Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
The Lenovo ThinkStation PGX is an excellent compact AI-development appliance, but it is not a conventional expandable workstation. Built around NVIDIA’s GB10 Grace Blackwell Superchip, it combines a 20-core Arm CPU, an integrated Blackwell GPU, 128GB of soldered unified memory, NVIDIA DGX OS, and unusually fast networking in a chassis measuring roughly 1.1 liters.
That makes the PGX compelling for local inference, RAG, embeddings, model experimentation, and small multi-node AI labs. It is a poor choice for buyers who need Windows, upgradeable memory, multiple GPUs, extensive storage, professional graphics performance, or server-grade redundancy.
What is the Lenovo ThinkStation PGX?
The PGX is Lenovo’s first ThinkStation workstation based on an NVIDIA system-on-chip. It is a corporate-oriented implementation of NVIDIA’s GB10 platform, designed as a local sandbox for developing, testing, and preparing AI workloads for larger cloud or data-center systems.
Lenovo positions it for prototyping and fine-tuning while keeping sensitive data and intellectual property on premises. It ships with NVIDIA DGX OS, rather than Windows, and includes NVIDIA’s AI software environment.
#1 Best Overall
- Next-Gen Power: Intel Core Ultra 7 265K processor(Upto 5.5 Ghz, 20 Cores,20 Threads,36 MB Total L2 Cache) for elite multitasking and compute performance
- Upto Massive 128GB DDR5 RAM: Seamlessly run multiple virtual machines, large datasets, and memory-hungry applications
- Upto 12TB High-Speed Dual SSD Storage (3X4TB SSDs): Faster boot, load times, and file transfers with RAID-ready flexibility
- Windows11 Pro: STREAMLIMED AND INTUITIVE UI | Intelligent desktop | Personalize your experience for simpler efficiency | Powerful security built-in and enabled.
- ISV Certified: Optimized and tested for professional software stability (AutoCAD, Revit, SOLIDWORKS, Adobe, and more) Easy to Upgrade & Service: Tool-less design for hassle-free maintenance and future expansion
The word workstation describes its professional deployment and support positioning. It does not mean the PGX has the expansion capacity of a traditional ThinkStation tower.
Key specifications
| Component | Specification |
|---|---|
| Platform | NVIDIA GB10 Grace Blackwell Superchip |
| CPU | 20 Arm cores: 10 Cortex-X925 and 10 Cortex-A725 |
| GPU | Integrated NVIDIA Blackwell GPU with 48 SMs |
| AI performance | Lenovo-rated up to 1 petaflop FP4 |
| Memory | 128GB LPDDR5X-8533 unified memory, soldered |
| Memory bandwidth | 273GB/s, according to Lenovo’s specification |
| Storage | 1TB or 4TB NVMe, configuration dependent |
| Operating system | NVIDIA DGX OS |
| Networking | 10Gb Ethernet and two 200Gbps ConnectX-7 ports |
| Wireless | Wi-Fi 7 2×2 and Bluetooth |
| Power supply | 240W USB-C adapter |
| Dimensions | 150 × 150 × 50.5mm |
| Weight | Approximately 1.2kg |
Specifications can vary by Lenovo part number and market. The reviewed configuration used 4TB of storage, while Lenovo also lists 1TB models and other configurations in the United States. Check the official PSREF before ordering.
Why the 128GB unified memory matters
Unlike a typical desktop with separate system RAM and graphics memory, the PGX’s 128GB is shared by the CPU and GPU. Both processors can access the same memory pool, reducing the need to copy data between ordinary RAM and a discrete GPU’s VRAM.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsThis is the PGX’s defining advantage. Models that cannot fit into a 16GB, 24GB, or 32GB graphics card may fit into its unified memory, making local experimentation possible in a very small system. CPU preprocessing, embeddings, retrieval pipelines, and GPU inference can also work from a common address space.
Capacity is not the same as speed, however. The memory is soldered, limited to 273GB/s of bandwidth, and shared with the CPU. A model that loads successfully may still deliver poor tokens-per-second, limited batch sizes, or unacceptable latency. Quantization, context length, memory pressure, and the specific framework remain critical.
In other words, 128GB unified memory is not equivalent to 128GB of dedicated high-bandwidth GPU VRAM, nor does it guarantee that every large model will run practically.
Workloads the PGX suits
The PGX is strongest when used as a personal or departmental AI development machine:
Free tools Windows power users keep installed
One-click scans. No signup required.
- Local inference with quantized language models.
- Retrieval-augmented generation and embedding workflows.
- Model prototyping and small-to-medium fine-tuning experiments.
- Computer-vision, speech, transcription, and multimodal development.
- CUDA application development.
- Jupyter Notebook and PyTorch experimentation.
- ComfyUI and comparable image-generation workflows, where the model and software support the platform.
