Institutional Investor Login

Day: September 29, 2026

Futurum Equities Follow 1,919 21,964

FuturumEquities
FuturumEquities avatar

Futurum Equities

@FuturumEquities ·
5h 2105007055308210369

ChatGPT is now deployed into Slack and Microsoft Teams without individual licenses, part of OpenAI's push toward 1.2B users.

Distribution through existing seats is how agentic tools scale past pilots, bypassing the per-user friction that stalls enterprise rollouts. $MSFT

Image for the Tweet beginning: ChatGPT is now deployed into Twitter feed image.
Image for the Tweet beginning: ChatGPT is now deployed into Twitter feed image.
Image for the Tweet beginning: ChatGPT is now deployed into Twitter feed image.
Image for the Tweet beginning: ChatGPT is now deployed into Twitter feed image.
Reply on Twitter 2105007055308210369 Retweet on Twitter 2105007055308210369 0 Like on Twitter 2105007055308210369 3 X 2105007055308210369
FuturumEquities avatar

Futurum Equities

@FuturumEquities ·
5h 2104994814471131420

SpaceX's compute agreement with Anthropic scaled to a potential $84.5 billion through 2029, up from about $45B disclosed earlier.

Most of that reportedly cancels on 90 days notice, so the committed figure sits well below the headline. $SPCX

Image for the Tweet beginning: SpaceX's compute agreement with Anthropic Twitter feed image.
Image for the Tweet beginning: SpaceX's compute agreement with Anthropic Twitter feed image.
Reply on Twitter 2104994814471131420 1 Retweet on Twitter 2104994814471131420 2 Like on Twitter 2104994814471131420 10 X 2104994814471131420
FuturumEquities avatar

Futurum Equities

@FuturumEquities ·
10h 2104931076179124307

"They're about to report Q4 earnings at 14 weeks instead of the normal 13 weeks." -@StockSavvyShay

Shay on the $MU print Wednesday. The Q1 guide goes back to 13 weeks, so growth could look like it is slowing and the numbers softer than they are. With @StockMarketNerd.

Image for the Tweet beginning: "They're about to report Q4 Twitter feed video.
Reply on Twitter 2104931076179124307 2 Retweet on Twitter 2104931076179124307 0 Like on Twitter 2104931076179124307 4 X 2104931076179124307
Retweet on Twitter Futurum Equities Retweeted
StockSavvyShay avatar

Shay Boloor

@StockSavvyShay ·
10h 2104919521740165365

$MU earnings tomorrow have a huge accounting nuance because Q4 has 14 weeks instead of the normal 13.

So Q1 guidance could look weaker sequentially simply because it goes back to 13 weeks even if the actual weekly revenue run rate stays strong so I wouldn't overreact to a

Image for the Tweet beginning: $MU earnings tomorrow have a Twitter feed video.
Reply on Twitter 2104919521740165365 33 Retweet on Twitter 2104919521740165365 23 Like on Twitter 2104919521740165365 323 X 2104919521740165365
Retweet on Twitter Futurum Equities Retweeted
StockSavvyShay avatar

Shay Boloor

@StockSavvyShay ·
22h 2104744468809421218

$MU at ~6x earnings tells you market is still pricing 2026 like peak AI memory and treating Micron like the old cycle despite SCAs and customers asking for more supply than Micron can deliver.

That gets harder to justify with HBM shipping across $NVDA, $AMD and custom

Image for the Tweet beginning: $MU at ~6x earnings tells Twitter feed video.
Reply on Twitter 2104744468809421218 43 Retweet on Twitter 2104744468809421218 34 Like on Twitter 2104744468809421218 580 X 2104744468809421218
FuturumEquities avatar

Futurum Equities

@FuturumEquities ·
28 Sep 2104694083767644386

AMD is paying $8.2B in stock for World Labs, whose models generate and simulate 3D environments from text, images and video.

The heterogeneous compute thesis widens: robotics and simulation are where the re-rating lives, and AMD is buying the stack to reach it. $AMD

Image for the Tweet beginning: AMD is paying $8.2B in Twitter feed image.
Reply on Twitter 2104694083767644386 1 Retweet on Twitter 2104694083767644386 0 Like on Twitter 2104694083767644386 8 X 2104694083767644386
Retweet on Twitter Futurum Equities Retweeted
StockSavvyShay avatar

Shay Boloor

@StockSavvyShay ·
28 Sep 2104673583142490269

Last week I shared my 5 favorite AI power names on the FE YouTube channel and $NVTS made the list because it is one of the few pure plays across the full grid to GPU power chain.

That means Navitas can benefit at multiple points in the power chain from GaN inside the rack to

Image for the Tweet beginning: Last week I shared my Twitter feed video.
Reply on Twitter 2104673583142490269 14 Retweet on Twitter 2104673583142490269 15 Like on Twitter 2104673583142490269 185 X 2104673583142490269
FuturumEquities avatar

Futurum Equities

@FuturumEquities ·
28 Sep 2104668514380321251

"You could be wrong and make money, and that's really cool." -@StockMarketNerd

Brad on the $MU bulls who say memory is now a smooth grower with no dip in demand coming. He thinks they are wrong, and that they will not lose money with that view for a while. With @StockSavvyShay.

Image for the Tweet beginning: "You could be wrong and Twitter feed video.
Reply on Twitter 2104668514380321251 2 Retweet on Twitter 2104668514380321251 1 Like on Twitter 2104668514380321251 9 X 2104668514380321251
Select Date

WHAT $AMD JUST REVEALED ABOUT THE NEW AI ECONOMY

  • AMD
  • Daily Shay

Podcast

Meta got the re-rating both hosts called at the start of 2026. META closed Thursday up 27% from the day Muse launched on September 8, AMD crossed $1 trillion on September 21, and Micron reports a 14-week quarter on September 30. Daniel Newman and Shay Boloor take the three in order on episode 58, then take your questions. Daniel says the AI winner will be the company with the most compute and energy at its disposal. Meta kept buying CPUs, GPUs and its own accelerators while the market wrote it off, and Muse puts that compute to work. Muse is free up to 100 million tokens a week, with $20 and $100 tiers above that, and Meta plans to take a small fee on each transaction. Daniel calls the transaction model the move Apple should have made. Shay's four qualities of an AI winner are proprietary data, install base, compute and energy. He says Meta had three of them all along, argues it now deserves a Google-like multiple for a consumer AI product, and warns that the next quarterly report will not show any of it. On AMD, Shay counts 40x 2027 earnings against 15x for NVDA and warns that warrants held by Meta and OpenAI will lift the share count in any 2030 model. Daniel says the market treats Nvidia as the incumbent with everything to lose and AMD as the entrant with everything to gain, and that the stock is priced as if Helios, AMD's first rack-scale system, carries almost no execution risk. On MU, Shay explains why the guide could look light: fiscal Q4 had 14 weeks and Q1 has 13. He is watching strategic customer agreements, HBM4 share and capex timing, and has not heard one plausible sign that memory pricing power is eroding. Daniel says Futurum's 2030 CPU forecast has more than doubled since January, walks through the follow-up reporting on the Coxon post, and says every AI scare so far has ended in a rally. He is long VKTX after Viking's maintenance data and says Novo Nordisk should buy it. The Q&A covers DigitalOcean and its move past GPU rentals, Everpure after its investor day, the rest of the year with the 10-year at its highest since 2007, Shay's agentic AI picks for 2027, AI power and 800-volt DC, Muse against Grok, the Anthropic IPO, Amazon in 2027, and why crypto has not rallied more. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. Chapters: 00:00 Episode 58: Daniel is back from Snapdragon Summit 04:09 Meta's re-rating on Muse 14:49 AMD at $1 trillion against Nvidia 22:31 Micron preview: the 14-week quarter 28:01 Futurum's CPU forecast doubled, and the Coxon post follow-up 34:54 Viking Therapeutics after its maintenance data 37:33 Q&A: DigitalOcean beyond GPU rentals 39:34 Q&A: Everpure after its investor day 42:48 Q&A: the rest of the year, the war and 5% yields 47:19 Q&A: agentic AI picks for 2027 49:25 AI power and 800-volt DC 50:38 Q&A: Muse or Grok, and the Anthropic IPO 53:07 Q&A: Amazon in 2027 and agentic commerce 55:29 Q&A: why crypto has not rallied 56:40 Micron's cycle and the close #FuturumEquities #DanielNewman #ShayBoloor #Meta #AMD #Stocks 58:28 Meta is up big on Muse. AMD is priced for no mistakes. Micron is next.

Meta is up big on Muse. AMD is priced for no mistakes. Micron is next.

