This week, I’m open sourcing a technique I’ve been using quietly for the last quarter to track AWS’s AI infrastructure buildout.
The idea is simple: AWS updates their regional services page daily. Every time they launch a new product, expand to a new region, or roll out AI services like Bedrock to new servers… it shows up on their website.
But here’s the problem:
There’s no historical log. No changelog. No press release every time a service goes live in a new geography.
If you wanted to track this manually, you’d need to check daily and keep your own records. That’s dozens of regions and hundreds of services. Nobody has time for that.
This is where AI automation comes in.
I’m going to show you how I use Manus to schedule a daily web scrape of AWS’s regional services data. The system:
- Scrapes AWS’s regional services page every morning at 8 AM ET
- Compares against the prior day’s data
- Flags any new regions, new services, or removals
- Generates an investment interpretation layer for each change
Why does this matter?
New AI service launches (especially Bedrock) are a proxy for GPU capacity buildout and customer demand. New region expansions signal incremental TAM and infrastructure investment. These are signals that don’t hit the P&L for 2-4 quarters… but if you’re a fundamental investor, this is how you get ahead of management commentary and street consensus.
Pod shops pay six figures for alternative data feeds like this. You can now build it yourself without writing a single line of code.
Let me show you how.
The Prompt
I like to run this in Manus. They combine (1) excel output format (2) low error rates on scraping (3) capability to schedule daily tasks so I can automate this research
As always, I never take paid promotions - we try to be as credibly neutral as possible here. I only write about tools I personally use in my investing and research stack.
I just find Manus the easiest to use for this task.
Prompt: AWS_Bedrock_Scraping_Prompt_Dave_Wang.pdf
The Result
Full output here: Link
Excel output: aws_region_services (1).xlsx
Our automated system produces a multi-sheet Excel report every morning at 8 AM ET.
Here's what makes this technique powerful.
You now have a proprietary dataset that doesn't exist anywhere else.
AWS doesn't publish a changelog. There's no press release when Bedrock goes live in ap-southeast-3. No analyst is tracking this systematically. But now you are.
Over time, you're building a historical record of AWS infrastructure buildout that you can map against earnings, guidance, and stock price. When management says "we're seeing strong AI demand" on the call... you'll have the receipts.
This is a leading indicator, not a lagging one.
New service launches and regional expansions are forward-looking signals. They reflect decisions AWS made 6-12 months ago based on customer demand forecasts. By the time this shows up in revenue, you're already 2-4 quarters behind.
If you see Bedrock light up in 5 new regions over a 2-week span, that's a proxy for:
- Incremental GPU capacity coming online
- Enterprise AI workload demand accelerating
- CapEx deployment translating to future revenue
This is the kind of signal that doesn't hit the P&L immediately... but if you're building a 12-month view on AMZN, NVDA, or AMD, this is exactly what you want.
The real edge is automation + accumulation.
Any analyst can check AWS's website once.
The edge comes from doing it every single day, automatically, and building a time series nobody else has. After 3 months, you'll have 90 data points. After a year, you'll be able to see seasonality, acceleration, and deceleration patterns that are invisible to everyone else.
Personal
My companies I've been building for corporate workshops, advisory, and online education have been growing fast this last quarter.
One challenge I've had has been scaling internal operations - I've been spending a lot of time finding the right folks, onboarding, building SOPs / systems, etc.
This is one skill I am working to improve .... and I'm sure many folks reading this have gone through this before.
If you've scaled teams fast, I'd love to hear your advice! Hit the reply button - I'm all ears and always looking to learn and improve.