- Multi-node experimentation using NCCL and the high-speed network ports.
It is much less suitable for training large models from scratch, high-throughput production inference, multi-GPU workloads, heavy 3D rendering, or applications that depend on mature x86-only binaries.
Rank #2
- Processor: Intel Core Ultra 9 285 vPro Processor (E-cores up to 4.60 GHz P-cores up to 5.40 GHz)
- Memory: 64 GB DDR5 Storage: 1 TB SSD M.2 2280 PCIe Gen4 Performance
- Graphic Card: NVIDIA RTX 2000 Ada Generation 16GB GDDR6 Warranty: 1 Years
- Dimensions (H x W x D): 415mm x 180mm x 370mm / 16.3″ x 7.1″ x 14.6″ Weight: Starting at 13.61kg / 30.0lbs
Always distinguish model fit from performance fit. A model can fit in 128GB and still be too slow for interactive use or commercial deployment.
Physical design and serviceability
The 150mm-square chassis is the main reason to consider the PGX. It can sit on a development desk, in a small office, or alongside other nodes without the space and power demands of a tower workstation.
Internally, the design is far less flexible. The CPU, GPU, and memory are integrated into the platform. There is no conventional PCIe expansion slot, no 2.5-inch or 3.5-inch drive bay, and no way to upgrade the 128GB memory.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The meaningful internal upgrade is the SSD. Lenovo uses an M.2 2242 slot, a shorter and less common format than the familiar 2280 drive. The slot is electrically PCIe Gen5 x4, although the reviewed 4TB configuration used a PCIe Gen4 TLC SSD. Replacing it requires checking drive length, thermals, firmware compatibility, and NAND type. Access involves removing covers and components; this is not a tool-free service design. ServeTheHome’s internal-hardware coverage documents the layout.
Ports and 200Gbps networking
Rear connectivity includes four USB-C ports rated at 20Gbps, DisplayPort Alt Mode over USB-C, HDMI 2.1, 10Gb Ethernet, Wi-Fi 7, and two 200Gbps ConnectX-7 networking ports. Power is supplied through USB-C from the 240W adapter.
The 200Gbps interfaces are extraordinary on a system this small. They can help connect several GB10 machines for distributed experimentation, shared storage, and NCCL-based workloads. They are also easy to overvalue.
ServeTheHome reports that the ConnectX-7 connectivity is divided across two PCIe Gen5 x4 links and appears as four network interfaces. That topology can require deliberate Linux and distributed-computing configuration; it does not create an effortless single-node 200Gbps experience in every workload.
A multi-node deployment may also need compatible cables or transceivers, a suitable 200Gbps switch, correctly configured interfaces, and software that can exploit the bandwidth. For a single developer, the ports may be expensive unused capacity.
Rank #3
- DISCLOSURE - Brand New Computer has been resealed to upgrade Memory/SSD. 1 Year warranty by Issaquash Highlands Tech
- TINY YET POWERFUL - Part of Lenovo's Tiny portfolio, the ThinkStation P3 Tiny delivers workstation-class performance beyond the ThinkCentre M70q, M80q, and M90q. As a replace(12TD001UUS) upgrade, its ultra-compact 3.01 lb design is up to 96% smaller than a traditional desktop while helping optimize valuable workspace without compromising performance or reliability. Built to meet military-grade durability standards and ISV-certified for AutoCAD, SOLIDWORKS, ArcGIS, Creo, and MATLAB, the P3 Tiny provides stability, compatibility, and performance professionals depend on for mission-critical workloads
- HIGH PRODUCTIVITY - With the Intel 14-Core i5-14500 vPro Processor and Intel UHD 770 Graphics, this system delivers exceptional speed and responsiveness for professional workloads and graphics-intensive business applications. Configurable with memory options from 8GB to 64GB DDR5 RAM and storage options from 256GB to 2TB PCIe NVMe M.2 SSD, enabling smooth multitasking and fast loading across a wide range of applications
- RICH CONNECTIVITY - Front: 2 x USB-A, USB-C, Headphone / microphone combo jack; Rear: DisplayPort 1.4, 4 x USB-A, HDMI 2.1, Ethernet; Native dual-display support with up to 8K@60Hz via DisplayPort 1.4 and 4K@60Hz via HDMI 2.1 for an expanded workspace; Full-size USB keyboard with number pad, USB mouse for seamless productivity. With Wi-Fi 6 and Bluetooth 5.1, ensuring fast wireless connectivity and compatibility with a wide range of peripherals
- OPERATING SYSTEM - Windows 11 Pro with Copilot AI enables AI‑assisted productivity across business applications, supporting task organization, information retrieval, and collaboration. Built‑in security protections and enterprise‑grade management features help safeguard data, enforce policies, and simplify system administration, delivering a stable and secure operating environment for professional computing
Software: DGX OS rather than Windows
NVIDIA DGX OS provides a Linux-based environment with NVIDIA drivers and AI tools prepared for the GB10 platform. The reviewed software experience included NVIDIA Sync for SSH setup and tunneling, DGX Dashboard, NVIDIA AI Workbench, and playbooks for workflows such as ComfyUI and multi-unit NCCL scaling. ServeTheHome’s software and topology coverage provides the relevant context.