Futurum Equities September 25, 2026 11:09 pm

Meta's Muse hit No. 1 on the US App Store ten days after launch, cybersecurity had its best week of the year on AI fear, and the Fed hiked for the first time since 2023 with both hosts on record for no hike. Daniel Newman and Shay Boloor open episode 57 a week after the pace the frontier call and grade what it changed. Daniel brings ETR survey data from about 50 enterprise CIOs: the buyers of AI are fine with a slowdown, want third-party evaluation, and will pay a premium for the testing. They cannot keep up with the releases, he says, and both hosts call the labs' slowdown push regulatory capture ahead of the OpenAI and Anthropic IPOs, which Shay has zero plans to buy. Nebius raised prices about 20%, Anthropic wants 5 gigawatts by year end, and Daniel repeats that the compute is the moat. Daniel calls the hike bullish for cash holders and a problem for anyone on a credit card or a 7.5% mortgage. Shay's rule is to separate demand from financing: the buildout does not slow, the neo clouds that borrow to build get replaced by Microsoft and Meta funding from cash flow. META is Shay's case of monetizing intelligence a company already owns. Muse turns a hotel ad into a booked reservation, which moves Meta from creating demand to participating in the transaction. Daniel calls it the Apple move Apple never made. Cybersecurity, the show's application winner call since the SaaSpocalypse, is now fully priced by both hosts' account. Daniel says the opportunity was the Mythos selloff. Shay holds CRWD and NET at about 8% of his portfolio, counts them at 100x free cash flow, and would not be surprised if they go nowhere for two years. His replacement trade is memory and power: MU, VRT, ON and NBIS, with SITM, AEHR and CRDO as the less obvious bottlenecks for 2027. Daniel backs memory, calls NVDA undervalued for the systems it builds, and puts META and GOOGL back on his list as stack owners. The Q&A covers whether the hike changes either host's buying, the 2027 themes, Daniel's long healthcare in OSCR and HIMS, ORCL as a 2028 story and a 0.3% trade, DELL at 21x against MU at 6x, and the first lightning round: short DELL and INTC, long MU and NVDA. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. If this one helped, a like puts the show in front of more investors. Subscribe for next Friday's live, and tell us in the comments whether you are with Shay on memory or holding your cyber names. Chapters: 00:00 Episode 57: still standing after the slowdown week 00:30 ETR's CIO survey: enterprise buyers are fine with slowing down 04:13 Shay: enterprises will pay a premium for third-party testing 04:27 Open models at pennies on the dollar and the regulatory capture question 05:40 Shay: pure gamesmanship ahead of the OpenAI and Anthropic IPOs 07:21 Daniel: every FUD moment has been a buying opportunity 07:35 Safety needs more compute: Nebius prices, Anthropic's 5 GW, compute is the moat 09:19 All-In, Dreamforce, and why Meta and Google already went slow 10:58 Shay: the smog check analogy and zero plans to buy either IPO 12:31 Daniel: overcooked valuations and the calls we got wrong 13:35 The rate hike both hosts got wrong 14:29 Who the hike hurts: credit cards, car loans, 7.5% mortgages 16:45 Shay: higher for longer through 2027 meets the first AI job trim 18:33 Bad economy, good market: separate demand from financing 19:59 Housing volume, the oil shock, and a Middle East resolution 21:52 The bond market is trading the war, then Meta and Muse at No. 1 23:29 Daniel: Muse is the Apple move Apple never made 24:58 Shay: from creating demand to participating in the transaction 26:43 Daniel on Zuckerberg: right on compute, AI and agents 27:47 Cybersecurity: the application winner at COVID level valuations 29:10 Daniel: two pure plays, a small pond with big fish 31:04 Q: are cyber stocks still worth buying? Shay: overpriced 32:18 The opportunity was the Mythos selloff 32:49 Shay: CrowdStrike and Cloudflare may not move for two years, memory and power are forgotten 34:09 Daniel: software outside security, Nebius at 4x 2028 EBITDA, own the stack 37:33 Shay: the previous winners can be the new winners 38:51 Q&A: does the hike make you cautious? Shay's QQQ 21 week EMA plan 40:31 Daniel: a bullish hike, 95% invested 41:44 Would the market be red without a hike? Fed independence 43:05 Q&A: 2027 themes, AI power, SiTime, testing, Credo 44:03 Daniel: infrastructure, energy, and long healthcare 45:47 Q&A: Oracle is a 2028 story and a 0.3% trade 47:07 Daniel: the $650B backlog and the price leader's balance sheet 49:02 Q&A: Dell at 21x vs Micron at 6x, one of those has to be wrong 50:09 Daniel: Dell and HPE fully valued, Nvidia at $10T is only a double 52:10 Lightning round: short Dell and Intel, long Micron and Nvidia 53:28 Wrap: a quiet FOMC week #FuturumEquities #DanielNewman #ShayBoloor #AI #Stocks #Meta 55:26 AI labs asked to slow down. The Fed hiked. Cybersecurity won the week. | Ep 57

AI labs asked to slow down. The Fed hiked. Cybersecurity won the week. | Ep 57

Futurum Equities September 18, 2026 11:04 am

An AI doom post from a former Anthropic researcher passed 100 million views this week, and Daniel Newman and Shay Boloor open episode 56 by grading it against what got signed and reported. Oracle grew cloud infrastructure revenue 121%, Amazon tied warrants on Qualcomm stock to as much as $60 billion of chip orders, and Meta's Muse agent ranked fourth on Signal65's Pinnacle benchmark. Daniel reads the Jacob Coxon post as a media event with nothing actioned after the tour. Shay files it under the doomsday narratives the market has cycled through and names the risks he actually watches: power constraints, capital intensity and higher rates for longer. Oracle is the test case. Daniel calls 121% cloud infrastructure growth the pace setter among hyperscalers and still cannot get past the cash Oracle does not have. Shay says nothing in the quarter changed the narrative on ORCL: half of a $664 billion backlog converts inside three years, the GPU fleet runs at 98% utilization, free cash flow is still negative, and only time fixes the stock. That leads into the dislocation both hosts see in DELL at 23x forward earnings, more than NVDA, while memory names like MU trade as a commodity. Daniel's answer is cash conversion. Shay would rather own memory. Qualcomm's segment is Shay's frame: QCOM only has to be at the table, and AWS just gave it a multi-generation seat in custom inference silicon and 1.6 terabit optics. Shay reads the deal as bullish for the whole data movement group, CRDO, ALAB, AVGO and MRVL included. Muse sits fourth on Pinnacle behind three frontier models, and Shay's case for META is a move from monetizing intention to monetizing action. The Q&A covers what a Fed hike next week does to a capex pipeline both hosts call price inelastic, AeroVironment and Kratos after the drone selloff, Daniel's covered call and put spread playbook, Shay's rule of buying when the VIX is above 20 and his 10% cash position, the prisoner's dilemma that keeps every hyperscaler spending, and Apple as the safe haven with the largest install base. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. If this one helped you separate the doom narrative from the receipts, a like puts the show in front of more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay on memory or Daniel on Dell. Chapters: 00:00 Episode 56: a new doomsday scenario in the AI narrative 01:27 Hike odds near 80% and Shay's 5 to 1 bet that the Fed holds 02:09 Leaving IREN without leaving the neo cloud theme 02:46 The Jacob Coxon post: a six week Anthropic hire goes viral 03:47 Anthropic's doom playbook: Dario, Mythos and the misuse report 06:55 Cancel the IPO or slow the buildout: Sacks vs Jensen 07:56 Daniel's Terminator thesis: humanoids plus autonomous weapons 09:36 Best case a slowdown campaign, worst case a psyop 10:08 Shay's list of doomsday narratives and the risks that matter: power, capital, rates 12:07 Nothing got actioned: the media tour and the follower count 12:47 Oracle: 121% cloud infrastructure growth and the cash it does not have 15:31 Shay: no narrative change, half the RPO converts in three years, 98% utilization 18:30 First to blink: why no hyperscaler can cut capex first 20:42 Dell rewarded, memory not: the market's dislocation 22:10 Daniel: Dell is the pick and shovel, Oracle's revenue moves right 23:40 Dell at 23x is more than Nvidia. Why not memory instead 25:50 Qualcomm and Amazon: Daniel on the deal and the abundance era 30:14 Shay: Qualcomm only has to be at the table, plus 1.6T optics 32:22 This will not be the last hyperscaler deal 33:34 Meta's Muse: fourth on the Pinnacle benchmark at a fraction of the cost 35:13 Shay: Meta from monetizing intention to monetizing action, 25x is fair 36:36 Amazon and Meta were the 2026 best ideas: taking the L 37:57 The Apple foldable, an engagement and Grandpa Dan 40:26 Q&A: does a Fed hike slow the AI capex pipeline 43:56 Shay: strong commitments hold, tier three neo clouds get dinged 45:48 Q&A: AeroVironment and Kratos, record backlogs in a broken down group 47:49 Anduril's IPO as the drone trade's SpaceX moment 48:49 Q&A: Daniel's option spreads and covered call playbook 51:50 Shay: buy above VIX 20, 10% cash in each portfolio 53:02 Q&A: the capex prisoner's dilemma and the earnings bubble 56:16 Apple as the safe haven with the biggest install base 57:30 Wrap: Fed week ahead #FuturumEquities #DanielNewman #ShayBoloor #AI #Stocks #Oracle 57:41 The AI doom post passed 100M views. Power and rates are the real risk. | Ep. 56

