AI Infrastructure Showdown: AMD’s EPYC Surge, Supermicro’s Blackwell Breakthrough & Astera’s Rack-Scale Revolution
Welcome, AI & Semiconductor Investors,
AMD earnings roared ahead with a 57% jump in Data Center revenue, Supermicro unleashed its Blackwell GPUs paired with next-gen liquid-cooling to slash TCO, and Astera Labs quietly cemented its lead in PCIe 6 and CXL rack-scale fabrics. With new export restrictions and tariff flashpoints looming, these three leaders are rewriting the playbook for delivering tomorrow’s AI infrastructure. Let’s Chip In.
What The Chip Happened?
📰 AMD Posts Q1 Earnings Surge: AI, EPYC, and Ryzen Drive Growth
🚀 Supermicro’s “Blackwell” Lift-Off
⚡ Astera Labs Rack-Scale Connectivity Takes Center Stage
[AMD Delivers Strong Q1 2025 Earnings on Data Center and AI Momentum]
Read time: 7 minutes
Advanced Micro Devices, Inc. (NASDAQ: AMD)
📰 AMD Posts Q1 Earnings Surge: AI, EPYC, and Ryzen Drive Growth
What The Chip: AMD reported its Q1 2025 results on May 6, 2025, highlighting a 36% year-over-year revenue increase to $7.4 billion. The Data Center segment soared 57% on the back of strong EPYC server CPU demand and expanding adoption of Instinct AI accelerators. Management also noted macro headwinds and new export license requirements for certain China-bound GPUs but reiterated confidence in delivering double-digit full-year growth.
The Situation Explained:
🔍 Data Center Takes Center Stage: Q1 Data Center revenue hit $3.7B (up 57% YoY). CEO Dr. Lisa Su said, “We delivered an outstanding start to the year... Growth accelerated... driven by strength in our core businesses and expanding data center and AI momentum.”
🚀 AI Accelerator Momentum: Instinct MI300 and MI325 sales climbed, with large hyperscalers and enterprise customers ramping new AI deployments. AMD expects significant growth from its next-gen MI350 accelerators later in 2025, despite a projected $1.5B revenue impact from new China export restrictions.
💻 Client & Gaming Strong: AMD’s Client & Gaming revenue rose 28% YoY to $2.9B. Desktop Ryzen CPUs gained traction with gamers, while newer AI-enabled Ryzen mobile notebooks showed strong uptake. Dr. Su highlighted a “record client CPU ASP,” driven by higher-end, higher-margin products.
🎮 Gaming GPU Upswing: Radeon 9070 launch drove “record first-week sell-out” at retail, and AMD introduced FSR 4 machine-learning-based rendering to boost gaming performance and frame rates.
✈️ Embedded Market Rebound: Embedded revenue dipped 3% YoY to $823M as industrial, aerospace, and test/measurement demand slowly recovers. Management expects a return to year-over-year growth in the second half of 2025 as these end markets improve.
🤝 ZT Systems Acquisition: AMD recently acquired ZT Systems’ design team to bolster rack-level AI solutions. The move adds system design expertise and complements AMD’s CPU/GPU portfolio for customers looking to deploy large-scale AI clusters more quickly.
📈 Margins & Guidance: Non-GAAP gross margin reached 54% in Q1. Q2 outlook includes an $800M charge for inventory/reserves tied to restricted GPU shipments. Excluding that charge, gross margin would be around 54% again, and AMD still forecasts strong top-line growth fueled by EPYC CPUs and next-gen Instinct GPUs.
Why AI/Semiconductor Investors Should Care: AMD’s Q1 results underscore the rising tide of AI infrastructure spending. Its EPYC processors keep gaining share among cloud and enterprise customers, while the Instinct AI accelerator family aims for a bigger slice of the massive HPC and AI opportunity. Despite tariff and export control pressures, AMD’s combination of high-performance CPUs, GPUs, and newly acquired system design capability positions it favorably for sustained data center and AI market expansion.
Super Micro Computer, Inc. (NASDAQ: SMCI)
🚀 Supermicro’s “Blackwell” Lift-Off
What The Chip: Supermicro just reported Q3 2025 revenues of $4.6 billion on May 6, 2025, up 19% year-over-year, but down sequentially as customers paused orders for upcoming Blackwell GPUs. Management believes this slowdown is temporary as fresh AI platforms, combined with Supermicro’s direct liquid cooling (DLC) technology, power future quarters.
