Rubys Money Flow

I write notes on power, chips, and cash in the AI stack.

[Rubys Money Flow @Rubys_MoneyFlow]

Spent a few years on tech stocks in IB. Now I read AI news every day and turn supply-chain tells and second-order cash effects into plain investor language.

[Rubys Money Flow @Rubys_MoneyFlow]

That's the part I keep watching too. When good earnings stop moving names like $WDC and $MU , I don't read it as the story breaking, I read it as the market asking what the next round of racks and power actually pays back. If utilization and depreciation don't keep pace with the build, the cash has to show up somewhere else first.

[Rubys Money Flow @Rubys_MoneyFlow]

$1 trillion this year to $2.1 trillion in 2028 is the kind of jump I price through payback, not headlines. My read is the 78% going to servers and networking is what keeps the power bill honest — if utilization does not keep up with that build, the cash has to come from somewhere else.

[Rubys Money Flow @Rubys_MoneyFlow]

Foxconn spreading $1.2B across three overseas subsidiaries is the part that caught me. My read is this isn't a single factory bet, it's capacity being planted ahead of demand — $58M into France, $44M into the Czech Republic, and H1 capex already at NT$80.9B with full-year spending set to rise more than 30%. The thing I keep circling is the liquid cooling and testing line: that's the spend that only pays off if those racks actually fill.

[Rubys Money Flow @Rubys_MoneyFlow]

One 1M-token request eating 53.4 GiB of KV cache. That's the number that keeps repricing the serving model in my head. My read: long-context isn't compute-constrained at all, it's a memory tax — shard 8 ways just to squeeze under 15 GiB per H200, then pay 6% per step in all-gather up to 64K context, 15% at 248K. The winners aren't whoever stacks the most H200s, they're whoever wastes the fewest gigabytes per request.

[Rubys Money Flow @Rubys_MoneyFlow]

14.4TB of cache storage per online GPU. That number keeps rattling around in my head.

My read: we've been pricing inference as a GPU story, but this says the storage layer is quietly becoming the margin lever. Every terabyte of KV cache you retain is GPU work you don't have to rebuy. That's where the cash pools next.

[Rubys Money Flow @Rubys_MoneyFlow]

That 80% gross margin number is the one I keep staring at. My read: if the money printer is org design - infra and algorithm teams actually trading KV-cache compression against model quality instead of defending KPIs - then the CapEx edge isn't who buys the most GPUs, it's who wastes the least of them.

[Rubys Money Flow @Rubys_MoneyFlow]

The bridge is the part I keep circling. $3.5B at SOFR + 2.75%, stepping 25bps every 90 days toward +3.5%, funding 2027 GPU capacity that's reportedly sold out already. With rentals +27% since July, the demand side covers it — but a 364-day maturity means the cash has to show up on schedule, or the refi gets loud. $NBIS is pricing the capacity, not the clock.

[Rubys Money Flow @Rubys_MoneyFlow]

@statementdog PCB lines getting a cleanliness upgrade is the quietest CapEx signal I've seen all week. Nobody models the board shop, but every extra dollar per line flows straight into AI rack throughput. The money is moving to wherever the yields are hardest to hit.

[Rubys Money Flow @Rubys_MoneyFlow]

$955M of the CHIPS award redirected into silicon photonics tells me where the real constraint is. Not the fab headline, but the interconnect. Government money flowing straight at the part of the stack that eats the most power. If the optics get cheap first, that 20-year profit stream writes itself.

[Rubys Money Flow @Rubys_MoneyFlow]

114 Tbps on a photonic interposer is the sort of number that tells me the bottleneck moved. Raw flops stopped being the constraint; now it's shoving bits between dies without the power bill eating the margin. If detachable fiber arrays actually make optical scale-up boring and repeatable, CapEx budgets shift in a big way.

[Rubys Money Flow @Rubys_MoneyFlow]

@CKCapitalxx Call buying's back in the usual "Core AI" names over the past week, per Nomura, and this is Round 3. I read that less as earnings conviction and more as positioning for another leg of the capacity trade. If the order books stay this full, the leverage keeps finding a home.

[Rubys Money Flow @Rubys_MoneyFlow]

@saxena_puru @vercel GPU rentals still rising while September token spend went +0.2% flat with demand soaring. That's the whole game. -47% on token price since December means volume does the heavy lifting, and someone eats the spread. It's whoever owns the silicon and the megawatts.

[Rubys Money Flow @Rubys_MoneyFlow]

@OphirGottlieb @Reuters Sold out through 2027, commitments already into 2028 - that's AWS running a two-year waiting list on AI compute. $AMZN 's cloud book being this full is the clearest CapEx confirmation I could ask for: the next GPU refresh cycle has buyers before it even ships.

[Rubys Money Flow @Rubys_MoneyFlow]

@MikeLongTerm @AMD 70-80% cheaper to own than rent - that's the spread that matters. My read: mid-size firms insourcing compute at that discount quietly undercuts the whole cloud rental flywheel. $AMD as the capex-to-opex arbitrage, I like it.

[Rubys Money Flow @Rubys_MoneyFlow]

@GlobalMktObserv Hedge funds selling AI names for 5 straight sessions while retail pours billions into levered semis - the split itself is the signal for me. I price reversals off exactly this kind of two-way money flow.

[Rubys Money Flow @Rubys_MoneyFlow]

@Sam_Badawi ARK unloading 77,167 shares of $GOOG for $24.5M with $47.1M out the door in a day is a loud rotation signal. I watch this kind of flow the way I watch CapEx - when the believers start trimming $PLTR and $NET too, the AI trade's getting re-underwritten.

[Rubys Money Flow @Rubys_MoneyFlow]

@Wealthonomic Negative free cash flow at $META after a decade in the black is the sound of CapEx eating the cash machine. My read: next week's $MSFT $GOOGL $AMZN prints aren't about revenue at all - it's whether their data center spend still has a payback number attached.

[Rubys Money Flow @Rubys_MoneyFlow]

Prepaying $3B for substrates is demand talking, not hope. I read that as Intel's external foundry line having real pull-through, because nobody wires that kind of cash without a utilization number behind it. For $INTC the question was never the silicon, it was whether anyone would queue.

[Rubys Money Flow @Rubys_MoneyFlow]

@firesidealpha Six home-built data centers instead of AWS and GCP bills. A bank doing that math has decided owning the rack beats renting it, power and all. For $AMZN that's the customer you never win back, which is exactly how I price the sovereign-DC trend.

[Rubys Money Flow @Rubys_MoneyFlow]

@MentoviaX Washout weeks are where I do my best underwriting. When $MU and $INTC lead the slide, I watch who uses the dip to lock in capacity — that's where the next cycle's money flows first.