The AI doom post passed 100M views. Power and rates are the real risk. | Ep. 56

Futurum Equities September 11, 2026 11:16 pm

Dell grew revenue 58% to a record $47 billion, raised its full year earnings guide to $25.50 a share, and closed up 16%. Every other AI beat this week sold off: Palo Alto down 9%, MongoDB down 13%, Credo down 20%. Daniel Newman and Shay Boloor spend episode 55, a Wednesday recording, on why valuation decided every reaction, and on the bar Broadcom, Snowflake and HPE have to clear after the close. The guys open on Dell. Daniel calls it a near perfect quarter: $60.9 billion of AI server orders, a $95 billion backlog, and AI factory customers up from 5,000 to 6,500 with almost all of that growth from enterprises. Shay adds the industry read, inference token volume up roughly 90x by 2030, and Daniel brings in Signal65's new Pinnacle benchmark on the cost of a correct outcome. Palo Alto grew next generation security ARR 63% to $9.1 billion and still sold off after walking in at 90x earnings against a 50 to 55x historical mean and a 76x COVID high. MongoDB grew Atlas 29% for the fifth straight quarter and fell 13% on a guide that implies deceleration, with the ETR spending data pulled live. Credo is where the hosts part ways: revenue grew 115%, the stock fell 20% on deceleration and customer concentration, Shay added 25% to a top five position at 22x forward earnings, and Daniel counters with Apple as the bellwether for what a hyperscaler eventually does to a dependent supplier. The preview block sets the bar for the afternoon: the one Broadcom number that matters, the $100 billion AI figure, networking at 40% of AI revenue, and Snowflake at 70x free cash flow. The Q&A covers Aehr Test Systems after a 50% drop in two weeks, the packaging trade in TSMC, Amkor and Intel, memory staying tight through 2030 and Micron vs SK Hynix, Shay's rule of never owning a business under 20% gross margins, the 10 year yield keeping the data center trade on pause until the war is over, and HPE's enterprise led quarter. The show is live every Friday. Bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on which beats get paid, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay or Daniel on Credo at 22x. Chapters: 00:00 Episode 55: a Wednesday recording before Shay's first family vacation 02:33 Earnings season is not over for growth investors 03:15 Dell: a near perfect quarter, up 16% 05:06 Inference tokens up 90x by 2030 and agentic AI as Dell's largest workload 06:01 Signal65's Pinnacle benchmark: cost per correct outcome 08:23 The AI bear tangent: why the doom keeps getting airtime 12:18 Palo Alto: strong quarter, red stock, 90x going in 14:24 Up 74% year to date and 120% in six months: trade vs invest 17:31 Palo Alto against its own history: 50 to 55x mean, 76x COVID high 18:59 MongoDB: Atlas 29% for the fifth straight quarter, the guide is the buzzkill 23:10 ETR data on MongoDB: pervasion up, spending intent cooling 25:00 Credo down 20%: set up to fail 25:59 Shay defends Credo at 22x and adds 25% to the position 30:03 Apple as the bellwether: how a hyperscaler squeezes a supplier 31:58 The rotation: software out, quality infrastructure in 32:25 Astera Labs: from 130x to 48x, Shay sold half at $415 to $420 34:45 Broadcom preview: the $100B AI number is the only one that matters 36:09 Custom silicon programs, networking at 40% of AI revenue, financing 39:07 Snowflake preview: 70x free cash flow against a 55x norm 43:02 Q&A: Aehr Test Systems fell from $150 to $79 in two weeks 44:46 Q&A: packaging stocks, TSMC, Amkor, Intel 46:20 Q&A: memory stays tight to 2030, Micron vs SK Hynix 48:40 Q&A: Dell at 25x earnings and the 20% gross margin rule 51:15 Q&A: 10 year yields and the data center trade on pause 53:40 Rates, bonds, and why housing has not broken 55:16 Q&A: HPE, enterprise tells the story 56:22 Wrap: Broadcom tonight, back to Fridays next week #FuturumEquities #DanielNewman #ShayBoloor #AI #Earnings #Stocks 57:18 Four AI sectors reported. Only the servers got paid. | Ep. 55

Four AI sectors reported. Only the servers got paid. | Ep. 55

Futurum Equities September 3, 2026 1:06 am

NVIDIA guided fiscal 2028 to 70% revenue growth against a street at 44%, and CFO Colette Kress says even that number is capped by supply, with demand running closer to 100%. Daniel Newman and Shay Boloor spend episode 54 on the NVDA print that took out two bear theses at once, Kevin Warsh's hawkish Jackson Hole debut, and the week Shay sold IREN with the power thesis still intact. The guys open on the print: a top line beat near $4 billion that sold off until the fiscal 2028 guide landed, roughly $40 billion of the $89 billion data center quarter now coming from outside the hyperscalers, and AWS ordering about 2 million more GPUs even while it builds its own silicon. Daniel spoke with Kress on earnings day, and her point was that the constraint is supply, so the competition thesis and the deceleration thesis died in the same print. The one crack is margins: a 74% guide against 75% expected, with memory inflation the stated reason and price increases coming next year. Shay's answer is that anything above 70% at this scale is remarkable, and he restates NVIDIA as the place his excess cash goes. Then the mood flips to IREN. One of the worst conference calls of the year, a deal the company legally cannot announce, and a capex raise well above what Nebius spends. Daniel is holding for what he thinks is underneath it. Shay sold his position on corporate governance, from the revenue goalposts to the Warriors jersey sponsorship, while conceding the power thesis has only gotten stronger. On Jackson Hole, neither host believes a Trump appointed chair raises rates, but the market prices perception, and a higher floor on rates pushed money toward quality: Amazon, Meta, Microsoft and ServiceNow all caught bids on a red day. The software conversation runs through Dario Amodei's televised concession next to Marc Benioff, the SaaS apocalypse graduating into what Shay calls the SaaSaPalooza, and CrowdStrike raising its ARR growth guide to 34% on AI security demand, with machine identities set to grow several hundred percent while headcount grows low single digits. The Q&A block covers Shay's beach stock rule for any position over 10%, Rocket Lab and AST SpaceMobile in the digestion phase after the SpaceX IPO, why nobody can undercut Nebius while power stays the bottleneck, memory stocks still priced as cyclical, and the Marvell print Shay left more bullish than he entered. The dip lists close it out: Lumentum, Nebius under $150, SpaceX under $80, and Pure Storage after its inventory build. Next week brings Credo, Palo Alto, MongoDB, GitLab, Broadcom, Snowflake, UiPath and Zscaler, plus a Monday episode where Shay and Brad go deep on Zscaler. The show is live every Friday. Bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the NVIDIA guide, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay or Daniel on IREN. Chapters: 00:00 Episode 54: Jackson Hole turns the screens red 00:54 Six Five Summit recap and Super Bowl earnings week 03:54 NVIDIA earnings: underwhelmed until it overwhelmed 05:37 Two bear theses die in one guide 07:53 Jensen answers every bear argument 09:46 Custom silicon and the non hyperscaler second wind 10:43 The Hugging Face rumor: the GitHub of AI 13:03 The margin crack: memory inflation 15:05 NVIDIA as the T-bill 15:41 IREN: the worst conference call of the year 19:14 Shay sells IREN on corporate governance 24:28 Owning power vs contracting power 26:31 Warsh's hawkish Jackson Hole debut 29:20 No hike coming, but the floor got higher 32:50 Did Dario just concede? 34:12 The SaaSaPalooza and the cybersecurity gold mine 36:35 Machine identities and the return to quality 38:26 Q&A: beach stocks and position sizing 41:34 Rocket Lab and ASTS after the SpaceX IPO 45:09 Q&A: can anyone undercut Nebius? 49:17 Q&A: memory stocks priced as cyclical 52:57 Cash levels into the midterms 53:53 Marvell: more bullish after the print 56:50 The dip list: Lumentum, Nebius, SpaceX, Pure Storage 59:19 Cyber's biggest days are ahead 1:00:58 Wrap: Monday's Zscaler deep dive #FuturumEquities #DanielNewman #ShayBoloor #AI #Nvidia #Stocks 1:1:32 NVIDIA guided 70% growth. Its CFO says it could double with supply. | Ep 54.

NVIDIA guided 70% growth. Its CFO says it could double with supply. | Ep 54.

