If you are studying Data Science, AI, or working in data analytics with a budget of 22–55 million VND, you may have encountered a familiar paradox: a machine powerful enough to run notebooks isn’t necessarily capable of training models. A data science laptop needs to handle Python, SQL, and large notebooks while maintaining stable multitasking and remaining lightweight enough to carry to campus. The four options below feature Core Ultra, Ryzen AI, up to 64GB of RAM, and 14–16 inch displays. Three of these models rely on integrated GPUs, meaning heavy CUDA tasks will require Colab or a server; only one model features a discrete NVIDIA RTX GPU for local CUDA execution, though it is limited to 8GB of VRAM.
What should be your priority when choosing a data science laptop?
The CPU determines processing speed for data, running multiple notebooks, and executing source code. The Core Ultra 7 165H features 16 cores, 22 threads, and a 28W TDP; the Ryzen AI 5 340 offers 6 cores, 12 threads, a 28W TDP, and a 50 TOPS NPU; while the Core Ultra 7 255HX delivers up to 20 cores and a 55W TDP for heavy computational demands. For students studying data science, Hung Phat recommends prioritizing high core-count CPUs if you frequently run Python, data visualizations, and multiple applications simultaneously.
16GB of RAM is sufficient for data science assignments, SQL, Power BI, and medium-sized datasets. 32GB is more suitable for running virtual machines or processing large data tables, while 64GB provides a significant advantage for massive notebooks. Note that LPDDR5x RAM on the Surface and OmniBook models is soldered, whereas the Legion 5 uses DDR5 SO-DIMM, which is upgradeable. A 512GB SSD is a reasonable starting point, but 1TB is more convenient if you plan to store significant amounts of data locally.
The integrated GPUs on the three Surface and OmniBook models are excellent for rendering charts and light AI tasks but do not support CUDA. Only the Legion 5 features a discrete NVIDIA RTX GPU, allowing for local CUDA execution; however, the 8GB VRAM still limits model size. You should also consider screen aspect ratios, weight, and data security. Incidents like the Education Data Leak demonstrate that academic data requires serious backup and access control.
How do these four data science student models differ?
Microsoft Surface Laptop 6 15-inch is ideal for data scientists requiring large amounts of RAM and a display that can show many rows of data. The specs include a Core Ultra 7 165H, 64GB LPDDR5x RAM, a removable 1TB PCIe 4.0 SSD, and a 15-inch touchscreen with 2496 × 1664 resolution. The 3:2 aspect ratio is highly convenient for viewing notebooks, spreadsheets, and source code. The estimated price is around 49 million VND. The downsides are the 60Hz screen, soldered RAM, and lack of RTX/CUDA support.
HP OmniBook X Flip Next Gen AI 16 is suited for those who need a large screen for desk-based study, with prices ranging from 25.5 to 28.5 million VND. It features a Ryzen AI 5 340, 16GB soldered LPDDR5x-7500 RAM, a 512GB PCIe 4.0 SSD, and Radeon 860M graphics. The 16-inch screen is available in either IPS (1920 × 1200) or OLED (2880 × 1800 120Hz) depending on the version, providing ample space for opening multiple data tables and code files simultaneously.
HP OmniBook X Flip Next Gen AI 14 offers similar specs to the 16-inch version but uses Radeon 840M graphics, a 59Wh battery, and weighs from 1.4kg. Priced between 22.5 and 28.5 million VND, it is perfect for students on the move. This is the lightest model in the group, making it convenient for carrying to class and taking direct notes on the foldable touchscreen.
Lenovo Legion 5 15IAX10 is the only option with a discrete NVIDIA GPU for those needing local CUDA, priced around 50.3–54.8 million VND. It utilizes a 20-core Core Ultra 7 255HX, upgradeable 32GB DDR5-5600 RAM, a 1TB PCIe Gen4 SSD, and either an RTX 5060 or RTX 5070 with 8GB GDDR7, paired with a 15.1-inch 2560 × 1600 165Hz OLED display. At 1.9kg, it is heavier than the other three models, but it offers the ability to train small-scale data science models directly on the device.
