About this video
The era of cloud-dependent AI is ending as personal hardware catches up. In this video, I break down my transition to the M5 Max MacBook Pro with 128GB of RAM and why local compute is the final nail in the coffin for many SaaS models. Key Takeaways: - Why 128GB of RAM is the essential threshold for professional local LLM performance. - The reality of M1 vs M5: Speed gains for traditional tasks are surprisingly marginal. - Why the 'SaaS is dead' narrative is driven by local-first AI workflows. - The hidden costs and thermal challenges of the 14-inch M5 Max under heavy load. - Why 'Personal Software' is the next big shift for founders and CEOs.
The Age of Personal Software: Why I Bet My Workflow on 128GB of Local RAM
Stop renting your intelligence from Silicon Valley. The era of the SaaS-dependent founder is reaching its natural conclusion, and the catalyst isn't a new cloud subscription, it is the hardware sitting on your desk. I recently upgraded to the M5 Max MacBook Pro with 128GB of RAM, and whilst the price tag of over £5,000 is enough to make any CFO sweat, the implications for personal software are profound.
The Marginal Gains of Traditional Compute
For the average CEO or founder, the difference between an M1 and an M5 in daily tasks is almost invisible. If you are just switching tabs, editing the occasional video, or managing a team in Slack, the speed increases are marginal. In my testing, rendering times in Final Cut Pro only improved by roughly 10 to 20 percent. If you are upgrading for the 'snappiness' of the OS, you are likely wasting your capital.
Why 128GB is the New Minimum
The real shift happens when we discuss Local Large Language Models (LLMs). The 'SaaS is dead' narrative gains traction when you realise that the most expensive part of your AI workflow can be brought entirely in-house.
On my previous 64GB machine, complex codebase analysis would take minutes of painful waiting. On the M5 Max with 128GB of unified memory, those same tasks take seconds. This is not just a speed upgrade; it is a fundamental shift in how we interact with software. When you own the compute, you own the privacy, the uptime, and the destiny of your data.
The Local-First Future
Apple has effectively built an AI workstation disguised as a consumer laptop. The 128GB of unified memory is the 'sweet spot' for running high-quality quantised models that rival GPT-4 in specific tasks.
As founders, we must ask: why are we paying monthly seats for features that our laptops can now run natively while we sleep? The transition to personal software means our tools will soon be as bespoke as our strategies, running locally, tuned to our specific needs, without a single API call to a third party.
Final Verdict
If your day consists of emails and spreadsheets, stay with your M1. But if you are building the future of local-first AI and you want to lead the charge away from SaaS dependency, the M5 Max is the first true tool for the age of personal software. Just be prepared for the fans to kick in when the thinking gets heavy.
Transcript▾
I have been teasing this purchase for long enough: the M5 Max MacBook Pro with 128GB of RAM. I was quite lucky with the timing, as Apple’s prices have since soared. I previously used an M1 Max with 64GB, which was top-tier at the time, but I wanted to see if the M5 Max was a genuine upgrade for a creative and technical workflow.
I opted for the 14-inch model for portability. It features a 40-core GPU, nano-texture display, and 2TB of storage. The current price for this spec is nearly £7,000, which is staggering. My daily work involves video editing, coding with tools like Warp and Claude, and occasional Docker usage. Interestingly, I found that for standard video editing and day-to-day multitasking, the performance gains over the M1 Max were subtle rather than revolutionary.
The true differentiator is local AI. On my M1, local LLM tasks were slow and cumbersome. On the M5 Max, the speed difference is incredible; what once took minutes now takes seconds. This machine is built for local inference. While the chip itself is fast, the real bottleneck for AI has always been RAM.
However, it is not all perfect. The 14-inch chassis struggles more with heat than the 16-inch, and the fans can be quite loud under load. Furthermore, rumours of a MacBook 'Ultra' suggest even more RAM might be on the horizon. For now, unless you are heavily invested in local AI, the M1 Max remains a formidable machine. But for the future of personal, local-first software, the M5 Max is in a league of its own.