Futurum Equities August 28, 2026 11:25 am

Anthropic put an October IPO on the table, and the neo clouds sold off hard into the news. Daniel Newman and Shay Boloor spend episode 53 on the narrow window where Anthropic can be the only public frontier AI stock, the margin test NVIDIA faces next Wednesday, and the Marvell warrant deal the market read backwards. The guys open on the data center politics souring the AI narrative into the midterms, then get into the week's headline: Anthropic closed a 39 point pervasion gap on OpenAI in 12 months per the ETR data and crossed over, while the early Q3 read shows OpenAI re-accelerating and Codex pervasion doubling in a quarter. Shay's read is a scarcity window: OpenAI can't list until 2027, so Anthropic gets nearly a year as the only direct public market expression of frontier AI. The catch is capital absorption. A raise that could approach $100 billion has to be funded from somewhere, and the obvious source is the AI proxies. That feeds the neo cloud conversation: Ken Griffin dumped most of the Situational Awareness book, CoreWeave borrows near 10% against a 15% return hurdle, and higher for longer rates squeeze every levered builder. On NVIDIA, the market is pricing the stock under a market multiple into Wednesday's report, and the counter is Vera Rubin holding 75% gross margins after Blackwell, SpaceX growing from roughly 4% of revenue toward 10% by the end of next year, and token margins climbing from 58% on Blackwell to an expected 90% on Feynman while paid token demand runs up 7x this year. Marvell gets the correction segment: the Google deal is about 7% dilution over 6.5 years against $120 billion of spend, TPU ecosystem attach rather than the socket, and roughly ten years of MRVL's entire run rate from one customer. The Q&A block covers OpenAI's consumer monetization problem, Coherent and the CPO revenue that only starts in the December quarter, Cipher at $15 after three round trips to $30, the poll showing 70% of Americans against data centers, and what Kevin Warsh's first Jackson Hole means two weeks before the September FOMC. The show is live every Friday. If you're tracking AI infrastructure, frontier AI, or the NVIDIA print, bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the Anthropic IPO, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you're buying the IPO on day one. Chapters: 00:00 Episode 53: the eye of the storm week 02:21 Data center politics hit the AI narrative 04:08 Golf courses use 3x the water of data centers 04:47 The spokesperson problem: Dario and Sam 06:59 Anthropic's October IPO: the scarcity window 10:10 The ETR data: Anthropic crossed OpenAI 11:35 Compute is the moat and Codex re-accelerates 13:20 Everybody buys the IPO anyway 14:17 A $100B raise has to come from somewhere 16:20 Why the neo clouds sold off 17:42 The cost of capital math on CoreWeave 20:20 Crypto rips 20% and the rotation trade 24:31 NVIDIA preview: the Super Bowl print 26:40 NVIDIA's earnings aren't for NVIDIA anymore 28:39 Daniel's checklist: margins, Rubin, customers 31:25 Jensen's answer to circular financing 34:59 The misleading token chart, corrected 37:36 Marvell and Google: the real warrant math 40:44 Q&A: OpenAI and consumer monetization 43:35 Q&A: Coherent and the CPO ramp 47:55 Q&A: Cipher and the neo cloud ladder 49:25 Q&A: the 70% anti data center poll 53:23 Jackson Hole: Warsh's debut 55:52 Cancer vaccines and AI's public case 59:31 Wrap: next week is a doozy #FuturumEquities #DanielNewman #ShayBoloor #AI #Nvidia #Stocks 1:0:12 Anthropic IPOs in October. The only frontier AI stock for a year. | Ep. 53

Anthropic IPOs in October. The only frontier AI stock for a year. | Ep. 53

Futurum Equities August 21, 2026 11:07 pm

A prominent memory bull flipped short term bearish over the weekend, and Sandisk answered with an investor day that guides to 80% gross margins out to 2030. Daniel Newman and Shay Boloor spend episode 52 on the memory wall, the week the GPU depreciation bear case broke, and Futurum's new $12.7 trillion AI capex number. The guys open on the memory debate: why every semiconductor and systems company is trying to engineer around the memory wall with stacking, optics, and software, and why Daniel still calls memory a 5 to 10 year constraint. Sandisk's investor day gets the full treatment: mid to high teens growth guided through fiscal 2030, 80% non GAAP gross margins, free cash flow at 50% of revenue, KV cache reaching 35% of AI data center NAND workloads by 2030, and the first HBF die coming in 2027. Shay explains his Micron 2027 call position and why memory stays rationed while GPU capacity commands premiums. Then the neo clouds: CoreWeave signed five year A100 contracts on silicon that is already six years old, grew its order book twice as fast as revenue, and passed $400 million in annualized non GPU business. Nebius ran the first Blackwell capacity auction and cleared 15% above its own best pricing, with management saying all of 2027 could book at those levels. Plus Applied Materials' largest sequential growth quarter in its 59 year history, Lenovo's 98% infrastructure growth, Shay's new Vertiv and ON Semiconductor positions for AI power, and the note that chips alone reach roughly $5 trillion of spend by the end of the decade. The Q&A block keeps growing and this week it earned the time: the humanoid robotics SPAC that fails the sniff test, the best AI energy stock between GE Vernova, Vertiv, and Bloom, Aehr Test Systems after a 15x run, AppLovin's make or break Q3, Cipher's messy quarter and the CEO stock sale, Credo against Astera Labs, and how Michael Burry and Jim Cramer keep getting the AI trade wrong. The show is live every Friday. If you're tracking AI infrastructure, memory, or the neo clouds, bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the memory trade, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments which constraint you think the dollars rotate to next. Chapters: 00:00 Episode 52 and the Six Five Summit slate 02:38 A memory bull flips bearish 04:05 The memory wall: a 5 to 10 year constraint 07:15 KV cache and memory beyond HBM 09:23 Sandisk investor day: the 2030 targets 11:14 Redesigning the memory hierarchy for inference 13:45 A100s contracted through 2030 15:25 CoreWeave: layers of revenue, real risks 18:48 Power is the gate: Vertiv and ON adds 21:01 Futurum's $12.7 trillion capex number 22:27 Nebius and the Blackwell auction 25:46 Selling puts on space names 28:49 ASTS and Rocket Lab: the execution risk era 31:34 Applied Materials, Cisco, and Lenovo prints 33:56 Elevated expectations, tepid reactions 36:06 Q&A: the humanoid robotics SPAC 38:37 Best AI energy stock: GEV, Vertiv, Bloom 40:42 Aehr Test Systems after a 15x run 42:06 AppLovin's make or break Q3 45:39 Cipher's miss and the CEO sale 48:29 Credo vs Astera Labs 51:22 Burry, Cramer, and the 5% that is overbuilt 55:18 SaaSpocalypse winners 57:43 The eye of the storm before NVIDIA #FuturumEquities #DanielNewman #ShayBoloor #AI #Semis #Earnings 58:20 Memory is a 5 to 10 year constraint. The 2030 guides just proved it. | Ep. 52

Memory is a 5 to 10 year constraint. The 2030 guides just proved it. | Ep. 52

Futurum Equities August 14, 2026 11:09 pm

SpaceX just delivered its first earnings report as a public company, and the number nobody modeled stole the whole show. Daniel Newman and Shay Boloor spend episode 51 on the print, the lockup everyone was selling ahead of, and why a weak jobs report has the entire AI trade breathing again. The guys go deep on SpaceX: 92% growth in an AI cloud business that was never supposed to exist at this scale, a revenue path from $44 billion this year toward $260 billion by 2029, and the math behind Shay buying the lockup fear at $109 to $110 before the print. Elon's on-record line that SpaceX runs only on Nvidia gets its own moment. From there it's Palantir's rerating: 93% growth, commercial up 150%, net dollar retention at 157%, roughly $6 billion of US commercial contract value closed, and 73 deals worth over $10 million each. Daniel calls the end of the SaaS apocalypse and explains why compute is the moat while Anthropic's street ARR pushes $85 billion. AMD gets the honest treatment: the MI350 and EPYC beat, why Lisa Su's conservative guide is the long game, and the Helios rack scale platform with its claimed 30% more tokens per dollar. Plus the no rate hike case under Kevin Warsh after a weak jobs Friday, Shay's new Vertiv position at $220 for AI power exposure, Micron over SK Hynix on HBM rate of change, and the reveal that Shay sold every share of his Tesla on a SpaceX merger theory. The back half is all viewer Q&A: the Palantir and Mercury digital twin partnership, Daniel's put selling playbook that pays him 5 to 10% of his portfolio a year in cash flow, the Oscar Health case study, T1 Energy and EOS Energy as the speculative power trades, and who actually wins when models get cheap. The show is live every Friday. If you're tracking AI infrastructure or trading this earnings season, bring your tickers and questions to the chat and the guys will take them on air. If this one helped you make sense of the tape, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and come tell Shay whether the Tesla exit was genius or tin hat in the comments. Chapters: 00:00 Episode 51: back from the Schwab event 04:19 Jobs Friday: the no rate hike case 09:40 Why weak data is bullish: the 10 year toward 4% 10:57 Buying the July panic and the gaslighting 15:02 Vertiv and the AI power gap 17:22 SpaceX's first earnings as a public company 20:04 The $44B to $260B ramp and the lockup buy 25:56 Sizing SpaceX and the space portfolio 28:12 AMD: Lisa Su's conservative guide and Helios 34:56 Palantir's rerating: 157% retention 38:48 The SaaSpocalypse is over 41:54 Models as commodities and the scarcity premium 44:26 Q&A: Palantir and Mercury's digital twins 46:28 Daniel's put selling playbook 51:37 T1 Energy, EOS, and Shay selling all his Tesla 54:19 Memory: Micron over SK Hynix 57:58 Who wins when models get cheap #FuturumEquities #DanielNewman #ShayBoloor #AI #SpaceX #Earnings 1:1:33 SpaceX guided to $260B by 2029. Shay bought the lockup fear. | Ep. 51

