A $7,000 AI Laptop. What Exactly Is Revolutionary?

A $7,000 AI Laptop. What Exactly Is Revolutionary?

I have worked with computers since the MS-DOS era. I respect Jensen Huang and what NVIDIA and Microsoft have built. That is precisely why their new RTX Spark laptops disappoint me.

Let's start with the price.

Microsoft's Surface Laptop Ultra starts at $2,599 with 24GB of unified memory. The 128GB version reviewed at launch costs $5,899 and includes a 1TB SSD. Best Buy advertises a 128GB ASUS ProArt P16 for $6,999.

And this is being sold as a new era of personal computing.

What is the breakthrough? Fitting a large language model into memory? Up to one petaflop of FP4 arithmetic? Neither tells me whether a demanding job gets finished.

Apple is hardly standing still. The M5 Max supports up to 128GB of unified memory and 614GB/s of bandwidth, versus RTX Spark's 300GB/s. Bandwidth is not a complete benchmark, and CUDA remains a major NVIDIA advantage. But early reviews also find the Surface behind MacBook Pro in CPU performance and battery life.

Then there is the software.

According to NVIDIA's own launch announcement, native Photoshop and Premiere support arrives in beta. Substance comes later.

This is not a garage startup asking enthusiasts to finance its first experiment. These are Microsoft and NVIDIA. At professional prices, beta software is an awkward centerpiece for a supposedly ready professional computer.

I own a DGX Spark. My AI systems run local models. Yet for demanding software engineering, they often turn to Codex: they set tasks, assess results and continue their work.

That is not a failure. It is sensible division of labour.

I can direct serious development from my phone. I do not need to carry the entire compute infrastructure to own it.

A local 70B model is a useful capability. It is not automatically a senior engineer, a mature personal AI environment or a substitute for stronger external tools. Future models will improve. But tomorrow's intelligence should not be included in today's sales pitch.

I want local AI. I want NVIDIA, Apple, AMD and Microsoft competing fiercely.

I also want a different competition: verified work completed, software that runs, genuine autonomy, total cost of ownership and the freedom to choose the best tool.

When a laptop costs more than many Europeans earn in a month, the buyer deserves a finished product, not a promise with a premium display.

A real personal AI revolution would make capability more accessible.

The invoice is certainly ready. Is the revolution?