Nvidia Ends Jetson TX2 and Xavier Supply Early

Nvidia is pulling the plug on several of its older Jetson embedded AI modules ahead of schedule, citing critical LPDDR4 memory shortages as the primary driver. Canadian supplier Connect Tech revealed the accelerated timeline, warning customers that final orders for affected Jetson TX2 and Xavier products must be placed before July 1, 2025.

The move impacts a range of modules still widely deployed in industrial automation, robotics, and edge AI applications — raising urgent questions for developers and enterprises that rely on these platforms for long-term projects.

Key Facts at a Glance

• Affected products: Jetson TX2 NX, Jetson TX2i (all SKUs), Jetson AGX Xavier 32GB Industrial, and Jetson Xavier NX (8GB/16GB)
• Order deadline: July 1, 2025 — no new orders accepted after this date
• NCNR status: All TX2 and Xavier models are now marked 'Non-Cancellable, Non-Returnable'
• Existing orders: Convert to NCNR status on July 15, 2025
• Final shipment date: July 15, 2026
• Newer Orin and Thor modules remain unaffected due to their use of LPDDR5 memory

LPDDR4 Shortages Drive the Decision

The root cause behind Nvidia's accelerated end-of-life decision is a growing shortage of LPDDR4 memory chips. As the semiconductor industry shifts its production capacity toward newer memory standards — particularly LPDDR5 and LPDDR5X — the supply of older LPDDR4 components is drying up faster than anticipated.

Memory manufacturers like Samsung, SK Hynix, and Micron have been gradually reallocating their fabrication resources to meet surging demand for next-generation memory. This demand is fueled largely by the AI boom, with data centers, smartphones, and automotive systems all requiring higher-bandwidth, more power-efficient memory solutions.

For Nvidia, this creates an untenable supply chain situation. The Jetson TX2, originally launched in 2017, and the Xavier series, introduced in 2018, were both designed around LPDDR4 memory. Securing sufficient quantities of these older chips to sustain production has become increasingly difficult and cost-prohibitive.

This is not an isolated phenomenon. Across the electronics industry, legacy components are becoming harder to source as foundries and memory fabs prioritize newer, more profitable product lines. The AI hardware boom has only accelerated this transition.

Which Products Are Affected — and Which Are Safe

The specific modules facing early termination represent Nvidia's previous-generation embedded AI lineup. Here is a detailed breakdown:

Products being discontinued:
- Jetson TX2 NX — compact module popular in drones and small robotics
- Jetson TX2i (all SKUs) — industrial-grade variant designed for harsh environments
- Jetson AGX Xavier 32GB Industrial — high-performance module for factory automation
- Jetson Xavier NX 8GB and 16GB — mid-range modules widely used in edge AI deployments

Products that remain available:
- Jetson Orin series (Orin NX, Orin Nano, AGX Orin) — uses LPDDR5 memory
- Jetson Thor — Nvidia's upcoming next-generation platform, also built on LPDDR5

The distinction is clear: any module relying on LPDDR4 memory is at risk, while products built around LPDDR5 remain fully supported and in active production. This effectively draws a line between Nvidia's legacy and current-generation embedded AI offerings.

Tight Deadlines Create Urgency for Customers

Connect Tech's announcement outlines a compressed timeline that leaves customers with limited room to maneuver. All TX2 and Xavier models have already been flagged with NCNR (Non-Cancellable, Non-Returnable) status, meaning any orders placed cannot be modified or reversed.

New orders must be submitted by July 1, 2025. After that date, the window closes permanently. Existing orders that are still in the pipeline will automatically convert to NCNR status on July 15, 2025, with a final shipment deadline of July 15, 2026.

This gives affected customers roughly 14 months from the NCNR conversion date to receive their last shipments. While that may sound like a reasonable Runway, the reality for many industrial and embedded systems customers is more complicated.

Long product lifecycles are the norm in industrial settings. Manufacturing lines, autonomous vehicles, and medical devices often rely on specific hardware for 5 to 10 years or more. An unexpected early termination can trigger costly redesign efforts, requalification processes, and supply chain disruptions.