SpaceX guided to $260B by 2029. Shay bought the lockup fear. | Ep. 51

Futurum Equities August 7, 2026 11:26 am

Episode 50 lands in a strange tape. Google grew its cloud business 82% and the stock got sold anyway, and Daniel Newman and Shay Boloor spend the show on what that selling actually means for everyone holding the AI trade. The guys go deep on Alphabet's print: a $25 billion cloud quarter growing 82% with 36% operating margins, a $15 billion capex raise, and the earnings math behind Shay's call that Google is "a great business, bad investment" until EPS clears $20 again in 2029. Nvidia at 18x earnings gets the other side of that trade. From there it's the $11.7 trillion cumulative AI capex bull case through 2030, TSMC raising capex $8 billion in a quarter while flagging a bigger 2027, Intel's blowout on server CPU pricing and Shay's rule that INTC has to prove an external foundry customer to earn its price above $100, Tesla converting record deliveries into a 1% operating margin, Micron and the HBM demand curve the market keeps mispricing, and IREN's $2.8 billion AI cloud contract with roughly half the GPU capex prepaid by customers. Plus SpaceX lockup math and entry levels, Meta at 18x with Shay's $1,200 thesis, and Jensen Huang joining X mid show. The back half is all viewer Q&A: ServiceNow as the benchmark for AI software, Shay's tier one and tier two software framework, whether Meta is the best Mag 7 buy right now, IREN position sizing, and the drone bucket ahead of the Anduril IPO. The show is live every Friday. If you're tracking AI infrastructure or trading this earnings season, bring your tickers and questions to the chat and the guys will take them on air. If this one helped you make sense of the tape, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and come argue with Shay about NVDA at $100 in the comments. Chapters: 00:00 Episode 50: Jensen joins X and the inverse Cramer tape 01:31 Strong prints keep getting sold 03:15 AI turns political and the Apple tell 07:47 The $11.7T capex bull case 09:42 Bottleneck bros and the rate of change problem 12:56 Semis supply constrained for 4 to 5 years 16:10 Google vs Nvidia: the two year problem 18:56 Daniel's defense: capex is existential 21:37 Memory: the market is pricing the peak wrong 23:20 TSMC raises capex, the grown up signal 26:50 Intel's $100 foundry gate 34:07 Tesla: record deliveries, 1% margin 37:47 SpaceX: merger math, lockups, entry levels 43:31 Q&A: ServiceNow and the software tiers 48:57 Meta at 18x and the $1,200 case 52:46 IREN's $2.8B contract 55:45 Drones, Anduril, ASTS 59:06 The put selling playbook and NVDA $100 on record #FuturumEquities #DanielNewman #ShayBoloor #AI #Semis #Earnings 1:2:40 Google grew cloud 82% and still got sold. Where the money went. | Ep 50

Google grew cloud 82% and still got sold. Where the money went. | Ep 50

Futurum Equities July 24, 2026 11:19 am

sentiment

powered by

DAILY RIP

powered by

Meta got the re-rating both hosts called at the start of 2026. META closed Thursday up 27% from the day Muse launched on September 8, AMD crossed $1 trillion on September 21, and Micron reports a 14-week quarter on September 30. Daniel Newman and Shay Boloor take the three in order on episode 58, then take your questions. Daniel says the AI winner will be the company with the most compute and energy at its disposal. Meta kept buying CPUs, GPUs and its own accelerators while the market wrote it off, and Muse puts that compute to work. Muse is free up to 100 million tokens a week, with $20 and $100 tiers above that, and Meta plans to take a small fee on each transaction. Daniel calls the transaction model the move Apple should have made. Shay's four qualities of an AI winner are proprietary data, install base, compute and energy. He says Meta had three of them all along, argues it now deserves a Google-like multiple for a consumer AI product, and warns that the next quarterly report will not show any of it. On AMD, Shay counts 40x 2027 earnings against 15x for NVDA and warns that warrants held by Meta and OpenAI will lift the share count in any 2030 model. Daniel says the market treats Nvidia as the incumbent with everything to lose and AMD as the entrant with everything to gain, and that the stock is priced as if Helios, AMD's first rack-scale system, carries almost no execution risk. On MU, Shay explains why the guide could look light: fiscal Q4 had 14 weeks and Q1 has 13. He is watching strategic customer agreements, HBM4 share and capex timing, and has not heard one plausible sign that memory pricing power is eroding. Daniel says Futurum's 2030 CPU forecast has more than doubled since January, walks through the follow-up reporting on the Coxon post, and says every AI scare so far has ended in a rally. He is long VKTX after Viking's maintenance data and says Novo Nordisk should buy it. The Q&A covers DigitalOcean and its move past GPU rentals, Everpure after its investor day, the rest of the year with the 10-year at its highest since 2007, Shay's agentic AI picks for 2027, AI power and 800-volt DC, Muse against Grok, the Anthropic IPO, Amazon in 2027, and why crypto has not rallied more. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. Chapters: 00:00 Episode 58: Daniel is back from Snapdragon Summit 04:09 Meta's re-rating on Muse 14:49 AMD at $1 trillion against Nvidia 22:31 Micron preview: the 14-week quarter 28:01 Futurum's CPU forecast doubled, and the Coxon post follow-up 34:54 Viking Therapeutics after its maintenance data 37:33 Q&A: DigitalOcean beyond GPU rentals 39:34 Q&A: Everpure after its investor day 42:48 Q&A: the rest of the year, the war and 5% yields 47:19 Q&A: agentic AI picks for 2027 49:25 AI power and 800-volt DC 50:38 Q&A: Muse or Grok, and the Anthropic IPO 53:07 Q&A: Amazon in 2027 and agentic commerce 55:29 Q&A: why crypto has not rallied 56:40 Micron's cycle and the close #FuturumEquities #DanielNewman #ShayBoloor #Meta #AMD #Stocks 58:28 Meta is up big on Muse. AMD is priced for no mistakes. Micron is next.

Meta is up big on Muse. AMD is priced for no mistakes. Micron is next.

Futurum Equities September 25, 2026 11:09 pm

Meta's Muse hit No. 1 on the US App Store ten days after launch, cybersecurity had its best week of the year on AI fear, and the Fed hiked for the first time since 2023 with both hosts on record for no hike. Daniel Newman and Shay Boloor open episode 57 a week after the pace the frontier call and grade what it changed. Daniel brings ETR survey data from about 50 enterprise CIOs: the buyers of AI are fine with a slowdown, want third-party evaluation, and will pay a premium for the testing. They cannot keep up with the releases, he says, and both hosts call the labs' slowdown push regulatory capture ahead of the OpenAI and Anthropic IPOs, which Shay has zero plans to buy. Nebius raised prices about 20%, Anthropic wants 5 gigawatts by year end, and Daniel repeats that the compute is the moat. Daniel calls the hike bullish for cash holders and a problem for anyone on a credit card or a 7.5% mortgage. Shay's rule is to separate demand from financing: the buildout does not slow, the neo clouds that borrow to build get replaced by Microsoft and Meta funding from cash flow. META is Shay's case of monetizing intelligence a company already owns. Muse turns a hotel ad into a booked reservation, which moves Meta from creating demand to participating in the transaction. Daniel calls it the Apple move Apple never made. Cybersecurity, the show's application winner call since the SaaSpocalypse, is now fully priced by both hosts' account. Daniel says the opportunity was the Mythos selloff. Shay holds CRWD and NET at about 8% of his portfolio, counts them at 100x free cash flow, and would not be surprised if they go nowhere for two years. His replacement trade is memory and power: MU, VRT, ON and NBIS, with SITM, AEHR and CRDO as the less obvious bottlenecks for 2027. Daniel backs memory, calls NVDA undervalued for the systems it builds, and puts META and GOOGL back on his list as stack owners. The Q&A covers whether the hike changes either host's buying, the 2027 themes, Daniel's long healthcare in OSCR and HIMS, ORCL as a 2028 story and a 0.3% trade, DELL at 21x against MU at 6x, and the first lightning round: short DELL and INTC, long MU and NVDA. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. If this one helped, a like puts the show in front of more investors. Subscribe for next Friday's live, and tell us in the comments whether you are with Shay on memory or holding your cyber names. Chapters: 00:00 Episode 57: still standing after the slowdown week 00:30 ETR's CIO survey: enterprise buyers are fine with slowing down 04:13 Shay: enterprises will pay a premium for third-party testing 04:27 Open models at pennies on the dollar and the regulatory capture question 05:40 Shay: pure gamesmanship ahead of the OpenAI and Anthropic IPOs 07:21 Daniel: every FUD moment has been a buying opportunity 07:35 Safety needs more compute: Nebius prices, Anthropic's 5 GW, compute is the moat 09:19 All-In, Dreamforce, and why Meta and Google already went slow 10:58 Shay: the smog check analogy and zero plans to buy either IPO 12:31 Daniel: overcooked valuations and the calls we got wrong 13:35 The rate hike both hosts got wrong 14:29 Who the hike hurts: credit cards, car loans, 7.5% mortgages 16:45 Shay: higher for longer through 2027 meets the first AI job trim 18:33 Bad economy, good market: separate demand from financing 19:59 Housing volume, the oil shock, and a Middle East resolution 21:52 The bond market is trading the war, then Meta and Muse at No. 1 23:29 Daniel: Muse is the Apple move Apple never made 24:58 Shay: from creating demand to participating in the transaction 26:43 Daniel on Zuckerberg: right on compute, AI and agents 27:47 Cybersecurity: the application winner at COVID level valuations 29:10 Daniel: two pure plays, a small pond with big fish 31:04 Q: are cyber stocks still worth buying? Shay: overpriced 32:18 The opportunity was the Mythos selloff 32:49 Shay: CrowdStrike and Cloudflare may not move for two years, memory and power are forgotten 34:09 Daniel: software outside security, Nebius at 4x 2028 EBITDA, own the stack 37:33 Shay: the previous winners can be the new winners 38:51 Q&A: does the hike make you cautious? Shay's QQQ 21 week EMA plan 40:31 Daniel: a bullish hike, 95% invested 41:44 Would the market be red without a hike? Fed independence 43:05 Q&A: 2027 themes, AI power, SiTime, testing, Credo 44:03 Daniel: infrastructure, energy, and long healthcare 45:47 Q&A: Oracle is a 2028 story and a 0.3% trade 47:07 Daniel: the $650B backlog and the price leader's balance sheet 49:02 Q&A: Dell at 21x vs Micron at 6x, one of those has to be wrong 50:09 Daniel: Dell and HPE fully valued, Nvidia at $10T is only a double 52:10 Lightning round: short Dell and Intel, long Micron and Nvidia 53:28 Wrap: a quiet FOMC week #FuturumEquities #DanielNewman #ShayBoloor #AI #Stocks #Meta 55:26 AI labs asked to slow down. The Fed hiked. Cybersecurity won the week. | Ep 57

