Artificial intelligence is rapidly transforming the working methods of businesses, researchers, developers, and innovation centers.
Large language models, conversational assistants, advanced data analytics, and intelligent automation are creating a growing demand for platforms capable of providing significant computing power directly at the workstation.
In many organizations, cloud infrastructures represent an excellent solution. However, some teams also wish to keep their data in-house, reduce recurring cloud usage costs, or have a development environment independent of an external connection.
It is precisely with this in mind that Lenovo has developed the ThinkStation PGX, a workstation designed from the ground up for the development, experimentation, and inference of advanced artificial intelligence models.
Unlike a conventional workstation primarily intended for CAD or classic engineering applications, the ThinkStation PGX has been designed to meet the needs of modern AI.
- AI model development;
- fine-tuning of generative models;
- local inference;
- data science;
- computer vision;
- intelligent automation;
- academic research;
- innovation and prototyping;
- development of AI-based products.
This platform marks an important step in the evolution of professional workstations by bringing the power usually reserved for data centers directly to the specialist’s desk.
The Lenovo ThinkStation PGX enables the local execution of next-generation artificial intelligence models while maintaining control over data, the development environment, and computing power.
Why is the demand for artificial intelligence exploding?
A few years ago, many artificial intelligence projects could be developed on relatively conventional professional computers.
The arrival of generative AI has profoundly changed this reality.
Today, organizations want to:
- create their own AI assistants;
- develop conversational solutions;
- automate document processing;
- analyze large volumes of data;
- leverage advanced language models;
- deploy intelligent applications;
- keep their data confidential;
- reduce their dependence on the cloud;
- accelerate their innovation cycles.
These new use cases require more memory, more bandwidth, and more computing power than traditional workstations.
That is why a new category of equipment is emerging today: platforms designed specifically for artificial intelligence.
The automotive analogy
To understand the positioning of the Lenovo ThinkStation PGX, let’s imagine four categories of vehicles:
- a family car;
- a commercial van;
- a specialized vehicle;
- a race car.
A family car is for daily commuting. A van is mainly for work. A specialized vehicle meets specific technical needs. A race car aims for extreme performance.
The IT market works in a similar way.
A consumer computer meets daily needs. A professional computer promotes productivity and security. A traditional workstation meets the needs of design and engineering. An AI platform like the ThinkStation PGX is designed for massive computations and modern generative models.
Positioning of the Lenovo ThinkStation PGX
| Automotive World | IT Category | Primary Use | Examples |
| Family car | Consumer computer | Web browsing, office work, and entertainment | Lenovo IdeaPad, HP OmniBook, Dell Inspiron |
| Commercial van | Professional computer | Productivity and business management | ThinkPad, EliteBook, Latitude |
| Specialized vehicle | Workstation | Engineering, architecture, simulation, and design | ThinkPad P, HP ZBook, Dell Precision |
| Race car | Advanced AI platform | AI development, training, and inference | Lenovo ThinkStation PGX |
Introducing the Lenovo ThinkStation PGX
The Lenovo ThinkStation PGX represents a new generation of workstations designed for artificial intelligence.
Its architecture has been optimized to enable the development, experimentation, fine-tuning, and inference of advanced models directly from the desk.
The platform notably includes:
- up to 1 petaflop of AI performance;
- 128 GB of unified LPDDR5X memory;
- memory bandwidth of up to 273 GB/s;
- 1 TB M.2 SSD storage;
- 10 Gb Ethernet connectivity;
- high-speed QSFP interfaces;
- Bluetooth and wireless networking;
- pre-installed NVIDIA DGX OS;
- a ready-to-use AI software stack.
With its 128 GB of unified memory and its ability to locally run models with up to 200 billion parameters, the ThinkStation PGX clearly targets developers, researchers, and organizations working with advanced artificial intelligence.
Impressive AI power directly on your desk
One of the most remarkable features of the ThinkStation PGX is its computing power dedicated to artificial intelligence.
According to Lenovo, the platform can deliver up to 1 petaflop of AI performance, power that allows for the development and execution of large-scale generative models without systematically relying on external infrastructure.
This capability opens the door to numerous use cases:
- corporate conversational agents;
- AI application development;
- intelligent document analysis;
- scientific research;
- computer vision;
- machine learning;
- natural language processing;
- research and innovation laboratories.
For many organizations, the ability to run these models locally can also help improve data privacy and reduce certain operating costs related to cloud resources.
128 GB of unified memory for modern models
One of the main challenges in artificial intelligence concerns the memory available to run models and manipulate data.
The ThinkStation PGX is equipped with 128 GB of unified LPDDR5X memory, offering a bandwidth of up to 273 GB/s.
This architecture allows for processing large datasets while running next-generation AI models.
