muji @ hhhypergrowth

Today's free post is on the Layers of AI - a tech explainer into what is underlying this wave of Generative AI and LLMs. #AI #GenerativeAI #LLMs hhhypergrowth.com/layers-of-ai/

@PiyDW what a dis, calling out the stock i 10x'd on in a year and half

@hhhypergrowth Why should I believe a guy who was pitching cloudflare $net at 100x sales. 🤣🤣

at what point does GitLab get swallowed up by a hyperscsaler that wants to compete with Microsoft Github? $GTLB asking for a friend

@JonahLupton @BitcoinAIGuy the whole point of Coreweave's backlog is that it has the power runway lined up $CRWV it secures the powershell at the point of signing the cust

@BitcoinAIGuy $CRWV has a large backlog but how valuable is a backlog if they don’t have the power to monetize it? Power is becoming more valuable and every day since there appears to be a finite amount available for data center growth over the next 2-3 years.

$CRWV has 3.1 GW+ contracted power and is guiding $30b+ ARR in 2027 $IREN has 4.5 GW, AND it’s trading 70% lower than CRWV today, but doesn’t have to pay colocation fees. Is it really that simple? If so, $IREN can 7-15x within 3 years

@CloudflareDev who is Greg Kinear calling on the phone tho

10¢, 10s, 10s across the board and a $0, because it's free for the first 24 hours!

@mayurzzz @aakashgupta probably all that free time he has apart from being a substacker, podcaster, and having a toddler

@aakashgupta How do you know so much a out everything. 😂

Sundar just shipped the only image model that knows what the world looks like right now. Everyone else is rendering from memory. Every other image model on the market, Midjourney, DALL-E, Flux, Stable Diffusion, generates from frozen training data. They’re rendering what the world looked like when the model was last trained. Nano Banana 2 generates from live web search results and real-time weather data. That’s a fundamentally different architecture. The “Window Seat” demo is the tell. It pulls Brighton Beach’s actual weather conditions, renders the West Pier from reference images found via Google Search, and overlays live temperature and wind data. No other image model can do this because no other image model is plugged into the world’s largest search index. Think about what this unlocks. Marketing teams generating product images with real store locations in the background. Travel companies rendering accurate destination photos with current season conditions. News organizations generating contextual imagery grounded in real events. Every use case where “the image needs to reflect reality” now defaults to Google. The competitive moat here is Google Search itself. Midjourney would need to build or license a web-scale search engine to replicate this. OpenAI would need Bing to be as comprehensive as Google, and it isn’t. The 750 million monthly active Gemini users are generating training signal on which grounded images perform best, creating a flywheel that compounds the advantage. Google shipped three Nano Banana models in six months. August, November, February. That cadence matters because each version absorbs user feedback at a scale nobody else operates at. Nano Banana Pro was so popular they had to throttle access due to demand. Image accuracy at scale requires search infrastructure nobody else has built. And that gap is only getting wider.

@OptionWell nimble... or, just an admission they couldn't figure it out on their own ("Hey Siri!, why have you sucked for years now?")

I love that $AAPL isn't spending fortunes on an AI buildout, instead leveraging the work of others. Although this approach has risk, it provides flexibility to partner with the best over time (from an OpenAI deal in 2024, to now Gemini); also, financial freedom, avoiding getting caught on endless refresh cycles, while avoiding many risks.

@StockMarketNerd i love you nerd

"Wait $NVDA had a great quarter why is it down a few %."

Agree ... time to find some deals out there.

@StockMarketNerd little bit

$NVDA -- kinda good:

today in premium: Understanding Reddit $RDDT part 2 tomorrow covers the financials, AI positioning, and moves

@antibearthesis @amitisinvesting because everyone and their mom own the stock already

@amitisinvesting HOW ARE WE NOT UP 10% 🤯🤯

UMMMMMMMMM the street wanted $73B on the $NVDA Q1 guide they just said “yeah we’ll probably do $79.8B” the only thing that mattered was the guide for this earnings LETS. GO.

Axon $AXON reported strong Q4 and FY26 guide FY25 Revenue $2.8B +33% ARR $1.3B +35% Bookings $7.4B +46% Future contracts $14.4B +43% Net inc margin 4.5% aEBITDA 25.5% guide 2026 revenue +27-30%, aEBITDA 25.5% new 2028 target of $6B ARR, aEBITDA 28%

@WisemanCap Considering they were growing 2% a year ago, this revenue rebound is definitely a good thing

@hhhypergrowth The only point market cares is the revenue beat and raise which was surprisingly good

Fastly $FSLY continues to disappoint on completely ignoring its land motion. They added 31 net new customers over the last year. Fun fact, Fastly added more large customers than it did overall customers in a year (+32 vs +31). At least they are no longer shedding customers like in FY24. But you won't know, because they aren't going to report it anymore.

@KurtMilneStocks Its been getting close on net… but is not a focus. Op margin just hovering

@hhhypergrowth Do you think they will ever be profitable? Looks like they almost make profit year after year.

today in premium: Cloudflare's stellar Q4, its GTM revamp and rise of agentic AI, and a look at Fastly's Q4 too $NET $FSLY

@StockMarketNerd @invest_alph margins impacts aren't a worry, look at how margins impacted by free users shifting to paid under GAAP (S&M to COGS) . this is an accounting shuffling.

@hhhypergrowth @invest_alph Everything amazing except the 2026 profit guide and I would put a lot of $ on their 2026 EBIT ending up above pre-call consensus. & when they’re accelerating this meaningfully and this well positioned in AI… spending makes sense!

$NET earnings snapshot:

@matt_solomon10 excellent take (as someone that follows them heavily)

$NET (continuation — resharing my prior threads above) The final component.

@matt_solomon10 narrator: there's not

Transcript is still not out so I'm waiting. But my gut is there is some juicy insight here.

Gotta dig into Fastly’s ($FSLY) earnings — it’s a clean case study for how to think about the agent-driven internet.