This can significantly shorten the path from unboxing to CUDA experimentation, but it also creates dependencies:
- DGX OS is not Windows.
- Arm Linux compatibility must be checked for every application and package.
- Drivers, CUDA, containers, and frameworks need compatible versions.
- NVIDIA’s release cadence influences the software environment.
- Lenovo firmware and system support may follow a different update schedule.
Before buying several units, test the exact dependency stack, including proprietary binaries, Python packages, containers, and deployment tools. Version-sensitive installation instructions should always be tied to the specific DGX OS image and software releases; a static command or menu path should not be assumed to remain valid.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Performance: capable, but not a benchmark champion
The GB10 platform is standardized, so different partner systems tend to perform similarly. In ServeTheHome’s testing, the PGX was within approximately 0.5% of the Gigabyte AI TOP ATOM in multi-core CPU performance and held an approximately 3.7% single-core advantage in that comparison.
In longer heat-soaked AI testing, the PGX was a few percent ahead of NVIDIA DGX Spark in several tests. Those differences are modest and benchmark-specific. Cooling is one of the areas where GB10 vendors can differentiate, but the platform’s workload behavior is more important than small ranking changes.
Lenovo’s “1 petaflop” figure is a peak FP4 AI-performance claim. It is not a general-purpose performance score and should not be compared directly with FP16, FP8, CPU, gaming-GPU, or professional-GPU results. Actual inference and development speed depend on model architecture, quantization, context, batch size, framework, and memory traffic.
The PGX’s strongest performance argument is therefore the combination of usable memory capacity, CUDA compatibility, compactness, and deployment simplicity—not that it replaces a high-end discrete GPU in every task.
Recommended Free Tools
Power and cooling
ServeTheHome measured approximately 38–40W at idle. Peak consumption was just under 200W, while sustained tested loads generally ranged from approximately 104W to 160W depending on the workload. Disconnecting the ConnectX-7 networking connection reduced tested power by roughly 18W.
Rank #4
- DISCLOSURE: Brand New Computer has been resealed to upgrade Memory/SSD. 1 Year warranty by oemgenuine.
- WORKSTATION-CLASS PERFORMANCE: The Lenovo ThinkStation P3 Tiny workstation is powered by the Intel Core i7-14700 processor (20 cores, 8 Performance + 12 Efficiency cores, up to 5.4GHz max turbo frequency, 33MB Intel Smart Cache), delivering exceptional processing power for CAD, engineering applications, content creation, data analysis, virtualization, and demanding professional workloadsall in a compact Tiny form factor.
- HIGH-SPEED MEMORY & STORAGE: Equipped with 16GB DDR5 memory and a 512GB PCIe NVMe SSD. Experience rapid boot times, seamless multitasking, fast application launches, and reliable performance for professional software, large datasets, and resource-intensive workflows.
- PROFESSIONAL GRAPHICS & MULTI-DISPLAY SUPPORT: Featuring Intel UHD Graphics 770 with support for multiple high-resolution displays via DisplayPort and HDMI, the ThinkStation P3 Tiny is ideal for business productivity, engineering projects, design work, financial analysis, and professional multitasking environments.
- ADVANCED CONNECTIVITY & FEATURES: Equipped with Wi-Fi 6, Bluetooth, and RJ-45 Gigabit Ethernet for fast, reliable networking. Includes DisplayPort, HDMI, and multiple USB ports for versatile connectivity. Its compact Tiny design maximizes workspace efficiency while delivering workstation-grade reliability and performance.
The cooling system uses two Delta fans with a substantial heatsink and heat-pipe assembly. The 240W adapter describes the platform’s power envelope, not constant wall consumption. Actual heat and fan behavior vary with ambient temperature, firmware, workload, and networking state.
The supplied coverage does not establish acoustic measurements, so the PGX should not be advertised as silent or definitively quiet.
Is it corporate-ready?
The PGX is corporate-friendly primarily because of Lenovo’s procurement, warranty, support, and deployment channels. A standardized appliance can be easier for an IT team to purchase and replicate than a custom Linux workstation. Local processing may also help organizations keep sensitive data on premises.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →That does not make it a server. The PGX should not be assumed to include ECC system memory, redundant power supplies, hot-swappable parts, a server BMC, rack-mount infrastructure, or server-equivalent fleet management. Security still depends on OS hardening, credentials, network controls, patching, and organizational policy.
