The Cashflow Memo

AI Is Finally Paying. The Question This Week Was How The Suppliers Got Paid.

Micron’s 2026 memory is sold out, with customers writing prepayment checks for chips that don’t exist. AMD’s Q1 data center revenue was up 57% — but the strength was EPYC CPUs, not the OpenAI GPU deal, which contributed zero dollars to the quarter. Palantir printed an 85% revenue quarter at a Rule of 40 score of 145. And Disney’s first full quarter under D’Amaro printed streaming income up 88%.

The Telltales Weekend Update. Ava Cabot and analyst Marcus Graham walk through what happened this week — and what’s coming next — across the 86 companies in the Cash Flow Memo. About 13 minutes. No filler.

Download the memo at telltales.us. Hunt, Jason, and Mike are back Wednesday on episode 2620.

Chapter markers

* Time | Segment

* 0:00 | Opening disclaimer

* 0:15 | Cold open — throughline + E2619 callback + Mother’s Day

* 1:30 | Theme — How the suppliers got paid (MU vs AMD compare/contrast)

* 5:30 | Deep dive — Palantir

* 9:30 | Rapid-fire — CRCL / DIS / UBER / ABNB + forward-week earnings

* 13:00 | Close — Consensus Watch + Wednesday tease

* 13:30 | Closing disclaimer

Full transcript

Opening disclaimer

Ava: The following conversation is intended for informational purposes only. You should always do your own work to determine if an investment is suitable for you.

Cold open

Ava: You’re listening to the Telltales Weekend Update. I’m Ava Cabot.

Marcus: And I’m Marcus Graham — the cashflow desk.

Ava: Quick note: the show is produced entirely with AI tools, and both voices you’re hearing are AI-generated. Send feedback through the Substack.

Ava: And before we get into it, happy Mother’s Day to all the moms in the audience. Thank you for spending part of today with us.

Ava: Here’s the throughline for this week. AI is finally paying — but the unit economics of how suppliers capture that demand vary wildly. On Wednesday’s show, Hunt, Jason, and Mike walked the hyperscaler stack from the bottom up — Hunt made the case that Amazon’s networking edge may matter more than Nvidia’s chips[^ep-e2619]. That was the architecture conversation. Today is the unit-economics conversation, supplier side. Micron printed the cleanest version. Customers writing prepayment checks for memory that doesn’t exist yet[^news-mu-hbm-booked-20260507] — pricing power so strong it’s hitting cash this quarter. AMD printed the messier version. Data center revenue up 57%[^news-amd-datacenter-20260505], but the strength was in EPYC server CPUs and existing Instinct GPU shipments[^news-amd-10q-q1-20260505], not the OpenAI deal everyone is anchoring on. That deal — signed last October[^news-amd-openai-20251006], OpenAI got warrants for up to 10% of AMD[^news-amd-openai-warrant-20251006] — contributed zero dollars to Q1[^news-amd-10q-q1-20260505]. Shipments don’t start until the second half of this year[^news-amd-openai-20251006]. The CPU side of AMD’s print connects back to Intel two weeks ago — same demand signal, different name on the box[^news-intc-q1-20260423][^news-intc-cpu-ai-20260423]. That’s the show.

Theme — How the suppliers got paid

Ava: Two supplier prints this week, both validating AI demand, very different evidence about pricing power. Micron is charging customers in advance. AMD’s strength came from CPUs, not GPUs — same story Intel told the market two weeks ago[^news-intc-q1-20260423].

Ava: Page 8 — Micron. Micron’s 2026 high-bandwidth memory is fully booked, with customers signing multi-year prepayment agreements for supply that doesn’t exist yet[^news-mu-hbm-booked-20260507]. CEO Sanjay Mehrotra said Micron can only meet between 50 and 67% of demand from key customers[^news-mu-demand-constrained-20260507]. The four largest spenders on AI — Meta, Microsoft, Amazon, and Apple — all publicly cited memory cost and availability constraints. Stock hit an all-time high of $683 on May 7[^news-mu-ath-20260507]. Mizuho took the price target from $545 to $740[^news-mu-mizuho-20260509]. And Micron raised fiscal 2026 capex to $25B from $20B, a 25% step up[^news-mu-capex-raise-20260507].

