[Butler Post] AMD and Intel

AMD’s Cash Flow

AMD’s cash-flow engine is finally purring instead of coughing, its server-CPU and AI-accelerator share keep ratcheting up, and the stock is way cheaper than Nvidia—but still isn’t screaming “dollar-store“ bargain on classic value metrics. If you’ve got a 10-year horizon and can stomach semiconductor mood-swings, AMD is credible compounder-candidate, not a deep-value play.

1. Cash-flow check-up (yes, it’s looking healtier)

Key takeaways

• Double the horsepower: Operating cash flow almost doubled YoY in 2024, and Q4’s 17% OCF margin shows the model scaling.

• Cap-light build-out: CapEx ran ~$636M in 2024 (just 2.5% of revenue), leaving plenty of runway before AMD needs to tap debt markets again.

• Cash cushion: $3.8 B in cash vs. net debt near zero after repaying a $750 M note. Balance-sheet risk is now “small-yappy-dog“ rather than “angry-pit-bull.“

Verdict: Healthy. Not Nvidia-level cash-gushing yet, but AMD is no longer living paycheck-to-paycheck.

---

2. Market-share momentum (the flywheel that feeds the cash)

• Server CPUs (EPYC): Shares hit 27.2% in Q12025—another record, eating into Intel every quarter.

• AI GPUSs (Instinct MI300/350): Street models now pencil in $15B+ annual accelerator revenue by 2026, thanks to hyperscaler design-wins and even a cheeky 67% price hike on the MI350.

• Client CPUs: Ryzen 7000/8000 (“Zen5”) finally pushed notebook ASPs above Intel’s H1 2025, flipping a decade-old script.

Translation: Growing pie slice = optionality. Every 1 ppt of extra server share roughly adds $1B-$1.2B annual revenue at current pricing.

---

3. Valuation sanity check

Is that “undervalued“?

• Versus Nvidia: Yes, AMD is the discount rack.

• Versus the S&P 500 or legacy chipmakers: No, it’s still growth-stock territory.

• Free-CF needs to tripple before the sticker looks like a classic “value stock.”

---

4. 10-year scenario sketch (back-of-napkin, coffee-stained)

---

5. Red‑flag risk round‑up (a.k.a. why your future self might face‑palm)

1. Supply‑chain choke points (CoWoS capacity, HBM3E, export controls).

2. Gross‑margin whiplash if hyperscalers start demanding Costco pricing for AI GPUs.

3. Intel rebound + ARM wildcards could cap server upside.

4. Valuation compression once the AI sugar‑rush normalizes.

---

Bottom line—should you buy now?

• Cash flow: Solid and trending up—passes the sniff test.

• Undervaluation: Relative to Nvidia, sure. Absolute bargain? Nope.

• 10‑year lens: If you believe AMD can keep compounding share in both servers and AI accelerators, today’s premium could look cheap in hindsight.

• Risk budget: Treat this as a growth stock in a cyclical sector—position‑size accordingly and expect volatility spikes big enough to rattle your coffee mug.

My non‑advisory gut‑check: Start nibbling on red days, build a position over time, and keep Dramamine handy for the inevitable semiconductor roller‑coaster ride.

(Not financial advice. Do your own due diligence, double‑check the numbers, and blame only the cat if you lose money.)

---

Intel’s Cash Flow

Cash‑flow face‑off: AMD vs Intel (2024 → mid‑2025)

What the numbers shout

1. AMD is throwing off real, discretionary cash.
   After the Xilinx digested, cap‑ex settled near $200 M/quarter, so >70 % of OCF drops through to FCF. The first quarter of 2025 kept that streak. Cash pile hit $6 B with < $4 B gross debt, leaving optionality for more AI‑GPU tape‑outs or bolt‑on deals.

2. Intel is still in cap‑ex purgatory.
   The core business generated >$10 B OCF in the last twelve months, but every nickel—and then some—gets sucked into fabs. Free‑cash‑flow stays negative while 18A and Ohio fabs soak up >$18 B gross cap‑ex in 2025. Partner contributions and government incentives mask the burn but don’t reverse it.

3. Balance‑sheet risk diverges.
   AMD can self‑fund aggressive R&D and still repurchase stock without borrowing. Intel is issuing debt, selling Mobileye shares, and cutting 24 k heads to keep liquidity intact.

4. Return math

   • AMD: Trailing FCF ≈ $2.7 B. At a $283 B cap, FCF yield ≈ 1 %. You’re paying for growth—but growth is visible in EPYC + MI350 share gains.

   • Intel: Trailing FCF negative; value hinges on future foundry profits plus a turnaround in CPUs/AI accelerators. If it works, multiple expands and FCF flips positive—huge upside, but only after years of heavy lifting.

10‑year bet — my chips go to…

💰 AMD. It already generates positive, scal­able free cash, has a clean balance sheet, and keeps nibbling share in both servers and AI GPUs. If Lisa Su keeps compounding FCF even at 15 % CAGR, today’s rich multiple looks reasonable.

🤞 Intel = lottery ticket. A foundry comeback could quintuple earnings, but you must finance years of red ink and trust a culture mid‑overhaul. Great if you need asymmetry and can wait, but risk‑adjusted it loses to AMD for most portfolios.

My (non‑advisor) move: 70 % of the silicon sleeve to AMD, 30 % to Intel only if you crave a speculative turnaround kicker. Otherwise, park the whole bet with AMD and sleep better.

—Butler ☕️