First things first

If there's one thing you can count on this newsletter for, it's that I will do whatever is convenient at the moment ;-) So, I'm breaking my own rule again and writing something for the newsletter, despite my claim that I would not do that anymore.

Also, I've annotated some links below, and left others blank. Expect to see generally the same pattern of randomness in schedule and format for the next couple of weeks. Travel has that effect on something like this that actually requires a lot of focused attention.


The most interesting thing to happen in enterprise IT this week didn't occur at Dreamforce or at Microsoft Ignite, but as part of Cloudflare's 8th-anniversary festivities, when the company announced the Bandwidth Alliance along with a collection of partners. Essentially, the partners are committed to eliminating or drastically reducing egress charges for data leaving their servers and hitting the Cloudflare network. If the traffic is going to another cloud provider, Cloudflare will then route traffic to the nearest endpoint where it has a peering connection with that provider.

The members of the Bandwidth Alliance are notable: Automattic, BackBlaze, DigitalOcean, Dreamhost, IBM Cloud, Linode, Microsoft Azure, Packet, Scaleway and Vapor. I bolded some of the bigger cloud providers here, but it's also worth noting that Cloudflare has a separate agreement in place with Google Cloud as part of its CDN Interconnect program.

You can get more details on this from the blog posts linked to below, and also from next week's ARCHITECHT Show podcast (it should publish on 10/4), where I speak with Cloudflare CEO Matthew Prince and BackBlaze CEO Gleb Budman about this at length:

Why should anyone care? Because this is one of the first legitimate attempts (that I've seen, at least) to actually facilitate multi-cloud architectures at the network level (albeit for things where low latency is not a pressing issue). Technologies like Kubernetes have certainly enabled some degree of cloud portability and multi-cloud at the infrastructure and DevOps level, but anybody trying to move data from one cloud to the other still got hit with those data egress fees. And pretty much everybody agrees that those are nearly pure profit for cloud providers.

The other big thing here is how the Bandwidth Alliance could potentially open up new business for smaller providers, which can't/don't provide the all-encompassing platforms of the mega cloud-providers. Whereas it historically hasn't made a lot of economic sense for most companies, or certainly individuals, to build best-of-breed systems for processing, storage, AI, etc, that looks like a much more realistic option going forward. One could argue this is kind of the opposite of the types of announcements we've heard this week between Salesforce and AWS, or Microsoft, SAP and Adobe, where it's essentially giant vendors creating partnerships that might benefit mutual customers but don't make anything more open from an ecosystem perspective.

Of course, this is not pure altruism on the part of Cloudflare (not that I'm arguing it should be). If this sounds like an arrangement too good to pass up, you'll need to become a Cloudflare customer. And if this takes off as currently set up, Cloudflare will have positioned itself as the integral piece, meaning it's in a position to extract value in any number of ways ranging from infrastructure sourcing to investor relations.

Can AI unbundle Silicon Valley?

On a tangentially related note, I couldn't help but note that a company called Idx, which raised $33 million for technology to detect eye diseases using artificial intelligence, is based in Iowa. The booming town of Coralville, Iowa, at that! I also read about a company called Olive, which is based in Columbus, Ohio, and uses AI to automate navigation of "crappy" health care software and also correlate information across data silos.

This made me think of a discussion on this week's podcast (linked to below), as well as others I've had, about how most AI startups won't be selling horizontal AI solutions and infrastructure, but will instead be selling vertical applications powered by AI. Because nobody actually wants to be in the business of maintaining AI infrastructure and building their own models.

This topic also expands beyond AI. One thing I've noted since joining Pivotal (disclaimer!) and even more so at our SpringOne Platform conference this week, is how many companies (many of which most people have never heard of) around the United States and the globe are doing really cool things and catapulting themselves into the 21st century thanks to advances in cloud-native architectures and development principles.

And these startups -- actually, they don't even need to be startups -- don't need to be located in Silicon Valley (or New York or Seattle) at all. They can be located anywhere it makes sense given their industries and talent base, which might be in Iowa (where there are a ton of universities), Minneapolis, Houston or wherever.

In fact, being located outside of Silicon Valley could actually be a benefit. Established companies implementing AI/containers/whatever into their products aren't going to move their headquarters, and it's likely a boon to be located near the people and industries the company is serving (e.g., agriculture, energy, automotive, etc.). Furthermore, there are a lot of people looking to get out of Silicon Valley for any number of reasons relating to cost of living, transportation, ideology, you name it. Moving somewhere closer to their homes, or at least to some semblance of sanity, while still getting to work on interesting technology could be a big draw.

The cliche is that all companies are becoming technology companies, and it's true to a large degree. But not all companies are going to be based in Silicon Valley. Regular readers might have noted that I have a lot of mixed feelings about AI and automation in the grand scheme of things, but I am heartened by the notion of software-based businesses spreading around the world and becoming imbued with some of those regional values. And by the knowledge that as programming abstractions advance, jobs in these companies should open up to more people.

That's not to say it's a 1:1 replacement for everything that has been and still might be lost, but spreading the technological wealth beyond a small part of California should ultimately be a good thing.

Read and share this issue online here.

The ARCHITECHT Show podcast

Cloud and infrastructure












Data and analytics