How much do CPU and GPU performance meet project demands?
The Core Ultra 7 165H achieves 1,759 single-core and 14,551 multi-core points in Cinebench R23. Geekbench 6 records 2,421.5 single-core and 12,230.5 multi-core points. The Intel Core Ultra 7 165H, based on the Meteor Lake-H architecture, features an 11 TOPS NPU and Arc 8 Xe-core graphics. This is well-suited for data science workflows involving Python, SQL, Power BI, and heavy multitasking, especially when paired with the 64GB RAM found on the Surface Laptop 6.
The Ryzen AI 5 340 scores 1,915.5 single-core and 12,532 multi-core in Cinebench R23, with Geekbench 6 scores of 2,781.5 and 10,765 respectively, and 19,584 in PassMark, alongside a 50 TOPS NPU on the AMD Ryzen AI 5 340 platform. In the higher tier, the Core Ultra 7 255HX reaches 32,612 multi-core in R23 and 48,542 in PassMark, providing significant headroom for big data processing and massive parallel workloads.
Regarding GPUs, the Radeon 840M on the OmniBook 14 features 256 shaders and scores 4,854 in Fire Strike; the Radeon 860M on the OmniBook 16 is more powerful with 512 shaders and 6,908 in Fire Strike. Both Intel Arc and Iris Xe do not support CUDA. Conversely, the Legion 5’s RTX 5060 has 3,328 CUDA cores and the RTX 5070 has 4,608 CUDA cores, both equipped with 8GB GDDR7. While sufficient for preprocessing and small model training, the 8GB VRAM remains a bottleneck for running large LLMs or CNNs.
Which configuration should you prioritize based on your needs?
Students who need a lightweight machine should prioritize the HP OmniBook X Flip 14. Weighing from 1.4kg, it is approximately 0.49kg lighter than the 16-inch version, making it much easier to carry to class. The foldable design and touchscreen support reading documents, note-taking, or presenting projects. While 16GB of RAM is sufficient for foundational data science studies, large datasets should be moved to the cloud to avoid memory shortages.
Deep data analysts should consider the Surface Laptop 6 15-inch. The 64GB of RAM provides a buffer for numerous notebooks, virtual machines, and heavy data tables. The 3:2 screen also displays more vertical content compared to a standard 16:9 screen. This is a top recommendation for data science students if your budget is near 55 million VND and you do not require CUDA.
If you absolutely must train data science models with CUDA locally, the Legion 5 15IAX10 is the logical choice thanks to its RTX 5060 or RTX 5070 and 20-core CPU, though the 8GB VRAM will require you to optimize your batch sizes. If you need a large display while keeping your budget under 30 million VND, the OmniBook X Flip 16 with its 16-inch screen and Radeon 860M offers a more balanced alternative.
What hardware limitations should you identify before buying?
The most critical limitation is VRAM. Integrated Arc, Iris Xe, and Radeon 840M/860M graphics are sufficient for general data science tasks but lack CUDA support; the RTX 5060 and RTX 5070 on the Legion 5 have CUDA but only 8GB of VRAM, which is insufficient for very large models. If your coursework requires heavy training, you should still budget for Colab, cloud services, or university servers. Do not choose a machine based solely on “AI PC” branding or NPU scores.
Soldered RAM is a drawback on the Surface Laptop 6 and both OmniBook models; the 16GB version may become a bottleneck when running multiple tabs, IDEs, Docker, and notebooks simultaneously. The Legion 5 solves this with upgradeable DDR5 RAM, though at the cost of weight and battery life. You should choose a capacity that will last at least three years from the start and check your curriculum’s software requirements before committing.
OmniBook screen specifications should be verified against the actual version, as IPS and OLED models differ in resolution and refresh rates. Battery life for the Surface Laptop 6, both OmniBook models, and the Legion 5 has not been independently measured. Upon receiving your device, you should inspect the physical condition, battery health, system configuration, and set a password following the guidelines on How to check password strength.
How should you check prices and warranties in Ho Chi Minh City?
