Journal / July 2, 2026

Gartner just put a $234 billion price on the SaaS everyone stops…

On July 1 Gartner attached a number to the fear every software vendor has been dodging. About $234 billion of enterprise application spend is exposed by 2030…

July 2, 20264 signals

Direct answer

What changed in AI on July 2, 2026?

On July 1 Gartner attached a number to the fear every software vendor has been dodging. About $234 billion of enterprise application spend is exposed by 2030…

Published by Bles SoftwareUpdated July 2, 20264 linked signalsHow we build the roundup

Signal digest

The sources worth opening

Gartner just put a $234 billion price on the SaaS everyone stops opening

On July 1 Gartner attached a number to the fear every software vendor has been dodging. About $234 billion of enterprise application spend is exposed by 2030 to what it calls agentic arbitrage, the moment an agent finishes a job across five tools and nobody logs into any of them. When the outcome arrives without the interface, the old link between seat count and revenue snaps. For an operator the opening is exactly as large as the risk. If you can sit as the agent layer across a company's tools and deliver a measured result instead of another dashboard, that budget is loose and looking for somewhere to land.

Source: Original source

Money is pouring into agent startups at prices that skip the usual proof

The last few days brought a wall of capital aimed at agents. Unconventional AI, two months old and led by former Databricks AI head Naveen Rao, took $475 million in seed at a $4.5 billion valuation. Legora raised $550 million at $5.55 billion, and Harness pulled $240 million to automate the after-code phase. The pattern is investors paying frontier prices for teams with almost no track record, betting the agent layer gets built fast. You do not need to chase those rounds. The signal is that buyers now expect agent-shaped products, so a working one is far easier to sell than it was six months ago.

Source: Original source

The market moved from spending big on tokens to spending well

A quiet reversal is showing up in real bills. Teams that spent the last year maxing out tokens are now optimizing for cost, and the swings are brutal. The CEO of Lindy moved all of his traffic off Claude to DeepSeek's cheaper open-weight models and watched his cost curve collapse. OpenAI itself said the ceiling on enterprise value is no longer model capability, it is picking the right use case, redesigning the workflow, and managing the change. For operators that is permission to stop obsessing over which model is smartest. The money is in the plumbing around the model, not the model.

Source: Original source

Google and Microsoft open their wallets to catch Claude Code

The coding agent race turned into a price war. Google and Microsoft are using their cloud balance sheets to pull developers away from Anthropic's Claude Code and OpenAI's Codex, with Google pitching a $100 a month developer tier and Sundar Pichai openly admitting the company is behind on agentic coding. Meanwhile Claude reached 56 million monthly users, up roughly 640% year over year. For anyone who builds software the takeaway is cheerful. Four giants fighting for your coding hours means the tools get better and cheaper on their dime, so commit to nothing long term and keep testing what actually ships your work fastest.

Source: Original source

← 2026-07-01 2026-07-03 →