ADVANCED MICRO DEVICES, INC. (AMD) 2024 Earnings Analysis
ADVANCED MICRO DEVICES, INC.2024 Earnings Analysis
67/100
AMD's FY2024 shows a company in transition — revenue rebounded to $25.8B with gross margin expanding to 49.4%, but the $24.8B goodwill overhang from Xilinx (35.9% of assets) and a 2.9% ROE reveal that Lisa Su's ambitious AI push via MI300X has yet to translate into returns commensurate with the capital deployed.
Core Dimension Scores
Evaluating competitive strength across earnings quality, moat strength, and risk sustainability
Overall Score Trend
Earnings Quality
Gross margin at 49.4% is respectable for a fabless semiconductor firm and marks a clear upward trend from 44.9% (FY2022) and 46.1% (FY2023). However, it remains far below NVIDIA's 72.7%, reflecting weaker pricing power in both GPU and CPU segments. The improving trajectory is encouraging but AMD still lacks premium pricing ability in its fastest-growing AI accelerator business.
Operating cash flow of $3.0B against $1.6B net income yields a 1.85x ratio — well above 1.0x. This large gap reflects heavy non-cash charges (primarily amortization of Xilinx intangibles ~$3.3B/year) depressing reported net income. The cash is real even if GAAP earnings look anemic.
Operating income of $1.9B on $25.8B revenue yields a 7.4% operating margin — notably thin for a high-growth semiconductor company. The 35.8% expense ratio (R&D + SG&A) is the main culprit, with R&D spending heavy as AMD races to build competitive AI software stacks and next-gen chip architectures.
Total operating expenses at 35.8% of revenue is very high. R&D alone likely exceeds $5.8B as AMD must invest aggressively across CPUs (Zen 5), GPUs (CDNA/RDNA), FPGAs (Xilinx), and the ROCm software ecosystem. Unlike NVIDIA whose revenue doubled to spread costs, AMD's revenue growth has been insufficient to dilute its cost structure.
Free cash flow of $2.4B covers net income 1.47x. The fabless model keeps capex low at $0.6B (2.5% of revenue), allowing most operating cash flow to pass through as FCF. This is a structural advantage — TSMC bears the multi-billion dollar fab investment burden.
AMD's earnings quality scores 68/100 — a mixed picture. The positive: gross margin is expanding (44.9% → 49.4% over three years), cash flow strongly backs earnings at 1.85x CF/NI, and the fabless model delivers clean FCF conversion. The negative: a 35.8% expense ratio eats into profitability, and operating margin at 7.4% is thin for a company with $25.8B in revenue. The Xilinx intangible amortization distorts GAAP profits — cash earnings are meaningfully higher than reported net income.
Moat Strength
ROE at 2.9% is strikingly low and the single biggest red flag. The $49B Xilinx acquisition inflated equity to $57.6B, creating a massive denominator that drowns out earnings. Adjusted for Xilinx goodwill/intangibles, 'operating ROE' would be higher — but the market paid real money for those assets, and returns must justify the price.
AMD's Zen architecture has been the CPU story of the decade. Server market share climbed from ~1% (2017) to ~25%+ (2024), taking share from Intel with superior performance-per-watt. EPYC processors now power workloads at every major cloud provider. This is a genuine technology moat built through superior execution.
MI300X is technically competitive with NVIDIA H100 on raw compute, but the ROCm software ecosystem remains AMD's Achilles' heel. CUDA's decade-long head start means most AI workloads are optimized for NVIDIA. AMD is winning some hyperscaler design wins but lacks the broad developer ecosystem that creates sticky moats.
Three consecutive years of margin expansion signal improving competitive positioning and product mix shift toward higher-margin data center products. However, 49.4% remains mid-tier for fabless semis — Broadcom, Qualcomm, and Marvell all achieve comparable or higher margins.