AI labs asked to slow down. The Fed hiked. Cybersecurity won the week. | Ep 57

Futurum Equities September 18, 2026 11:04 am

An AI doom post from a former Anthropic researcher passed 100 million views this week, and Daniel Newman and Shay Boloor open episode 56 by grading it against what got signed and reported. Oracle grew cloud infrastructure revenue 121%, Amazon tied warrants on Qualcomm stock to as much as $60 billion of chip orders, and Meta's Muse agent ranked fourth on Signal65's Pinnacle benchmark. Daniel reads the Jacob Coxon post as a media event with nothing actioned after the tour. Shay files it under the doomsday narratives the market has cycled through and names the risks he actually watches: power constraints, capital intensity and higher rates for longer. Oracle is the test case. Daniel calls 121% cloud infrastructure growth the pace setter among hyperscalers and still cannot get past the cash Oracle does not have. Shay says nothing in the quarter changed the narrative on ORCL: half of a $664 billion backlog converts inside three years, the GPU fleet runs at 98% utilization, free cash flow is still negative, and only time fixes the stock. That leads into the dislocation both hosts see in DELL at 23x forward earnings, more than NVDA, while memory names like MU trade as a commodity. Daniel's answer is cash conversion. Shay would rather own memory. Qualcomm's segment is Shay's frame: QCOM only has to be at the table, and AWS just gave it a multi-generation seat in custom inference silicon and 1.6 terabit optics. Shay reads the deal as bullish for the whole data movement group, CRDO, ALAB, AVGO and MRVL included. Muse sits fourth on Pinnacle behind three frontier models, and Shay's case for META is a move from monetizing intention to monetizing action. The Q&A covers what a Fed hike next week does to a capex pipeline both hosts call price inelastic, AeroVironment and Kratos after the drone selloff, Daniel's covered call and put spread playbook, Shay's rule of buying when the VIX is above 20 and his 10% cash position, the prisoner's dilemma that keeps every hyperscaler spending, and Apple as the safe haven with the largest install base. The show is live every Friday at 1 PM ET. Bring your tickers and questions to the chat and the guys will take them on air. If this one helped you separate the doom narrative from the receipts, a like puts the show in front of more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay on memory or Daniel on Dell. Chapters: 00:00 Episode 56: a new doomsday scenario in the AI narrative 01:27 Hike odds near 80% and Shay's 5 to 1 bet that the Fed holds 02:09 Leaving IREN without leaving the neo cloud theme 02:46 The Jacob Coxon post: a six week Anthropic hire goes viral 03:47 Anthropic's doom playbook: Dario, Mythos and the misuse report 06:55 Cancel the IPO or slow the buildout: Sacks vs Jensen 07:56 Daniel's Terminator thesis: humanoids plus autonomous weapons 09:36 Best case a slowdown campaign, worst case a psyop 10:08 Shay's list of doomsday narratives and the risks that matter: power, capital, rates 12:07 Nothing got actioned: the media tour and the follower count 12:47 Oracle: 121% cloud infrastructure growth and the cash it does not have 15:31 Shay: no narrative change, half the RPO converts in three years, 98% utilization 18:30 First to blink: why no hyperscaler can cut capex first 20:42 Dell rewarded, memory not: the market's dislocation 22:10 Daniel: Dell is the pick and shovel, Oracle's revenue moves right 23:40 Dell at 23x is more than Nvidia. Why not memory instead 25:50 Qualcomm and Amazon: Daniel on the deal and the abundance era 30:14 Shay: Qualcomm only has to be at the table, plus 1.6T optics 32:22 This will not be the last hyperscaler deal 33:34 Meta's Muse: fourth on the Pinnacle benchmark at a fraction of the cost 35:13 Shay: Meta from monetizing intention to monetizing action, 25x is fair 36:36 Amazon and Meta were the 2026 best ideas: taking the L 37:57 The Apple foldable, an engagement and Grandpa Dan 40:26 Q&A: does a Fed hike slow the AI capex pipeline 43:56 Shay: strong commitments hold, tier three neo clouds get dinged 45:48 Q&A: AeroVironment and Kratos, record backlogs in a broken down group 47:49 Anduril's IPO as the drone trade's SpaceX moment 48:49 Q&A: Daniel's option spreads and covered call playbook 51:50 Shay: buy above VIX 20, 10% cash in each portfolio 53:02 Q&A: the capex prisoner's dilemma and the earnings bubble 56:16 Apple as the safe haven with the biggest install base 57:30 Wrap: Fed week ahead #FuturumEquities #DanielNewman #ShayBoloor #AI #Stocks #Oracle 57:41 The AI doom post passed 100M views. Power and rates are the real risk. | Ep. 56

The AI doom post passed 100M views. Power and rates are the real risk. | Ep. 56

Futurum Equities September 11, 2026 11:16 pm

Dell grew revenue 58% to a record $47 billion, raised its full year earnings guide to $25.50 a share, and closed up 16%. Every other AI beat this week sold off: Palo Alto down 9%, MongoDB down 13%, Credo down 20%. Daniel Newman and Shay Boloor spend episode 55, a Wednesday recording, on why valuation decided every reaction, and on the bar Broadcom, Snowflake and HPE have to clear after the close. The guys open on Dell. Daniel calls it a near perfect quarter: $60.9 billion of AI server orders, a $95 billion backlog, and AI factory customers up from 5,000 to 6,500 with almost all of that growth from enterprises. Shay adds the industry read, inference token volume up roughly 90x by 2030, and Daniel brings in Signal65's new Pinnacle benchmark on the cost of a correct outcome. Palo Alto grew next generation security ARR 63% to $9.1 billion and still sold off after walking in at 90x earnings against a 50 to 55x historical mean and a 76x COVID high. MongoDB grew Atlas 29% for the fifth straight quarter and fell 13% on a guide that implies deceleration, with the ETR spending data pulled live. Credo is where the hosts part ways: revenue grew 115%, the stock fell 20% on deceleration and customer concentration, Shay added 25% to a top five position at 22x forward earnings, and Daniel counters with Apple as the bellwether for what a hyperscaler eventually does to a dependent supplier. The preview block sets the bar for the afternoon: the one Broadcom number that matters, the $100 billion AI figure, networking at 40% of AI revenue, and Snowflake at 70x free cash flow. The Q&A covers Aehr Test Systems after a 50% drop in two weeks, the packaging trade in TSMC, Amkor and Intel, memory staying tight through 2030 and Micron vs SK Hynix, Shay's rule of never owning a business under 20% gross margins, the 10 year yield keeping the data center trade on pause until the war is over, and HPE's enterprise led quarter. The show is live every Friday. Bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on which beats get paid, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay or Daniel on Credo at 22x. Chapters: 00:00 Episode 55: a Wednesday recording before Shay's first family vacation 02:33 Earnings season is not over for growth investors 03:15 Dell: a near perfect quarter, up 16% 05:06 Inference tokens up 90x by 2030 and agentic AI as Dell's largest workload 06:01 Signal65's Pinnacle benchmark: cost per correct outcome 08:23 The AI bear tangent: why the doom keeps getting airtime 12:18 Palo Alto: strong quarter, red stock, 90x going in 14:24 Up 74% year to date and 120% in six months: trade vs invest 17:31 Palo Alto against its own history: 50 to 55x mean, 76x COVID high 18:59 MongoDB: Atlas 29% for the fifth straight quarter, the guide is the buzzkill 23:10 ETR data on MongoDB: pervasion up, spending intent cooling 25:00 Credo down 20%: set up to fail 25:59 Shay defends Credo at 22x and adds 25% to the position 30:03 Apple as the bellwether: how a hyperscaler squeezes a supplier 31:58 The rotation: software out, quality infrastructure in 32:25 Astera Labs: from 130x to 48x, Shay sold half at $415 to $420 34:45 Broadcom preview: the $100B AI number is the only one that matters 36:09 Custom silicon programs, networking at 40% of AI revenue, financing 39:07 Snowflake preview: 70x free cash flow against a 55x norm 43:02 Q&A: Aehr Test Systems fell from $150 to $79 in two weeks 44:46 Q&A: packaging stocks, TSMC, Amkor, Intel 46:20 Q&A: memory stays tight to 2030, Micron vs SK Hynix 48:40 Q&A: Dell at 25x earnings and the 20% gross margin rule 51:15 Q&A: 10 year yields and the data center trade on pause 53:40 Rates, bonds, and why housing has not broken 55:16 Q&A: HPE, enterprise tells the story 56:22 Wrap: Broadcom tonight, back to Fridays next week #FuturumEquities #DanielNewman #ShayBoloor #AI #Earnings #Stocks 57:18 Four AI sectors reported. Only the servers got paid. | Ep. 55

