AMD has moved to expand its space-grade compute line-up, qualifying a new organic lidless package for its Versal AI Core XQRVC1902 adaptive SoC and extending space-flight ambitions out to 15 years. The packaging — engineered to withstand thermal and mechanical extremes — is being positioned for geosynchronous satellites, lunar programmes, and deep-space probes, and is undergoing Class Y qualification under MIL-PRF-38535.
The update forms part of a broader roadmap across AMD’s adaptive computing families. Space-grade variants of the Versal RF Series and Versal AI Edge Series Gen 2 devices are also being pushed through Class B and Class Y qualification. The latter introduces a reported ten-fold increase in scalar compute for on-board processing and data management, aimed at missions that increasingly rely on autonomous decision-making and localised AI workloads.
Class B and Class Y designations remain the benchmark for high-reliability components in aerospace. Class B is typically selected for low Earth orbit and shorter missions where cost pressures are acute, whereas Class Y is reserved for long-duration or deep-space operations that demand extended screening and documentation. AMD is positioning its enhanced packaging technology to close gaps in thermal performance, where lidless designs can offer advantages in heat dissipation during prolonged exposure to radiation and thermal cycling.
Ken O’Neill, Space Architect at AMD, said: “By combining proven adaptive technology with advanced packaging, we’re helping enable our customers to innovate faster to transform the future of space systems. These new space-rated devices extend our leadership in adaptive computing, combining unmatched functional integration, high-performance computing, and long-term reliability for the next generation of orbital, deep space, and exploration missions.”
Technical gains across the portfolio are significant. The Versal AI Core XQRVC1902 targets compute-heavy payloads, bringing higher vector algorithm throughput, more system logic cells, and increased on-chip memory compared with earlier space-grade devices. The Versal RF devices consolidate RF conversion, signal processing, and programmable logic into single monolithic packages — a consolidation move that will matter for mass- and power-constrained platforms. Meanwhile, the Versal AI Edge Gen 2 family integrates Arm Cortex-A78AE and R52 processors with next-generation AIE-ML v2 engines for real-time inference and on-board autonomy.
AMD claims these integration levels will allow designers to replace custom ASICs, which can run into tens of millions of dollars and years of qualification work. By offering flight-qualified, off-the-shelf SoCs with embedded AI acceleration and radiation tolerance, the company is tackling a longstanding bottleneck: delays in aligning commercial compute roadmaps with the slower cycles of aerospace engineering.
Spaceflight timelines remain central. Sampling for the XQRVC1902 is slated for 2026, with flight-qualified units due a year later. Space-rated versions of the Versal RF and Versal AI Edge Gen 2 devices follow in 2029 — a cadence that aligns with the next wave of satellite constellations and deep-space science missions. With multi-year procurement windows now common across government and commercial programmes, AMD’s roadmap gives integrators a clearer view of compute availability across the second half of the decade.
The company’s push reflects a wider shift in the sector: payloads are growing more software-defined, mission autonomy is expanding, and the compute envelope is migrating from ground operations to orbit. As aerospace designers look to cut mass, power demand, and qualification overhead, high-integration adaptive SoCs are becoming an increasingly practical alternative to bespoke hardware.