AMD's moat scores 65/100 — a company with genuine competitive advantages in x86 CPUs but still proving itself in AI accelerators. The Zen architecture renaissance is a real moat story: AMD went from near-bankruptcy to taking 25%+ server share from Intel. However, the 2.9% ROE cannot be ignored — the $49B Xilinx acquisition has yet to earn its cost of capital. In AI GPUs, AMD has the silicon but lacks the software ecosystem (ROCm vs CUDA) that creates lasting competitive advantages.
Capital Allocation
Capital intensity at 2.5% is very low, typical of a fabless semiconductor model. AMD spends only $0.6B on capex against $25.8B revenue. TSMC handles the capital-intensive wafer fabrication. This structural advantage means most incremental revenue flows to the bottom line.
FCF of $2.4B is solid but modest relative to AMD's $25.8B revenue scale and ~$200B+ market cap. FCF yield is below 1.5%, reflecting a company that is investing heavily in growth (R&D) rather than harvesting profits. For a high-growth fabless company, this is acceptable but not yet impressive.
Cash of $3.8B covers $1.7B long-term debt 2.2x — a very comfortable position. AMD has deleveraged significantly since the Xilinx acquisition. The conservative balance sheet provides flexibility for future M&A or accelerated R&D investment without distress risk.
The $49B Xilinx acquisition (2022) is the defining capital allocation decision. It created $24.8B in goodwill (35.9% of assets) and the jury is still out. FPGA/adaptive computing revenue has not yet justified the premium paid. The strategic logic — combining CPUs, GPUs, and FPGAs for AI edge workloads — is sound, but execution is years from proving out.
Capital allocation scores 78/100. The fabless model delivers clean capital efficiency: 2.5% capex ratio and $2.4B FCF on minimal maintenance investment. Cash/debt at 2.2x shows prudent balance sheet management post-Xilinx. The elephant in the room is the Xilinx acquisition itself — $49B for a business that has not yet generated returns commensurate with the capital deployed. If FPGA/adaptive compute becomes critical for AI edge inference, the deal will look visionary; if not, it will remain a drag on ROE for years.
Key Risks
CRITICAL: Goodwill at $24.8B represents 35.9% of total assets — an extremely high concentration from a single acquisition. If Xilinx underperforms or the FPGA market fails to grow as expected, AMD faces material impairment risk. A write-down of even 20% would be ~$5B, exceeding three years of current net income.
2.9% ROE is below the cost of equity by a wide margin — AMD is currently destroying value on a GAAP basis. While Xilinx intangible amortization depresses earnings, the reality is that $57.6B in equity is generating only $1.6B in net income. The capital base must earn substantially more to justify its size.
Total liabilities relative to total assets are conservatively low. With $57.6B equity against $69.2B assets, AMD's balance sheet leverage is minimal. Long-term debt of only $1.7B is easily serviceable. Financial distress risk is negligible.
NVIDIA controls ~80%+ of the AI training GPU market with an entrenched CUDA ecosystem. AMD's ROCm software stack is improving but remains years behind in developer adoption, library optimization, and enterprise support. Winning in AI GPUs requires winning the software war — not just the silicon war.
AMD competes simultaneously against Intel (CPUs), NVIDIA (GPUs), and specialized players in FPGAs and embedded — each requiring distinct R&D investment, software ecosystems, and go-to-market strategies. Spreading resources across four fronts with a 35.8% expense ratio risks underinvesting in each.
Risk profile scores 55/100 (higher = safer) — reflecting material structural concerns. The 35.9% goodwill/assets ratio from Xilinx is the most pressing risk: a single acquisition created an enormous intangible asset base that must earn its keep. The 2.9% ROE means AMD is currently not clearing its cost of capital. Competing against NVIDIA's CUDA moat in AI while simultaneously fighting Intel in CPUs stretches resources thin. On the positive side, the balance sheet is clean with low debt and ample cash coverage.
Management
Ask about this section
This analysis is for educational purposes only and does not constitute investment advice.