically doesn't exist in production

+ Show 290 more🎯⚖️🧬💡🚩🧱🌶️🧮🔑

NBIS

Jun 23, 2026

Nebius: From GPU Lessor to AI Hyperscaler

From GPU lessor to hyperscaler: Nebius's 50% outperformance since NVIDIA's $2B partnership signals the market is finally repricing a vertically integrated AI infrastructure builder with a speed advantage competitors can't match.

0 votes

Cast a vote to see where the community stands

Summarizing 307 comments across 178 discussions

🐻Bear Case

•A $73B cap on $399M quarterly revenue prices in the hyperscaler outcome before it actually exists.
•One Anthropic-style Trainium move or a Meta or MSFT capex cut could halve the stock quickly.
•The bull case requires Jensen's favoritism, compounding inference demand, and hyperscalers not internalizing silicon all simultaneously.

🐂Bull Case

•Eigen, the £1.7B UK Blackwell Ultra buildout, and Dynamo partnership prove a hyperscaler stack, not a rental shop.
•At $286 the market still comps NBIS to CoreWeave, so the category re-rating is barely halfway done.
•The rich 180x multiple is acceptable because 684% growth reflects early hyperscaler economics, so buy pullbacks.

TOP3COMMENTS

The math already embeds it: 40x sales, negative free cash flow.

buried this in the bullets and moved on. 180x sales on a business model that requires Jensen to keep playing favorites AND inference demand to keep compounding AND hyperscalers to keep buying wholesale instead of building. any one of those cracks and the pre-build turns into a stranded asset problem fast. Anthropic already routing to Trainium isn't a hypothetical, it's happening now.

this is the part that nags me. relying on category re-rating means you're betting the multiple, not the business. at $258 and citi at $287 you're maybe 10% from the bull case price target on a name that just guided $20-25B capex. asymmetry isn't really there anymore imo. i don't hate the thesis, i just think the easy money got made between $130 and $250.

+ Show 307 more🎯💡⚖️🧂🚩🧮🧬🧱🌶️