This capacity is particularly interesting for:
- inference of large language models;
- fine-tuning of existing models;
- advanced data analytics;
- research laboratories;
- computer vision;
- multimodal applications;
- corporate assistants.
For many developers, having such an amount of memory directly on their workstation can significantly accelerate experimentation phases.
NVIDIA DGX OS: a ready-to-use platform
Beyond hardware power, the ThinkStation PGX also distinguishes itself with its software environment.
The platform comes with NVIDIA DGX OS as well as an AI software stack designed to accelerate prototyping, development, and deployment.
This approach allows teams to devote more energy to innovation rather than configuring their technology environment.
Users can thus work in an environment already prepared for:
- machine learning;
- deep learning;
- generative models;
- data analytics;
- AI innovation projects;
- technology demonstrators.
Scalable architecture thanks to 200 Gbit/s QSFP
AI needs evolve rapidly. A high-performance infrastructure must be able to evolve with projects.
The ThinkStation PGX incorporates QSFP interfaces capable of supporting speeds up to 200 Gbit/s.
This architecture notably allows for:
- interconnecting multiple PGX systems;
- increasing computing capabilities;
- supporting even larger models;
- improving resource sharing;
- gradually building a local AI infrastructure.
Two ThinkStation PGX systems can be combined to support models with up to 405 billion parameters.
The ability to scale the platform represents a strategic advantage for organizations that anticipate growth in their artificial intelligence projects.
A compact and easy-to-deploy platform
Despite its impressive power, the ThinkStation PGX maintains a particularly compact form factor.
- 150 mm wide;
- 150 mm deep;
- 51 mm high;
- approximately 1.2 kg.
This compactness facilitates its integration into:
- offices;
- laboratories;
- research centers;
- academic environments;
- development teams;
- innovation spaces.
Multiple units can be deployed without having to immediately invest in a full server room.
Connectivity designed for professional environments
The Lenovo ThinkStation PGX offers connectivity tailored to modern infrastructures.
- 20 Gbit/s USB-C ports;
- HDMI output;
- 10 Gigabit Ethernet;
- QSFP connectors;
- Bluetooth;
- wireless networking;
- high-speed network connectivity.
This connectivity makes it easy to integrate the platform into existing technology infrastructures.
Who is the ThinkStation PGX for?
| Profile | Uses | PGX Benefits |
| AI Developer | Development, fine-tuning, and experimentation | 1 petaflop AI and 128 GB memory |
| Data Scientist | Advanced data analytics and statistical models | High memory bandwidth |
| University | Research and teaching | Local AI infrastructure |
| AI Startup | Prototyping and rapid innovation | Compact, ready-to-use solution |
| Research Center | Advanced models and specialized processing | Scalable QSFP architecture |
| Innovative Enterprise | AI assistants, automation, and document analysis | Local data control |
Cloud or local execution?
Cloud services play an important role in many artificial intelligence projects.
However, some organizations also wish to have a platform capable of running their models locally.
This approach can offer:
- better data control;
- a reduction in certain recurring expenses;
- increased autonomy;
- faster experimentation;
- better cost predictability;
- a controlled development environment.
The ThinkStation PGX does not necessarily replace the cloud, but it can become a strategic complement for many organizations.
Key strengths of the ThinkStation PGX
- up to 1 petaflop of AI performance;
- 128 GB of unified LPDDR5X memory;
- memory bandwidth of 273 GB/s;
- support for models up to 200 billion parameters;
- up to 405 billion parameters with two systems;
- 1 TB SSD storage;
- 10 Gb Ethernet;
- QSFP up to 200 Gbit/s;
- pre-installed NVIDIA DGX OS;
- ready-to-use AI platform;
- compact form factor;
- scalable architecture.
Points to check before purchasing
- Nature of the models to be executed;
- Volume of data to process;
- Long-term storage needs;
- Existing network infrastructure;
- Planned growth of AI projects;
- Compliance and confidentiality requirements;
- User training;
- Integration with existing tools.
Conclusion
The Lenovo ThinkStation PGX represents a new generation of computing platforms designed specifically for artificial intelligence.
With up to 1 petaflop of AI performance, 128 GB of unified memory, a scalable architecture, and a ready-to-use software environment, it offers researchers, developers, and businesses a particularly compelling solution for running advanced AI models locally.
Its compact form factor, high-speed connectivity, and ability to support very large models make it a platform capable of supporting organizations in their current projects while preparing for the innovations of tomorrow.
The Lenovo ThinkStation PGX brings the power of advanced artificial intelligence infrastructures directly to the desk of the professionals building the solutions of tomorrow.
Discover the Lenovo ThinkStation PGX at ÉQUIMAF.ca
Are you looking for a high-performance platform for the development, experimentation, or inference of artificial intelligence models?