Astera Labs reported record quarterly revenue of $230.6 million for the third quarter of fiscal 2025, up 20% from the prior quarter and 104% year over year. Growth was fueled by strong demand across its signal conditioning, smart cable module (SCM), and switch fabric product families as hyperscaler AI platforms entered production. GAAP net income rose to $91.1 million, with a 24% operating margin and 76.2% gross margin.
CEO Jitendra Mohan said Astera Labs continues to see strong traction for its PCIe 6 and Taurus Ethernet SCM portfolios, alongside expanding Scorpio fabric switch wins across hyperscaler platforms. The company expects Q4 revenue between $245 million and $253 million, with non-GAAP EPS of approximately $0.51. Astera also announced plans to acquire aiXscale Photonics GmbH, a fiber-chip coupling specialist, expanding its roadmap beyond copper to photonic interconnects for future rack-scale AI systems.
At the 2025 Open Compute Project Global Summit, Astera showcased a full suite of connectivity solutions spanning PCIe, UALink, Ethernet, and CXL, emphasizing open standards for AI Infrastructure 2.0. It also joined Arm Total Design to provide multi-protocol chiplet connectivity for Neoverse-based SoCs, broadening its reach into custom AI silicon development.
• Q3 revenue: $230.6M (+20% QoQ, +104% YoY)
• Non-GAAP gross margin: 76.4%; operating margin: 41.7%
• Scorpio fabric switch adoption expanding across hyperscalers
• aiXscale Photonics acquisition to accelerate optical roadmap
• Joined Arm Total Design for chiplet connectivity solutions
• Q4 revenue outlook: $245–253M; non-GAAP EPS ≈ $0.51
Highlights from the Investor Conference Call:
• Scorpio P-Series ramped with a lead hyperscaler; new design wins span multiple AI platforms using NVIDIA GB300/B200 and custom accelerators.
• PCIe 6 products now account for over 20% of revenue, confirming leadership in next-gen interconnects.
• Scorpio X Series enters production shipments in Q4 2025, with high-volume ramp expected in 2026.
• Astera engaged with 10 major AI platform providers across PCIe and UALink scale-up designs—expected to generate multi-year revenue through 2029.
• UALink adoption gaining momentum; first Astera-based UALink switches expected in 2026, with revenue contribution from 2027.
• Taurus Ethernet SCMs saw strong 400G shipments and early 800G engagements; 800G revenue expected to ramp in early 2026.
• Optical interconnects to begin generating material revenue in 2028–2029; aiXscale acquisition adds key packaging and photonics expertise.
• Scorpio X as anchor socket drives multi-product pull-through—retimers, controllers, and cable modules—boosting content per accelerator to “thousands of dollars.”
• China opportunity rising as local hyperscalers adopt PCIe-based scale-up due to export restrictions on high-speed interconnect IP.
• Q4 growth expected across all product families; Taurus SCMs cited as strongest contributor.
• Company expects to end 2025 with over 700 employees, up 60% year to date.
• CFO reaffirmed long-term gross margin model of ~70%, with near-term mix-driven fluctuation due to increased hardware module sales.
“Astera Labs delivered strong financial results in Q3 with revenue growing by 20% sequentially to a new record level of $230.6 million,” said Jitendra Mohan, CEO of Astera Labs. “Our rack-scale vision is further strengthened by the proposed acquisition of aiXscale Photonics, which will help us meet the performance and reach requirements of future AI systems.”
🌐 Analysis:
Astera Labs is capitalizing on the rapid buildout of AI-scale infrastructure, positioning itself at the core of the PCIe 6 and UALink ecosystems. Its Scorpio fabric and Taurus SCM platforms are emerging as foundational elements for rack-scale AI systems, directly aligning with hyperscaler roadmaps. The aiXscale acquisition marks a strategic move into optical connectivity, competing with efforts from Broadcom, Marvell, and Nvidia’s CPO initiatives. With multi-generational design wins and growing hyperscaler penetration, Astera is shaping the connective backbone of AI Infrastructure 2.0.