The prices listed are estimates and not daily real-time inventory quotes. The Surface Laptop 6 is around 49 million VND, the OmniBook X Flip 16 is approximately 25.5–28.5 million VND, the OmniBook X Flip 14 is around 22.5–28.5 million VND, and the Legion 5 15IAX10 is roughly 50.3–54.8 million VND depending on the GPU version. You should confirm the exact GPU version and configuration that fits your data science needs before finalizing your purchase.
| Model | Est. Price | CPU | RAM | GPU / Display | Best For |
|---|---|---|---|---|---|
| Microsoft Surface Laptop 6 15″ | ~49 million | Core Ultra 7 165H | 64GB LPDDR5x (Soldered) | Arc 8 Xe-core, 15″ 2496×1664 Touch 60Hz | Large datasets, 3:2 screen |
| HP OmniBook X Flip 16 | 25.5–28.5 million | Ryzen AI 5 340 | 16GB LPDDR5x-7500 (Soldered) | Radeon 860M, 16″ IPS/OLED | Wide screen, 50 TOPS NPU |
| HP OmniBook X Flip 14 | 22.5–28.5 million | Ryzen AI 5 340 | 16GB LPDDR5x-7500 (Soldered) | Radeon 840M, 14″ IPS/OLED | Ultra-portable, 1.4kg |
| Lenovo Legion 5 15IAX10 | 50.3–54.8 million | Core Ultra 7 255HX (20-core) | 32GB DDR5-5600 (Upgradeable) | RTX 5060/5070 8GB, 15.1″ OLED 165Hz | Local CUDA, High performance |
When purchasing in Ho Chi Minh City, always ask to confirm if the machine is new or used, the exact RAM and SSD capacity, the screen type, and the warranty period. For the OmniBook, check if it is the IPS or OLED version; for the Surface, verify the RAM and configuration as it cannot be upgraded; for the Legion 5, confirm whether it has an RTX 5060 or RTX 5070. Hung Phat will provide real-time availability, warranty policies, and current promotions via the purchase links above.
To decide quickly: the Surface Laptop 6 is best for heavy dataset processing; the OmniBook X Flip 14 is ideal for students constantly on the move; the OmniBook X Flip 16 is for those needing maximum screen real estate; and the Legion 5 15IAX10 is for those who require local CUDA capability. In all cases, cross-reference your specific data science curriculum to ensure the hardware meets your required software needs before finalizing your choice.
Frequently Asked Questions (FAQ)
How much budget should I set aside for a data science laptop?
A budget of 22–29 million VND is suitable for students using Colab, SQL, and medium-sized notebooks, which covers the two OmniBook X Flip models. If you require 64GB of RAM and a 3:2 screen, your budget will increase to nearly 49 million VND for the Surface Laptop 6. If you need an NVIDIA GPU with CUDA, the Legion 5 15IAX10 falls in the 50–55 million VND range.
Should I choose Windows or macOS for AI and data science?
Windows is compatible with all four products listed and is highly convenient for many Python, SQL, and Power BI tools. While macOS can handle many academic tasks, it is not included in this specific evaluation. If your curriculum requires CUDA, the deciding factor remains NVIDIA GPU compatibility or access to a remote server.
Which model offers the best balance of portability and performance in this segment?
The HP OmniBook X Flip 14, weighing from 1.4kg and featuring the Ryzen AI 5 340 with a 50 TOPS NPU, is ideal for students who frequently carry their devices to class. The trade-off is the 16GB of soldered RAM and the Radeon 840M’s lack of CUDA support, meaning heavy training should still be offloaded to the cloud.
Which model is best if I need to run CUDA locally?
The Lenovo Legion 5 15IAX10 is the only option in this list with a discrete NVIDIA GPU, featuring an RTX 5060 or RTX 5070 with 8GB GDDR7 and a 20-core Core Ultra 7 255HX—a rare configuration for a portable data science machine. It can run CUDA for preprocessing and small models, but the 8GB VRAM limit means you may still need to supplement with a GPU server for very large models.