The Situation Explained:
🔋 AI Pipeline Growth: Over 70% of Supermicro’s Q3 revenues came from AI GPU products. Founder and CEO Charles Liang noted, “We have a very strong demand and also our coming soon DCBBS [Data Center Building Block Solutions].”
🔧 Inventory Write-Down: Non-GAAP EPS of $0.31 (down from $0.66 last year) took a hit from writing down old-generation GPUs. The transition to Blackwell-based gear caused customers to delay Hopper GPU purchases, impacting margins.
💧 Liquid Cooling 2.0: Supermicro’s new DLC-2 (direct liquid cooling) tech promises up to 40% energy savings, quieter data centers (~50 dB), and significantly faster deployments. Liang aims to ramp these solutions in the coming quarters.
🏗️ Data Center Building Block Solutions (DCBBS): This new offering integrates racks, cooling, compute, storage, and software into “plug-and-play” data center solutions, accelerating time-to-online (TTO). Management claims DCBBS can cut total cost of ownership (TCO) by 30%.
💼 Q4 Guidance: Management forecasts Q4 revenues of $5.6–$6.4 billion with about a 10% gross margin. They remain cautious about tariffs and macro conditions, but highlight a strong new GPU cycle.
📈 Global Capacity: New manufacturing sites in the U.S., Malaysia, and Taiwan expand total capacity. CFO David Weigand said the company’s flexible global supply chain mitigates risks from tariffs and helps serve customers worldwide.
⚠️ Risks: Ongoing macro pressures and potential changes in export rules could dampen short-term visibility. However, Supermicro’s domestic manufacturing footprint and robust AI pipeline should bolster resilience.
Why AI/Semiconductor Investors Should Care: Supermicro’s Q3 shows the growing pains of a fast-evolving AI market. Yet next-gen Blackwell GPUs, paired with Supermicro’s DLC-2 and DCBBS, offer a strong competitive edge for large-scale AI data center builds. With management expecting sequential rebounds in the coming quarters, investors may want to keep an eye on how quickly these new technologies translate into higher sales and improved margins.
Astera Labs, Inc. (NASDAQ: ALAB)
⚡ Astera Labs Rack-Scale Connectivity Takes Center Stage
What The Chip: Astera Labs held its Q1 2025 Earnings Call on May 6, unveiling $159.4 million in revenue, up 13% sequentially and 144% from last year, fueled by strong AI infrastructure demand. The team highlighted new design wins in scale-up and scale-out connectivity, plus momentum building for its upcoming Scorpio fabric switches.
The Situation Explained:
🔌 Robust Quarterly Performance: Astera Labs posted $159.4 million in Q1 revenue, with non-GAAP gross margin around 75%. CFO Michael Tate guided Q2 revenue to $170 - $175 million, citing steady demand despite macro headwinds.
⚙️ Aries & Taurus Growth: Aries PCIe retimers and Taurus Ethernet smart cable modules (SCMs) both expanded in AI rack deployments. Management said Aries retimers are increasingly used across GPUs and custom AI ASICs for scale-up connectivity, while Taurus helps in Ethernet-based scale-out systems at multiple hyperscalers.
🚀 New Scorpio Ramps: Scorpio P-Series is ramping in customized GPU-based racks, and Scorpio X-Series preproduction shipments start late this quarter. President/COO Sanjay Gajendra said, “Scorpio will be central to next-generation AI racks,” targeting scale-up interconnects with advanced diagnostics and telemetry.
🔭 PCIe 6 Leadership: The Aries 6 “Smart Gearbox” and PCIe 6 over optics demos solidify the company’s first-to-market push for next-generation speeds. These offerings bridge older PCIe 5.0-based systems to emerging PCIe 6.0 components, expanding Astera’s dollar content per platform.
🔒 CXL Memory Expansion: Astera’s Leo CXL controllers shipped preproduction volumes in Q1, and broader production ramps are expected in late 2025. This technology will let data centers seamlessly scale memory pools for both AI and general-purpose compute.
🔗 UALink Potential: Management highlighted the new Ultra Accelerator Link (UALink) 1.0 specification, aimed at delivering 200G per lane with up to 1,024 accelerators. “This can represent a multibillion-dollar market opportunity by 2029,” said CEO and Co-Founder Jitendra Mohan.
