Posts

“We processed over 100t tokens this quarter, up 5x year over year, including a record 50t tokens last month alone.”

If the market harbored any doubt for the insatiable demand for AI, this statement during Microsoft’s quarterly earnings yesterday, quashed it.

What could this mean for a run rate? Using some basic assumptions¹, this implies :

So AI is roughly between 2 to 22% of Azure revenue. Error bars here are quite large, though.

By 2026, AI agents will consume 10x more enterprise data than humans, but with none of the contextual understanding that prevents catastrophic misinterpretations.

In this presentation I shared yesterday, this is the main argument.

Historically, our data pipelines have served people. We’ve architected complex pipelines to ingest, filter, and transform information in different systems of record: cloud data warehouses, security information and event management systems (SIEMs), and observability platforms.

We then interpreted these outputs and acted upon them.

If every employee starts managing agents, how does a company change?

First, “83% of global leaders say AI will let employees take on more complex, strategic work earlier in their careers.”

One executive recently framed this transition of teams evolving to three areas of work : operational, tactical, & strategic. Operational work can be mostly fully automated today. Agents are chomping away at tactical work - better accuracy will improve their share. Humans will focus on strategy work, but likely assisted by AI.

The ServiceNow team is raising the bar by innovating on our own platform, taking an end-to-end agentic AI-first approach to running our business. We see this in a 16x improvement from lead-to-sale conversion & an over 86% deflection of the soul-crushing work people used to do themselves.

ServiceNow reported earnings yesterday alongside Google & many other companies. All the insights out of those reports were interesting, but this was the one that stood out the most to me.

When you buy a stock or a bond, do you know which database that transaction is running on?

I’m sure the answer is no, but the database will change & that has far-reaching implications.

data via Allium

Blackrock’s tokenized Treasury fund, BUIDL, has amassed $2b of treasuries so far, with a recent significant surge. This fund is still a bit smaller than BlackRock’s classic Treasury Fund BYTXX at $21b.

Two years ago I published an image of a rabbit on a fire truck to demonstrate how effective image generation was for blog posts.

The conclusion at the time : the images were unusable.

In 2025, the opposite is true - with one caveat.

Prompt: Coupa & Thoma Bravo logos each in front of a cloud with a plus sign between them.

2023	2025

Two years has brought tremendous professionalism to new image. Corporate street art is now pixel-perfect professional. The clouds and the + sign are perfect, and the logos look the part.

Software systems work best when they’re connected to each other. For years, incumbents use deep integrations as a competitive moat.

But AI upends this dynamic.

A few of our portfolio companies are starting to develop integrations with AI in a matter of hours, completely upending the two or three quarter timeframes of classic enterprise integration development.

This enables two important impacts to the sales cycle.

First, the integration a customer desires can be built during the sales cycle, demonstrating the startup’s technical agility.

I want to see how you use AI.

The biggest challenge to AI adoption is reimagining workflows, AI is best discovered socially.

Midjourney’s launch on Discord was brilliant. As a new user, I knew enough to ask for an image of a rabbit on a firetruck. But by watching others, I discovered I could create vector images, recast photos into Ghibli-style art, change aspect ratios, or convert photos to pencil-shaded drawings.

We don’t often hear the words speed & ERP in the same sentence. But when researching Doss, we heard it again & again. We’ve always believed fast sales cycles create competitive advantage - nowhere is this more true than enterprise software.

Imagine purchasing software knowing there’s a 75% chance it won’t deliver value after a 12-month deployment. If selling software is the business of selling promotions, traditional ERP offers a ponderous path to career advancement.

I’ve been watching my behavior evolve as I use AI more.

Coding new projects, I started first by describing a big idea : generate a new website that ingests podcasts and transcribes them. But I quickly discovered this approach was like asking someone to build a skyscraper without blueprints—the AI choked on requests too broad & lacking structure.

So I broke the project down into smaller blocks. Write a function to download the list of podcasts in the last week from these sites. But I noticed how low-leverage that felt : I knew I was yoking a strong animal to pull along a child’s wagon.