Four AI sectors reported. Only the servers got paid. | Ep. 55

Futurum Equities September 3, 2026 1:06 am

NVIDIA guided fiscal 2028 to 70% revenue growth against a street at 44%, and CFO Colette Kress says even that number is capped by supply, with demand running closer to 100%. Daniel Newman and Shay Boloor spend episode 54 on the NVDA print that took out two bear theses at once, Kevin Warsh's hawkish Jackson Hole debut, and the week Shay sold IREN with the power thesis still intact. The guys open on the print: a top line beat near $4 billion that sold off until the fiscal 2028 guide landed, roughly $40 billion of the $89 billion data center quarter now coming from outside the hyperscalers, and AWS ordering about 2 million more GPUs even while it builds its own silicon. Daniel spoke with Kress on earnings day, and her point was that the constraint is supply, so the competition thesis and the deceleration thesis died in the same print. The one crack is margins: a 74% guide against 75% expected, with memory inflation the stated reason and price increases coming next year. Shay's answer is that anything above 70% at this scale is remarkable, and he restates NVIDIA as the place his excess cash goes. Then the mood flips to IREN. One of the worst conference calls of the year, a deal the company legally cannot announce, and a capex raise well above what Nebius spends. Daniel is holding for what he thinks is underneath it. Shay sold his position on corporate governance, from the revenue goalposts to the Warriors jersey sponsorship, while conceding the power thesis has only gotten stronger. On Jackson Hole, neither host believes a Trump appointed chair raises rates, but the market prices perception, and a higher floor on rates pushed money toward quality: Amazon, Meta, Microsoft and ServiceNow all caught bids on a red day. The software conversation runs through Dario Amodei's televised concession next to Marc Benioff, the SaaS apocalypse graduating into what Shay calls the SaaSaPalooza, and CrowdStrike raising its ARR growth guide to 34% on AI security demand, with machine identities set to grow several hundred percent while headcount grows low single digits. The Q&A block covers Shay's beach stock rule for any position over 10%, Rocket Lab and AST SpaceMobile in the digestion phase after the SpaceX IPO, why nobody can undercut Nebius while power stays the bottleneck, memory stocks still priced as cyclical, and the Marvell print Shay left more bullish than he entered. The dip lists close it out: Lumentum, Nebius under $150, SpaceX under $80, and Pure Storage after its inventory build. Next week brings Credo, Palo Alto, MongoDB, GitLab, Broadcom, Snowflake, UiPath and Zscaler, plus a Monday episode where Shay and Brad go deep on Zscaler. The show is live every Friday. Bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the NVIDIA guide, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you side with Shay or Daniel on IREN. Chapters: 00:00 Episode 54: Jackson Hole turns the screens red 00:54 Six Five Summit recap and Super Bowl earnings week 03:54 NVIDIA earnings: underwhelmed until it overwhelmed 05:37 Two bear theses die in one guide 07:53 Jensen answers every bear argument 09:46 Custom silicon and the non hyperscaler second wind 10:43 The Hugging Face rumor: the GitHub of AI 13:03 The margin crack: memory inflation 15:05 NVIDIA as the T-bill 15:41 IREN: the worst conference call of the year 19:14 Shay sells IREN on corporate governance 24:28 Owning power vs contracting power 26:31 Warsh's hawkish Jackson Hole debut 29:20 No hike coming, but the floor got higher 32:50 Did Dario just concede? 34:12 The SaaSaPalooza and the cybersecurity gold mine 36:35 Machine identities and the return to quality 38:26 Q&A: beach stocks and position sizing 41:34 Rocket Lab and ASTS after the SpaceX IPO 45:09 Q&A: can anyone undercut Nebius? 49:17 Q&A: memory stocks priced as cyclical 52:57 Cash levels into the midterms 53:53 Marvell: more bullish after the print 56:50 The dip list: Lumentum, Nebius, SpaceX, Pure Storage 59:19 Cyber's biggest days are ahead 1:00:58 Wrap: Monday's Zscaler deep dive #FuturumEquities #DanielNewman #ShayBoloor #AI #Nvidia #Stocks 1:1:32 NVIDIA guided 70% growth. Its CFO says it could double with supply. | Ep 54.

NVIDIA guided 70% growth. Its CFO says it could double with supply. | Ep 54.

Futurum Equities August 28, 2026 11:25 am

Anthropic put an October IPO on the table, and the neo clouds sold off hard into the news. Daniel Newman and Shay Boloor spend episode 53 on the narrow window where Anthropic can be the only public frontier AI stock, the margin test NVIDIA faces next Wednesday, and the Marvell warrant deal the market read backwards. The guys open on the data center politics souring the AI narrative into the midterms, then get into the week's headline: Anthropic closed a 39 point pervasion gap on OpenAI in 12 months per the ETR data and crossed over, while the early Q3 read shows OpenAI re-accelerating and Codex pervasion doubling in a quarter. Shay's read is a scarcity window: OpenAI can't list until 2027, so Anthropic gets nearly a year as the only direct public market expression of frontier AI. The catch is capital absorption. A raise that could approach $100 billion has to be funded from somewhere, and the obvious source is the AI proxies. That feeds the neo cloud conversation: Ken Griffin dumped most of the Situational Awareness book, CoreWeave borrows near 10% against a 15% return hurdle, and higher for longer rates squeeze every levered builder. On NVIDIA, the market is pricing the stock under a market multiple into Wednesday's report, and the counter is Vera Rubin holding 75% gross margins after Blackwell, SpaceX growing from roughly 4% of revenue toward 10% by the end of next year, and token margins climbing from 58% on Blackwell to an expected 90% on Feynman while paid token demand runs up 7x this year. Marvell gets the correction segment: the Google deal is about 7% dilution over 6.5 years against $120 billion of spend, TPU ecosystem attach rather than the socket, and roughly ten years of MRVL's entire run rate from one customer. The Q&A block covers OpenAI's consumer monetization problem, Coherent and the CPO revenue that only starts in the December quarter, Cipher at $15 after three round trips to $30, the poll showing 70% of Americans against data centers, and what Kevin Warsh's first Jackson Hole means two weeks before the September FOMC. The show is live every Friday. If you're tracking AI infrastructure, frontier AI, or the NVIDIA print, bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the Anthropic IPO, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments whether you're buying the IPO on day one. Chapters: 00:00 Episode 53: the eye of the storm week 02:21 Data center politics hit the AI narrative 04:08 Golf courses use 3x the water of data centers 04:47 The spokesperson problem: Dario and Sam 06:59 Anthropic's October IPO: the scarcity window 10:10 The ETR data: Anthropic crossed OpenAI 11:35 Compute is the moat and Codex re-accelerates 13:20 Everybody buys the IPO anyway 14:17 A $100B raise has to come from somewhere 16:20 Why the neo clouds sold off 17:42 The cost of capital math on CoreWeave 20:20 Crypto rips 20% and the rotation trade 24:31 NVIDIA preview: the Super Bowl print 26:40 NVIDIA's earnings aren't for NVIDIA anymore 28:39 Daniel's checklist: margins, Rubin, customers 31:25 Jensen's answer to circular financing 34:59 The misleading token chart, corrected 37:36 Marvell and Google: the real warrant math 40:44 Q&A: OpenAI and consumer monetization 43:35 Q&A: Coherent and the CPO ramp 47:55 Q&A: Cipher and the neo cloud ladder 49:25 Q&A: the 70% anti data center poll 53:23 Jackson Hole: Warsh's debut 55:52 Cancer vaccines and AI's public case 59:31 Wrap: next week is a doozy #FuturumEquities #DanielNewman #ShayBoloor #AI #Nvidia #Stocks 1:0:12 Anthropic IPOs in October. The only frontier AI stock for a year. | Ep. 53

