Finvest
HUBS Software · CRM · SaaS · AI software · Thesis updated July 12, 2026

Growth is back, expansion is the question

01 Running thesis

New logos are carrying the story

HubSpot had a mixed Q1. Revenue growth sped up to 23% year-over-year, and large deals were strong. Deals over $120,000 in annual recurring revenue, or ARR, grew 64% year-over-year. That says bigger customers are buying the unified platform, not only small teams trying a cheap tool.

The hard part is net revenue retention, or NRR. NRR shows how much the same customers spend after upgrades, downgrades, and churn. It fell to 103% from 105% in Q4. That matters because the prior bull case leaned on expansion getting better in 2026.

The bull case is still alive. If the NRR drop is short lived, HubSpot can pair strong new customer wins with better expansion from existing customers. That would support growth above 20% for longer.

The bear case is also clearer now. New customers may be hiding slower growth inside the existing base. If NRR stays near 103% or falls again, the company may need heavier sales spending to keep growth high, which would limit margin upside.

May 2026→Q1 revenue accelerated to 23% year-over-year, helped by strong large-deal momentum. The offset was NRR falling to 103%, plus a new EU AI Act risk in the 10-Q.
Feb 2026▲Q4 2025 supported the bull case as NRR rose to 105% and full-year net new ARR grew faster than revenue. AI agent adoption also kept improving.
Nov 2025▲Q3 showed steady growth and stronger AI product adoption. The XFunnel deal added a clear answer to the risk that AI search weakens traditional SEO.
Aug 2025▲Q2 made the AI monetization plan clearer, with agents tied to a credit model. Revenue growth also improved modestly in constant currency.
Feb 2025→The initial view framed HubSpot as an AI-first customer platform. The key debate was whether AI could drive a growth re-acceleration without letting foundation models weaken HubSpot's position.
02 Business model

Seats first, AI credits next

HubSpot mainly makes money from subscriptions to its cloud customer platform. In Q1 2026, subscription revenue was 98% of total revenue. Professional services and other revenue, which includes on-boarding, training, consulting, and Payments, was the other 2%.

The core model is seat-based. Customers buy product plans, seats, and more hubs as they grow. The Smart CRM is the common data layer, and HubSpot tries to make customers more valuable by getting them to use more of the same platform.

AI adds a second money path. HubSpot embeds AI features into its hubs to make paid seats more useful. It also sells usage through credits when AI agents do work, such as answering support questions or helping sales teams prospect. That credit model is promising, but it is still early and not yet a major revenue line.

The model breaks if customers stop expanding. HubSpot can still grow by adding new customers, but lower NRR would make the growth less efficient. It also breaks if AI tools help adoption but do not turn into paid usage or higher seat upgrades.

03 Product portfolio

One platform, many entry points

Cash cow

Smart CRM

The Smart CRM is the shared customer database and context layer. It comes with product plans and helps power HubSpot's AI tools across the platform.

Growth engine

Marketing Hub and Content Hub

These tools help customers create content, run campaigns, manage websites, and reach buyers. HubSpot is also adding Answer Engine Optimization tools as AI search changes how buyers find brands.

Growth engine

Sales Hub

Sales Hub helps teams track deals, manage outreach, and use AI for prospecting. It benefits from larger customers buying more seats and more hubs.

Growth engine

Service Hub

Service Hub helps support teams manage tickets and customer help. Customer Agent is tied closely to this area because it can answer questions and resolve support work.

Steady

Operations Hub and Commerce

Operations Hub helps clean, connect, and sync customer data. Commerce and Payments add transaction tools that make the platform more useful for growing businesses.

Option

AI Agents and Breeze

HubSpot is moving from AI assistants that suggest work to AI agents that execute work. By year-end 2025, Customer Agent had been activated by over 8,000 customers, and Prospecting Agent by over 10,000 customers.

Option

Ecosystem and LLM connectors

HubSpot connects with outside tools and large language models such as ChatGPT, Claude, and Gemini. This can make HubSpot data more useful, but it also raises the risk that the main user screen shifts outside HubSpot.

04 Business segments

Mostly subscription revenue

Subscription98%growing fast
Professional services and other2%modest

HubSpot reports as one operating segment, so this mix uses Q1 2026 revenue lines from its Form 10-Q. Subscription revenue was 98% of total revenue, while professional services and other revenue was 2%.

05 Risk factors

What could break the thesis

NRR stays stuck

High impact · Medium odds

Net revenue retention fell to 103% in Q1 from 105% in Q4. That is a warning sign because expansion from existing customers was expected to improve in 2026. If the same customers do not buy more seats, hubs, or credits, HubSpot may need more new customers to hit the same growth rate.

We watchQuarterly NRR, especially whether it moves back toward 105% or slips below 103%.

AI credits do not pay off

High impact · Medium odds

AI agent adoption is growing, but the credit-based model is still young. Free or bundled AI can help sell seats, but investors need proof that paid usage becomes a real revenue stream. If customers use agents only inside free allowances, the AI story will look more like a feature cost than a profit driver.

We watchManagement disclosure on paid AI credits, agent pricing, and revenue contribution from Customer Agent, Prospecting Agent, and other agents.

Marketing playbook shifts too fast

Medium impact · Medium odds

HubSpot has said traditional search engine optimization is becoming less effective as buyers use AI answers. The company is building Answer Engine Optimization tools and bought XFunnel to help customers show up in AI answers. The risk is that the new playbook does not replace lost search traffic quickly enough for HubSpot or its customers.

We watchCustomer adoption of AEO tools, XFunnel integration updates, and HubSpot commentary on lead generation from AI search.

Platform competition gets sharper

Medium impact · High odds

HubSpot competes with point tools and larger customer platform vendors. Its LLM connectors can make the product more useful, but they can also move the user experience toward ChatGPT, Claude, or Gemini. If that happens, HubSpot's CRM data layer could become easier to replace.

We watchWin rates in larger deals, multi-hub adoption, and whether customers keep using HubSpot as the main workflow system.

AI rules add cost

Medium impact · Medium odds

The EU AI Act begins to apply, with some exceptions, as of August 2, 2026. HubSpot's platform is becoming more AI-centered, so new rules could raise compliance costs or require product changes. This risk matters because international revenue is a large part of the business.

We watchEU AI Act implementation updates, new AI compliance costs, and changes to HubSpot's AI product disclosures.
06 Quick answers

In one breath

What does HubSpot actually sell?

HubSpot sells a customer platform used by marketing, sales, service, content, operations, and commerce teams. Most revenue comes from subscriptions to its cloud software.

Why did HubSpot's Q1 2026 update look mixed?

Revenue growth sped up to 23% year-over-year, which was a strong positive. But NRR fell to 103% from 105% in Q4, which raised concern that existing customers may not be expanding as fast.

How does HubSpot make money from AI?

HubSpot uses AI in two ways. Some AI features make paid seats more valuable, while AI agents that do work can be monetized through credits.

What is the main metric to watch next?

NRR is the cleanest watch item. A rebound toward 105% would support the bull case, while more weakness would make the growth story less efficient.