Bhagat Capital

12/08/2026

COREWEAVE Q2 2026: STRONGER DEMAND—BUT ALSO A MUCH BIGGER CAPITAL BILL

CoreWeave shares jumped around 18% following the results. The market clearly liked the record revenue, expanding backlog and raised guidance. However, investors should look beyond the headline growth.

Revenue reached $2.575B, while backlog expanded to $104.2B. Importantly, that backlog excludes more than $25B of additional customer commitments signed during early Q3.

The most interesting management commentary was about GPU asset life. CoreWeave recently contracted an NVIDIA A100 architecture—originally introduced in 2020—through 2029 at what management described as an attractive price.

That challenges the argument that older GPUs quickly become commercially obsolete. They may be recontracted or redeployed into inference workloads after their original contracts expire. However, one A100 contract is not sufficient evidence for the entire fleet. Investors should monitor renewal pricing, utilization and contract duration.

CoreWeave is also becoming more than a pure GPU-rental company:

• Storage, CPU, networking and software exceeded $400M in ARR
• Managed inference booked ARR surpassed $100M
• The company expects at least $250M of managed-inference ARR by year-end
• Pricing increased approximately 25% across SKUs in July
• CoreWeave completed the first cloud bring-up and validation of NVIDIA Vera Rubin NVL72

The risk remains capital intensity. CoreWeave reported a $626M GAAP net loss and a $567M adjusted net loss. Q2 interest expense reached $640M, while 2026 CapEx guidance was raised to $35B–$39B.

Financing is improving: the weighted-average cost of debt has declined by almost 300 basis points, representing approximately $1.1B in annualized interest savings based on the Q2 debt load. But Q3 interest expense is still expected to rise to $860M–$940M as deployments accelerate.

My investor takeaway: demand visibility is exceptional, but the investment case ultimately depends on whether utilization, renewal pricing and higher-margin services can generate free cash flow faster than debt, depreciation and capital requirements increase.

Would you classify CoreWeave as durable AI infrastructure—or a debt-funded race against GPU obsolescence?



This post is for educational purposes and does not constitute investment advice.

12/08/2026

A GPU launched in 2020 just secured a CoreWeave contract extending through 2029. 👀

That challenges the belief that older AI chips quickly become obsolete—and suggests GPUs may generate incremental returns after their original contracts expire.

But one A100 deal does not prove the economics of the entire fleet. Investors still need to monitor:

• Renewal pricing
• GPU utilization
• Contract duration
• CapEx and interest costs

CoreWeave surged 18% after the results—but with a $626 million net loss, the real question is whether asset durability and pricing power can stay ahead of debt and depreciation.

Durable AI infrastructure—or a capital-intensive race?

11/08/2026

AI memory stocks exploded in 2026 — but the real question now isn’t demand. It’s margins.
Micron is operating around mid-80% gross margins, while SK hynix recently crossed 76% operating margin.
If 70–80% profitability becomes the new normal, these businesses may deserve a completely different valuation framework.
But if this is simply peak-cycle profitability, today’s seemingly cheap valuations could be deceptive.
So have AI memory margins already peaked — or is this a structurally different memory cycle?
Full analysis on Bhagat Capital.

09/08/2026

Everyone is focused on GPUs and HBM in the AI story.
But there’s another bottleneck investors should not ignore: how fast the data moves.

Inside AI infrastructure, copper still plays a major role today — especially in scale-up systems. But as bandwidth demand rises, the industry is gradually shifting toward optical connectivity and CPO (Co-Packaged Optics).

That’s where the next layer of opportunity may emerge.

The real question is no longer just who makes the fastest chip…
it’s also who helps move the data fast enough.

Companies like Coherent, Lumentum, Credo and Broadcom are becoming important names to watch in this transition.

Follow Bhagat Capital for Part 2: Who actually wins this optical transition?

07/08/2026

AMD’s AI revenue is exploding — but there’s a margin catch investors shouldn’t ignore.

Data Center AI is growing fastest, yet its gross margin is still slightly below AMD’s corporate average. That means massive revenue growth does not automatically translate into equivalent profit growth.

In this reel, I break down the 4 risks I’m watching:
Margin mix • Helios ex*****on • Deployment timing • Valuation expectations

The real question for AMD is no longer just: Can AI revenue grow?
It’s: Can that growth convert into margins, cash flow and shareholder returns?

Full AMD earnings + valuation analysis on Bhagat Capital.

06/08/2026

A Wall Street analyst estimated AMD could generate about $30B of AI revenue in 2027.

Lisa Su essentially said: you are underestimating us.

Yet AMD shares still fell after earnings.

My view: the market is no longer questioning demand—it is questioning whether that growth can convert into margins and cash flow.
Is AMD priced too high, or was the sell-off an opportunity?
Full Analysis on my Youtube

AMD reported record Q2 2026 results, guided above Wall Street expec...

05/08/2026

AMD’s AI story just got more interesting.
A Wall Street analyst estimated roughly $30 billion in AMD AI chip revenue for 2027.
Lisa Su’s response?
That estimate may be too conservative.
Yet AMD shares still fell after earnings.
That is the real debate: the demand opportunity is massive, but can AMD convert that growth into enough margin, cash flow and earnings to justify the valuation?
Full analysis on Bhagat Capital.

04/08/2026

ARM just reported another strong quarter. 📈
Revenue, royalty growth, margins, guidance—what actually mattered, and what did Wall Street miss?
In 60 seconds, here's the complete earnings breakdown with no hype and no jargon.
Would you buy ARM at today's valuation?

03/08/2026

Nvidia gets the headlines. Arm gets paid. 💰

Buy Nvidia or build your own chip — Vera, Graviton, Axion, Cobalt all run back to Arm.

Beat earnings AND guidance… stock still dropped 7%, then reversed. That contradiction IS the story.

Tollbooth of AI, or an expensive chipmaker in disguise? Full analysis on YouTube 🎥

Which Arm are you buying? 👇

01/08/2026

🚀 Amazon just revealed its biggest AI ambition yet.

Andy Jassy believes AWS has the potential to become a $1 TRILLION annual revenue business. 🤯

But that ambition comes with a price.

Amazon also raised its 2026 CapEx outlook from $200B to $220B—not because demand is slowing, but because AI infrastructure is becoming more expensive. Higher spending on memory, storage, and networking is pushing costs up as Amazon races to build the next generation of AI infrastructure.

The real question for investors:
Will this massive AI investment create enormous long-term value, or is Wall Street right to worry about the spending?

💬 What's your view—bullish or bearish on Amazon?