Ava: And AMD. After the close on May 5 — Q1 revenue $10.3B, data center revenue up 57% to $5.8B[^news-amd-q1-20260505][^news-amd-datacenter-20260505], Q2 guidance $11.2B[^news-amd-q2-guidance-20260505], stock up 16% the next day[^news-amd-stock-20260505]. Important read on that data center number. The growth came from EPYC server CPUs and the existing Instinct GPU ramp[^news-amd-10q-q1-20260505], not from the OpenAI deal. Per the 10-Q, the OpenAI warrants haven’t vested[^news-amd-10q-q1-20260505]. Shipments don’t start until the second half of this year[^news-amd-openai-20251006]. The deal contributed zero dollars to Q1[^news-amd-10q-q1-20260505]. Which connects directly to Intel two weeks ago. Intel printed Q1 revenue up 7% YoY to $13.6B[^news-intc-q1-20260423], with management telling the Street that the CPU-to-GPU ratio in AI deployments is shifting from 1:8 toward 1:4, and trending toward parity as multi-agent workloads scale[^news-intc-cpu-ai-20260423]. That’s the read across both prints. AI demand isn’t just GPUs. Hyperscalers are buying head-node CPUs alongside their accelerator clusters. CPU demand is having a renaissance after two flat years, and AMD just confirmed what Intel told the market. Marcus, the cashflow take.

Marcus: The Q1 print is good news about the business AMD already has — EPYC share, current Instinct shipments — not about the business AMD bought with equity last October. The OpenAI warrant is a future customer-acquisition cost that doesn’t hit the P&L until the second half of this year[^news-amd-openai-20251006]. When it does, it shows up in two places. As data center revenue ramping into 2027. And as up to 160 million shares of dilution if MI450 hits its deployment milestones[^news-amd-openai-warrant-20251006]. The memo had AMD at 167× trailing free cash flow on $4.4B of TTM FCF[^memo-amd-evfcf-20260328][^memo-amd-fcf-20260328] — that multiple is being defended by the forward book, not by what just printed. Q1 didn’t move the actual question. The actual question is whether the OpenAI relationship — paid for in October at a penny a share — produces enough revenue to justify the dilution. We don’t get a read on that until late 2026 at the earliest.

Marcus: The compare to Micron sharpens the point. Both names trade at multiples that need a story. Micron at 209× free cash flow has the leverage of a structural shortage[^memo-mu-evfcf-20260226][^memo-mu-fcf-20260226] — they’re charging in advance because they can, and the cash is hitting the bank this quarter. AMD at 167× paid customers in equity last October to lock in revenue that hasn’t started yet, while the current quarter prints fine on CPU demand and legacy GPU shipments. Both stocks are pricing forward stories. Only one of those stories has cash hitting the bank right now.

Ava: Two suppliers, two prints. Same end-market. Pick your fighter.

Deep dive — Palantir

Ava: Page 17 — Palantir. The cleanest AI is actually paying print of the cycle so far. After the close on May 4. Q1 revenue $1.63B, up 85% year-over-year — the fastest growth rate since the company went public in 2020[^news-pltr-q1-20260504]. Beat consensus by $90M. U.S. revenue alone hit $1.28B, up 104% — the first time U.S. growth has crossed 100%[^news-pltr-us-growth-20260504]. U.S. commercial revenue up 133%. Net income quadrupled to $870M, EPS $0.34[^news-pltr-netincome-20260504]. Adjusted free cash flow $925M on a 57% margin[^news-pltr-fcf-20260504]. Rule of 40 score: 145[^news-pltr-rule40-20260504]. Quick aside for readers who don’t follow software — Rule of 40 is the standard efficiency benchmark in SaaS. You add a company’s revenue growth and its profit margin. 40 is the bar for a healthy software company. Palantir is at 145. Roughly four times the bar. Back to the print — full-year guidance raised from $7.2B to $7.65B[^news-pltr-fy-guidance-20260504]. Marcus, the cashflow take.

Marcus: Palantir is the proof point that the rest of enterprise software has been promising and not delivering. The memo has Palantir at 160× trailing free cash flow on $1.9B of TTM FCF[^memo-pltr-evfcf-20260331][^memo-pltr-fcf-20260331]. We re-anchor when the Q1 10-Q files. The number that actually matters from this print is the U.S. commercial line — that’s a customer set that didn’t exist 18 months ago paying real money for AI deployment, not pilots.

Ava: 160× free cash flow. Rule of 40 at 145.

Marcus: 160× is not defensible on a static comp. It’s defensible only if you believe the U.S. commercial growth rate doesn’t normalize for another four to six quarters. So weight the outcomes. Bull case at 30% — government and commercial both compound from here. Base case at 50% — commercial decelerates and the multiple compresses by half on a stock that still works. Bear case at 20% — the multiple does what 160× multiples always do.

Ava: So the question is whether the customer set that just appeared keeps appearing.

Marcus: That’s the entire question. Argus upgraded to Buy on May 7, Citi took the target to $225[^news-pltr-argus-20260507][^news-pltr-citi-20260507]. The Street is pricing the bull. Watch the Q2 print for whether U.S. commercial holds north of 100% year-over-year, or steps down to 70. 70 is still a great number. It is also a different multiple.

Ava: 145 Rule of 40. That number alo...