AMD vs INTC: AMD Won the CPU War. Can Intel Pivot to Foundry?
AMD's EPYC server chips beat Intel's Xeon on performance-per-dollar, and hyperscalers are deploying them in growing volumes. The CPU war is effectively over in Intel's most profitable segment. Now Intel's survival plan is a bet on manufacturing — becoming a contract foundry to compete with TSMC. AMD is the proven share-gainer with an improving AI GPU story. Intel is a restructuring case with a massive binary outcome. These are very different risk profiles.
AMD's EPYC Victory Changed the Server Market Permanently
AMD's Zen architecture, launched in 2017, was the beginning of a turnaround that few believed possible. By Zen 3 (2020) AMD had closed the performance gap with Intel. By Zen 4 (Genoa, 2022) AMD was ahead on performance-per-dollar in most server workloads. By Zen 5 (Turin, 2024) the gap had become a chasm. Hyperscalers — who care deeply about total cost of ownership — started deploying AMD CPUs at scale because the economics justified it.
Intel's Xeon is still shipping and generating revenue, but the growth narrative is gone. Intel's data center CPU business is in secular decline on market share. The segment that once drove 60%+ of Intel's profits is now a defense play — holding onto existing contracts while AMD picks off new deployments.
Business Comparison
- EPYC CPUs gaining data center share vs Intel
- MI300X AI GPU: real inference deployments
- Fabless model — manufacturing via TSMC
- ~50% gross margins, improving trajectory
- Lower multiple than NVDA, more AI upside if GPU scales
- Legacy x86 CPU — still $30B+ revenue stream
- IDM 2.0: contract foundry for other chip designers
- Intel 18A process node — make or break
- CHIPS Act subsidies reduce foundry capex burden
- Trading near multi-decade lows — deep value or value trap
Why AMD Going Fabless Was a 15-Year Advantage
When AMD spun off its manufacturing operations in 2009 (creating GlobalFoundries), it looked like desperation. The company was near-bankrupt. In hindsight, it was the best strategic decision AMD ever made. Fabless means AMD designs chips but outsources production to TSMC. TSMC is the world's best chip manufacturer. AMD gets access to TSMC's cutting-edge process nodes without carrying the massive fixed-cost overhead of running fabs.
Intel chose the opposite path — owning its fabs is its competitive identity. But when Intel's manufacturing fell behind TSMC, it was stuck. No fab to outsource to. Intel was manufacturing its own chips on a process node that was worse than what TSMC could provide to AMD. That created a period from roughly 2017 to 2021 where AMD's chips were physically better because TSMC's manufacturing was better. Intel's IDM 2.0 strategy is an attempt to regain that manufacturing edge — but it requires billions in capex and flawless execution.
Who Should Buy Which
Technical Signals — What to Watch
- AMD RSI: AMD's RSI swings wider than Intel's — from below 30 on earnings misses to above 70 on AI GPU announcements. RSI below 35 after selloffs in AMD have been high-probability medium-term entries historically.
- INTC technicals: Intel near multi-decade lows is a base-building situation. Watch for a series of higher lows and a volume-backed breakout above key resistance before buying.
- AMD key catalyst: Quarterly MI300X revenue and full-year AI GPU guidance. Any significant upward revision to AI GPU revenue targets moves AMD materially higher.
- INTC key catalyst: Intel 18A yield confirmation and a major foundry customer announcement — these are the binary events that would break Intel out of its multi-year downtrend.
APEX scores both stocks daily across RSI, MACD, moving averages, volume, and 52-week position. Updated every market day.
Compare AMD vs INTC Live →