💬 Board Expansion: Astera announced the addition of Dr. Craig Barratt to its Board of Directors. “Craig brings a wealth of experience across execution and innovation that will help us expand our connectivity leadership in cloud and AI infrastructure,” Mohan said.
📈 Strong R&D Investment: To support future growth, the company will continue high R&D spend. “We are well positioned to continue to deliver above-market growth,” Mohan added, emphasizing deeper product breadth and better platform synergy with its COSMOS software suite.
Why AI/Semiconductor Investors Should Care: Astera Labs is quickly becoming an indispensable player for rack-scale AI connectivity, which is growing in complexity and magnitude as large language models and accelerated compute proliferate. Its leadership in PCIe 6, active cable modules, fabric switches, and UALink points to significant upside in an evolving data center ecosystem. Investors should watch Astera’s multi-protocol approach (PCIe, Ethernet, and CXL) and growing attach rate as signals that the company’s technology and the resulting revenue potential are poised to expand.
Youtube Channel - Jose Najarro Stocks
X Account - @_Josenajarro
Disclaimer: This article is intended for educational and informational purposes only and should not be construed as investment advice. Always conduct your own research and consult with a qualified financial advisor before making any investment decisions.
The overview above provides key insights every investor should know, but subscribing to the premium tier unlocks deeper analysis to support your Semiconductor, AI, and Software journey. Behind the paywall, you’ll gain access to in-depth breakdowns of earnings reports, keynotes, and investor conferences across semiconductor, AI, and software companies. With multiple deep dives published weekly, it’s the ultimate resource for staying ahead in the market. Support the newsletter and elevate your investing expertise—subscribe today!
[Paid Subscribers] AMD Delivers Strong Q1 2025 Earnings on Data Center and AI Momentum
Date of Event: May 6, 2025
Executive Summary
*Reminder: We do not talk about valuations, just an analysis of the earnings/conferences
Advanced Micro Devices, Inc. (AMD) recently reported its first quarter (Q1) 2025 earnings on May 6, demonstrating strong execution across Data Center, Client & Gaming, and Embedded segments. The company posted a 36% year-over-year revenue increase, reaching $7.4 billion, with non-GAAP gross margin at 54%. In the words of Dr. Lisa Su, AMD’s Chair and Chief Executive Officer, “We delivered an outstanding start to 2025 as year-over-year growth accelerated for the fourth consecutive quarter driven by strength in our core businesses and expanding data center and AI momentum.”
Notable metrics include non-GAAP operating income of $1.8 billion (up 57% year-over-year), reflecting a solid product mix shift towards high-margin Data Center offerings. Dr. Su added, “Despite the dynamic macro and regulatory environment, our first quarter results and second quarter outlook highlight the strength of our differentiated product portfolio and consistent execution.”
In addition to robust sales of EPYC™ server CPUs and Instinct™ AI GPUs, AMD’s Client & Gaming businesses also outperformed, propelled by strong adoption of high-performance Ryzen™ processors and new Radeon™ graphics cards. Meanwhile, CFO Jean Hu highlighted the company’s aggressive yet disciplined investment strategy to capture future growth opportunities. “We grew first quarter revenue 36% year-over-year and delivered significant earnings leverage as our business gains scale,” she said. Notably, there was a cautionary note on updated export licensing requirements, which reduced guidance for Data Center GPU revenue in the second quarter.
Overall, management maintains a confident outlook on achieving strong double-digit percentage revenue growth for fiscal year 2025, anticipating sustained Data Center expansion and improved demand in Embedded and Gaming.
Growth Opportunities
AMD is capitalizing on the surging demand for high-performance computing (HPC), artificial intelligence (AI), and cloud infrastructure, particularly in hyperscale and enterprise data centers. The latest 5th Gen EPYC™ Turin server processors continued to gain share, while AI-focused AMD Instinct GPUs benefited from rising needs in generative AI, inference, and distributed compute clusters.
AI Acceleration: The MI300 Series has attracted significant interest from cloud providers, HPC customers, and large AI model developers. Major cloud providers, such as Oracle and leading frontier model companies, plan to deploy clusters powered by AMD Instinct accelerators.