Warranty terms vary by exact part number, geography, and sales channel. One referenced US 1TB configuration lists a one-year Premier warranty; verify the coverage for the specific machine before purchase.
1TB or 4TB?
| Configuration | Best for | Trade-off |
|---|---|---|
| 1TB | Developers using NAS, external NVMe, or shared storage | Lower purchase price, but limited room for models, containers, and datasets |
| 4TB | Users who want models and working data locally | More convenient, but the premium may be substantial and the drive remains a less-common 2242 format |
The 1TB model is sensible when the PGX will use network storage or external drives. Choose 4TB when local access and simplicity matter more than minimizing the purchase price. Do not pay for a premium configuration unless its storage, warranty, or support terms justify the difference.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Price as of August 16, 2026
Lenovo’s US listings showed configuration-dependent prices rather than one universal PGX price. The 30KL0002US 1TB model was listed at approximately $5,079 on one buy page. Other listed part numbers were approximately $5,949 and $6,729, while other commercial pages showed prices as high as approximately $7,879.
These are date-stamped observed prices, not a permanent MSRP. Lenovo pricing can vary by part number, market, account status, promotion, membership, and business channel. Check the current Lenovo listings and identify the exact configuration before comparing it with alternatives.
Best Value
- Release Year: 2025 model featuring the latest workstation technology and design innovations
- AI-powered Technology: Yes, equipped with artificial intelligence capabilities for enhanced performance and productivity
- Number of Processors Supported: Single processor configuration with support for one processor unit
- Number of Processors Installed: One processor pre-installed and ready for immediate use
- Processor Manufacturer: Intel processor technology providing reliable and powerful computing performance
ServeTheHome’s March 10, 2026 review described a 1TB PGX at roughly $4,100 and compared it with a $4,699 DGX Spark configuration. That older comparison should not be substituted for current pricing.
PGX versus the alternatives
NVIDIA DGX Spark
DGX Spark is NVIDIA’s own GB10 personal AI supercomputer. Because the underlying platform is heavily standardized, expect broadly similar performance. The decision comes down to current price, storage, availability, support relationship, and whether NVIDIA’s own branding matters more than Lenovo’s corporate channel.
See the NVIDIA GB10 marketplace for current reference and partner products.
Other GB10 partner systems
Systems such as the Gigabyte AI TOP ATOM, ASUS Ascent GX10, Dell Pro Max with GB10, and HP ZGX Nano G1n should be compared on price, storage, cooling, warranty, regional availability, accessories, and firmware policy. No supplied evidence supports calling the PGX universally the fastest GB10 system.
AMD Ryzen AI Max+ systems
An AMD integrated-GPU system may be preferable when x86 compatibility, a conventional desktop operating system, flexible storage, or lower cost matters most. The trade-off is a different AI software environment without the PGX’s CUDA-first stack and ConnectX-7-oriented scale-out design.
Conventional NVIDIA workstations
A tower with an RTX or RTX PRO GPU is the better choice for Windows, professional graphics applications, multiple GPUs, more GPU-memory options, PCIe expansion, upgradeable components, and extensive local storage. It will generally consume more power and occupy more space.
Cloud GPUs
Cloud infrastructure remains a valid alternative, particularly for bursty workloads. Compare GPU rental, persistent storage, data egress, idle time, security and compliance, team access, utilization, support, and data-transfer costs. Local hardware is not automatically cheaper; its value improves when utilization is high and keeping data on premises is important.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Who should buy the PGX?
| Buyer | Verdict |
|---|---|
| Corporate AI prototyping team | Strong fit |
| Individual developer wanting local CUDA | Good fit if the budget allows |
| Large-model production inference | Usually not |
| Gaming or content creation | Poor fit |
| Windows-first user | Poor fit |
| Multi-node AI lab | Strong fit, with networking costs included |
| Upgrade-focused workstation buyer | Poor fit |
| Privacy-sensitive organization | Potentially strong fit |
Final verdict
The Lenovo ThinkStation PGX is a remarkably dense NVIDIA AI development appliance. Its 128GB unified memory, Blackwell software ecosystem, tiny chassis, and 200Gbps networking make it far more interesting than an ordinary mini-PC.
It earns its place in a corporate lab or departmental development environment when local data handling, standardized CUDA tooling, Lenovo procurement, and multi-node experimentation matter. The 1TB model is the practical value choice for teams with network storage; the 4TB version is more convenient for local model work.
Buy something else if you need a traditional workstation. Soldered memory, one internal M.2 2242 drive, no conventional expansion, Arm/Linux compatibility constraints, and the lack of server redundancy define the PGX as a specialized AI appliance—not a universal desktop replacement.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