Anthropic IPOs in October. The only frontier AI stock for a year. | Ep. 53

Futurum Equities August 21, 2026 11:07 pm

A prominent memory bull flipped short term bearish over the weekend, and Sandisk answered with an investor day that guides to 80% gross margins out to 2030. Daniel Newman and Shay Boloor spend episode 52 on the memory wall, the week the GPU depreciation bear case broke, and Futurum's new $12.7 trillion AI capex number. The guys open on the memory debate: why every semiconductor and systems company is trying to engineer around the memory wall with stacking, optics, and software, and why Daniel still calls memory a 5 to 10 year constraint. Sandisk's investor day gets the full treatment: mid to high teens growth guided through fiscal 2030, 80% non GAAP gross margins, free cash flow at 50% of revenue, KV cache reaching 35% of AI data center NAND workloads by 2030, and the first HBF die coming in 2027. Shay explains his Micron 2027 call position and why memory stays rationed while GPU capacity commands premiums. Then the neo clouds: CoreWeave signed five year A100 contracts on silicon that is already six years old, grew its order book twice as fast as revenue, and passed $400 million in annualized non GPU business. Nebius ran the first Blackwell capacity auction and cleared 15% above its own best pricing, with management saying all of 2027 could book at those levels. Plus Applied Materials' largest sequential growth quarter in its 59 year history, Lenovo's 98% infrastructure growth, Shay's new Vertiv and ON Semiconductor positions for AI power, and the note that chips alone reach roughly $5 trillion of spend by the end of the decade. The Q&A block keeps growing and this week it earned the time: the humanoid robotics SPAC that fails the sniff test, the best AI energy stock between GE Vernova, Vertiv, and Bloom, Aehr Test Systems after a 15x run, AppLovin's make or break Q3, Cipher's messy quarter and the CEO stock sale, Credo against Astera Labs, and how Michael Burry and Jim Cramer keep getting the AI trade wrong. The show is live every Friday. If you're tracking AI infrastructure, memory, or the neo clouds, bring your tickers and questions to the chat and the guys will take them on air. If this one sharpened your read on the memory trade, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and tell us in the comments which constraint you think the dollars rotate to next. Chapters: 00:00 Episode 52 and the Six Five Summit slate 02:38 A memory bull flips bearish 04:05 The memory wall: a 5 to 10 year constraint 07:15 KV cache and memory beyond HBM 09:23 Sandisk investor day: the 2030 targets 11:14 Redesigning the memory hierarchy for inference 13:45 A100s contracted through 2030 15:25 CoreWeave: layers of revenue, real risks 18:48 Power is the gate: Vertiv and ON adds 21:01 Futurum's $12.7 trillion capex number 22:27 Nebius and the Blackwell auction 25:46 Selling puts on space names 28:49 ASTS and Rocket Lab: the execution risk era 31:34 Applied Materials, Cisco, and Lenovo prints 33:56 Elevated expectations, tepid reactions 36:06 Q&A: the humanoid robotics SPAC 38:37 Best AI energy stock: GEV, Vertiv, Bloom 40:42 Aehr Test Systems after a 15x run 42:06 AppLovin's make or break Q3 45:39 Cipher's miss and the CEO sale 48:29 Credo vs Astera Labs 51:22 Burry, Cramer, and the 5% that is overbuilt 55:18 SaaSpocalypse winners 57:43 The eye of the storm before NVIDIA #FuturumEquities #DanielNewman #ShayBoloor #AI #Semis #Earnings 58:20 Memory is a 5 to 10 year constraint. The 2030 guides just proved it. | Ep. 52

Memory is a 5 to 10 year constraint. The 2030 guides just proved it. | Ep. 52

Futurum Equities August 14, 2026 11:09 pm

SpaceX just delivered its first earnings report as a public company, and the number nobody modeled stole the whole show. Daniel Newman and Shay Boloor spend episode 51 on the print, the lockup everyone was selling ahead of, and why a weak jobs report has the entire AI trade breathing again. The guys go deep on SpaceX: 92% growth in an AI cloud business that was never supposed to exist at this scale, a revenue path from $44 billion this year toward $260 billion by 2029, and the math behind Shay buying the lockup fear at $109 to $110 before the print. Elon's on-record line that SpaceX runs only on Nvidia gets its own moment. From there it's Palantir's rerating: 93% growth, commercial up 150%, net dollar retention at 157%, roughly $6 billion of US commercial contract value closed, and 73 deals worth over $10 million each. Daniel calls the end of the SaaS apocalypse and explains why compute is the moat while Anthropic's street ARR pushes $85 billion. AMD gets the honest treatment: the MI350 and EPYC beat, why Lisa Su's conservative guide is the long game, and the Helios rack scale platform with its claimed 30% more tokens per dollar. Plus the no rate hike case under Kevin Warsh after a weak jobs Friday, Shay's new Vertiv position at $220 for AI power exposure, Micron over SK Hynix on HBM rate of change, and the reveal that Shay sold every share of his Tesla on a SpaceX merger theory. The back half is all viewer Q&A: the Palantir and Mercury digital twin partnership, Daniel's put selling playbook that pays him 5 to 10% of his portfolio a year in cash flow, the Oscar Health case study, T1 Energy and EOS Energy as the speculative power trades, and who actually wins when models get cheap. The show is live every Friday. If you're tracking AI infrastructure or trading this earnings season, bring your tickers and questions to the chat and the guys will take them on air. If this one helped you make sense of the tape, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and come tell Shay whether the Tesla exit was genius or tin hat in the comments. Chapters: 00:00 Episode 51: back from the Schwab event 04:19 Jobs Friday: the no rate hike case 09:40 Why weak data is bullish: the 10 year toward 4% 10:57 Buying the July panic and the gaslighting 15:02 Vertiv and the AI power gap 17:22 SpaceX's first earnings as a public company 20:04 The $44B to $260B ramp and the lockup buy 25:56 Sizing SpaceX and the space portfolio 28:12 AMD: Lisa Su's conservative guide and Helios 34:56 Palantir's rerating: 157% retention 38:48 The SaaSpocalypse is over 41:54 Models as commodities and the scarcity premium 44:26 Q&A: Palantir and Mercury's digital twins 46:28 Daniel's put selling playbook 51:37 T1 Energy, EOS, and Shay selling all his Tesla 54:19 Memory: Micron over SK Hynix 57:58 Who wins when models get cheap #FuturumEquities #DanielNewman #ShayBoloor #AI #SpaceX #Earnings 1:1:33 SpaceX guided to $260B by 2029. Shay bought the lockup fear. | Ep. 51

SpaceX guided to $260B by 2029. Shay bought the lockup fear. | Ep. 51

Futurum Equities August 7, 2026 11:26 am

Episode 50 lands in a strange tape. Google grew its cloud business 82% and the stock got sold anyway, and Daniel Newman and Shay Boloor spend the show on what that selling actually means for everyone holding the AI trade. The guys go deep on Alphabet's print: a $25 billion cloud quarter growing 82% with 36% operating margins, a $15 billion capex raise, and the earnings math behind Shay's call that Google is "a great business, bad investment" until EPS clears $20 again in 2029. Nvidia at 18x earnings gets the other side of that trade. From there it's the $11.7 trillion cumulative AI capex bull case through 2030, TSMC raising capex $8 billion in a quarter while flagging a bigger 2027, Intel's blowout on server CPU pricing and Shay's rule that INTC has to prove an external foundry customer to earn its price above $100, Tesla converting record deliveries into a 1% operating margin, Micron and the HBM demand curve the market keeps mispricing, and IREN's $2.8 billion AI cloud contract with roughly half the GPU capex prepaid by customers. Plus SpaceX lockup math and entry levels, Meta at 18x with Shay's $1,200 thesis, and Jensen Huang joining X mid show. The back half is all viewer Q&A: ServiceNow as the benchmark for AI software, Shay's tier one and tier two software framework, whether Meta is the best Mag 7 buy right now, IREN position sizing, and the drone bucket ahead of the Anduril IPO. The show is live every Friday. If you're tracking AI infrastructure or trading this earnings season, bring your tickers and questions to the chat and the guys will take them on air. If this one helped you make sense of the tape, a like genuinely helps the show reach more investors. Subscribe and hit the bell so you catch the next live, and come argue with Shay about NVDA at $100 in the comments. Chapters: 00:00 Episode 50: Jensen joins X and the inverse Cramer tape 01:31 Strong prints keep getting sold 03:15 AI turns political and the Apple tell 07:47 The $11.7T capex bull case 09:42 Bottleneck bros and the rate of change problem 12:56 Semis supply constrained for 4 to 5 years 16:10 Google vs Nvidia: the two year problem 18:56 Daniel's defense: capex is existential 21:37 Memory: the market is pricing the peak wrong 23:20 TSMC raises capex, the grown up signal 26:50 Intel's $100 foundry gate 34:07 Tesla: record deliveries, 1% margin 37:47 SpaceX: merger math, lockups, entry levels 43:31 Q&A: ServiceNow and the software tiers 48:57 Meta at 18x and the $1,200 case 52:46 IREN's $2.8B contract 55:45 Drones, Anduril, ASTS 59:06 The put selling playbook and NVDA $100 on record #FuturumEquities #DanielNewman #ShayBoloor #AI #Semis #Earnings 1:2:40 Google grew cloud 82% and still got sold. Where the money went. | Ep 50

Google grew cloud 82% and still got sold. Where the money went. | Ep 50

Futurum Equities July 24, 2026 11:19 am

INSIGHTS

powered by

Feedback

Rate your experience