Impact on the Embedded AI and Robotics Ecosystem

The Jetson platform has been a cornerstone of the edge AI ecosystem for nearly a decade. Thousands of companies — from startups building delivery robots to large manufacturers deploying computer vision on factory floors — have built their products around Jetson modules.

The TX2 and Xavier families, in particular, found strong adoption in several key segments:

• Industrial automation: Vision inspection systems, predictive maintenance
• Autonomous machines: Agricultural robots, warehouse AMRs (autonomous mobile robots)
• Drone technology: Mapping, surveying, and delivery drones
• Medical devices: AI-assisted diagnostic equipment
• Smart cities: Traffic monitoring, surveillance, and analytics

For companies in these sectors, the early termination means an accelerated migration path to the newer Jetson Orin platform. While the Orin series offers significantly better performance — up to 275 TOPS (trillion operations per second) compared to Xavier's 32 TOPS — the migration is not always straightforward.

Hardware form factors differ, software stacks may need updating, and thermal and power profiles change. For products already certified and deployed in the field, swapping out the compute module can require extensive re-engineering and recertification.

Migration Path: Moving from Xavier to Orin

Nvidia has positioned the Jetson Orin family as the natural successor to both TX2 and Xavier. The Orin lineup spans a wide performance range, from the entry-level Orin Nano (delivering up to 40 TOPS) to the flagship AGX Orin (reaching 275 TOPS).

Key advantages of migrating to Orin include:

• Significantly higher AI performance — up to 8x improvement over Xavier
• LPDDR5 memory — faster bandwidth and long-term supply stability
• Ampere GPU architecture — more efficient inference and training capabilities
• Software compatibility — Nvidia's JetPack SDK maintains backward compatibility for most applications
• Longer support lifecycle — Orin is expected to remain in production well into the late 2020s

However, the transition cost is real. Companies must factor in hardware redesign, software porting, testing, and potential recertification depending on their industry. For cost-sensitive applications where TX2 provided an affordable solution, the jump to Orin may also come with a higher bill of materials.

Connect Tech and other Jetson ecosystem partners are likely to offer carrier boards and integration services to ease the transition, but the clock is ticking.

Broader Industry Context: Legacy Chips Face an Uncertain Future

Nvidia's decision reflects a broader trend sweeping the semiconductor industry. As AI drives unprecedented demand for cutting-edge silicon, older chip architectures and the components they depend on are being phased out faster than ever.

The LPDDR4 shortage is just one symptom. Legacy process nodes at foundries like TSMC and Samsung Foundry are also seeing reduced investment, making it harder to produce older chips economically. This creates a cascading effect: when one critical component becomes scarce, entire product lines can become unviable.

For the embedded and industrial markets, this trend poses a unique challenge. Unlike consumer electronics, where product cycles are measured in months, industrial systems operate on multi-year timelines. The mismatch between semiconductor industry cadence and industrial deployment cycles is becoming a significant pain point.

Companies that proactively plan for component obsolescence and maintain flexible hardware architectures will be better positioned to weather these disruptions.

What This Means for Developers and Businesses

If your organization currently deploys or plans to deploy Jetson TX2 or Xavier-based systems, the message is clear: act now. The July 1 ordering deadline leaves little time for deliberation.

Short-term, companies should audit their current inventory and project future needs to determine whether a final bulk order is warranted. Long-term, the focus should shift to evaluating the Jetson Orin platform and beginning the migration process.

For new projects, there is no reason to start with TX2 or Xavier at this point. The Orin family offers superior performance, better power efficiency, and — critically — a secure supply chain built on current-generation memory technology.

Looking Ahead: Nvidia's Jetson Roadmap

Nvidia continues to invest heavily in the embedded AI market. The upcoming Jetson Thor platform, based on Nvidia's next-generation architecture, promises even greater performance and is expected to target advanced robotics, autonomous vehicles, and humanoid robots.

With the TX2 and Xavier chapters closing, Nvidia is clearly signaling that the future of edge AI lies with modern architectures and current-generation components. The LPDDR4 shortage may have accelerated the timeline, but the direction was inevitable.

For the thousands of developers and companies in the Jetson ecosystem, the transition will require effort — but it also represents an opportunity to unlock significantly more capable AI at the edge. The key is to start planning today, before the final order window slams shut on July 1.