Happy Friday. Cloudflare just delivered its fastest revenue growth in nearly four years, and the numbers are hard to ignore: 39% year-over-year, $782 million in a single quarter, and GAAP profitability now expected before the end of this year instead of 2028.
The other two stories today are about where the next wave of AI infrastructure spending lands. Dropbox paid $350 million to get into object storage for AI workloads, and AWS formally retired Amazon Q Developer in favor of Kiro bots that run scheduled, persistent background agents inside a developer's workflow.
ENTERPRISE SOFTWARE
Cloudflare Hits $3B Run Rate, Pulls GAAP Profit Target Forward by Two Years
Cloudflare added more than 158,000 net new paying customers in Q3 2026, a 110% increase year-over-year, while dollar-based net retention climbed to 122%. The Workers Developer Platform now counts more than 10 million active developers, and CEO Matthew Prince credited agentic workloads as a primary driver of the outperformance.
The gross margin dip to 72.4%, down 290 basis points, is worth watching. Cloudflare attributed it to the lower-margin Workers platform mix, which means the fastest-growing part of the business is also the least profitable per dollar of revenue. If agentic traffic scales as fast as the company expects, that margin pressure is not going away. The first revenue from Act 4 products, which target agent traffic management and content micropayments, is early but signals Cloudflare is building monetization specifically for AI workloads. That is the number to track in Q4.
DEVELOPER TOOLS
AWS Kills Amazon Q Developer, Bets Everything on Kiro's Persistent Agent Bots
AWS made Kiro bots generally available on November 5, completing a consolidation that formally ends Amazon Q Developer. New Q Developer signups were blocked in May, and full end-of-support lands April 30, 2027. Migration is one-way: there is no path back to a Q Developer Pro subscription.
The coordination model is the real variable for engineering leaders evaluating adoption. Bots run parallel worker sessions in the background on scheduled intervals, which means token consumption and compute costs can compound fast without close monitoring. IDC's Dave McCarthy named this the coordination tax problem, and it is not theoretical. Any team running overnight CI reviews or automated release notes should cost-model usage carefully before moving to Pro or team plans. The Notes memory system also raises audit questions: tracing why a bot made a specific decision weeks ago, across multiple sessions, may require more discipline than most teams apply today.
FUNDING & M&A
Dropbox's $350M Tigris Buy Is a Direct Move Against Amazon S3
Dropbox agreed on Nov. 3 to pay approximately $350 million in cash for Tigris Data, a startup founded by former Uber storage engineers. The price is 14 times Tigris's October 2025 Series A. The deal seeds DBX Agent Cloud, an S3-compatible object storage product that eliminates egress fees and automatically moves data closer to the GPUs accessing it.
Dropbox's core business grew 0.5% year-over-year in Q3. That number explains why the company is writing a check equal to 31% of its June cash balance. The Tigris acquisition is a revenue diversification play, not a product extension. Eliminating egress fees is a direct challenge to Amazon S3's pricing model, and the customer list, Railway, Runpod, and Replicate, shows Tigris was already embedded with AI compute providers. The harder question is whether Dropbox can sell infrastructure to developers who have never thought of it as a storage vendor.
Created by the robots at The SaaS Sentinel