In a single week, investors poured billions into the software layer that lets AI agents actually do work — Baseten alone raised a $1.5 billion Series F at a $13 billion valuation — while Gartner warned that those same agents could siphon $234 billion away from the traditional SaaS vendors they run on top of. The message from the last seven days is hard to miss: the money and the risk are now flowing to the same place.
Key takeaways
- Baseten raised a $1.5B Series F at up to a $13B valuation, underscoring how inference has become AI's most contested infrastructure layer.
- Qualcomm agreed to acquire AI software startup Modular for ~$3.9B in stock to deepen its data-center software stack.
- Gartner now projects AI agent software spending will hit $206.5B in 2026, up roughly 139% year over year.
- A separate Gartner report warns agentic AI puts $234B of enterprise SaaS spend at risk by 2030 through “agentic arbitrage.”
- Governance is the new frontier: Runlayer landed a $30M Series A to police what AI agents are allowed to do.
AI Agents Just Had a Billion-Dollar Week
If you only glanced at the headlines, you might have missed how tightly this week's news clustered around one theme. Nearly every major funding round, acquisition, and analyst report pointed at the same shift: enterprises are no longer buying software that answers questions. They're buying software that takes action.
That distinction matters. Traditional SaaS charges per seat and lives inside a dashboard someone has to click through. AI agents skip the dashboard entirely — they read the data, make the decision, and execute across several systems at once. The companies raising money this week are building the plumbing that makes that possible.
The Funding: Where the Money Went
The standout was Baseten, which announced a $1.5 billion Series F on June 22 led by Altimeter Capital, Conviction, and Spark Capital, with the round structured across tranches at $13 billion and $11 billion valuations. The company now processes more than a billion inference calls a day, and revenue has grown roughly 20x year over year — a sign of how quickly demand for running models in production has scaled.
The week's biggest acquisition came from Qualcomm, which agreed to buy AI software startup Modular for about $3.9 billion in an all-stock deal announced June 24. Modular's software lets AI models run across CPUs, GPUs, NPUs, and custom chips without rewriting code — exactly the kind of hardware-agnostic layer Qualcomm needs as it pushes into data centers.
Smaller but telling: Runlayer closed a $30 million Series A led by Felicis for AI agent governance, and Italian startup Seltz raised $12.5 million to build a web-knowledge API for agents. The pattern is clear — capital is chasing the infrastructure, safety, and data layers that sit underneath the agents themselves.
| Company | Deal | Amount | Focus |
|---|---|---|---|
| Baseten | Series F | $1.5B | AI inference infrastructure |
| Qualcomm / Modular | Acquisition | ~$3.9B | Data-center AI software |
| Runlayer | Series A | $30M | AI agent governance |
| Seltz | Seed | $12.5M | Web-knowledge API for agents |
AI Agents vs. the SaaS Business Model
Here's the twist that made this week interesting. On July 1, Gartner published a report warning that agentic AI puts up to $234 billion of enterprise application software spend at risk by 2030 — roughly 20% of enterprise SaaS spending. The mechanism they describe is “agentic arbitrage”: when an AI agent completes a task by stitching together several systems, users stop logging into the individual apps.
That breaks the core assumption of the SaaS model. For two decades, software vendors have sold seats, and revenue rose as more people logged in. If agents do the work instead of humans, seat-based pricing starts to wobble — the software becomes invisible infrastructure rather than a destination.
Agentic systems deliver outcomes directly, making the software invisible and breaking the link between user growth and revenue growth.
The same report frames it as an opportunity, too. AI-native startups that act as the agentic layer across a company's tools can capture not just existing budgets but the incremental spend unlocked when agents actually deliver measurable results. In other words, the $234 billion doesn't vanish — it moves.
Why AI Agents Need Governance Now
The Runlayer round is a small deal with an outsized signal. As soon as agents can act on their own — moving money, editing records, sending messages — someone has to decide what they're allowed to touch. Governance, permissions, and audit trails are quickly becoming a product category of their own, much the way identity management became essential once cloud apps multiplied.
The parallel to early SaaS
It rhymes with the last platform shift. When SaaS exploded, a whole ecosystem of security, single sign-on, and compliance tooling grew up around it. AI agents are triggering the same wave, just faster. Expect “agent governance” to appear on enterprise shortlists within the next few budget cycles.
What AI Agents Mean for Small Businesses
You don't need a $1.5 billion war chest to benefit from this shift. The same agent patterns big enterprises are buying can be assembled today from affordable, no-code tools. Automation platforms like Make.com and Pipedream let you wire an AI model to your existing apps so routine tasks run without a human clicking through them.
On the workspace side, agent-friendly tools such as Notion, ClickUp, Coda, and Taskade now embed AI that can draft, summarize, and act on your data in place. For content, tools like Writesonic point the same way — software that produces the outcome instead of just giving you a blank editor.
The practical takeaway: start small. Pick one repetitive workflow, hand it to an agent, and measure the hours you get back before scaling up.
Frequently asked questions
What exactly are AI agents?
AI agents are software systems that don't just answer questions — they take action. An agent can read data, make a decision, and carry out multi-step tasks across several apps on its own, rather than waiting for a person to click through each screen.
Why is Gartner warning about $234 billion in SaaS spend?
Because agents can complete work by combining several tools at once, users may stop logging into individual apps. That undermines seat-based pricing, the model most SaaS companies rely on, putting an estimated 20% of enterprise SaaS spending at risk by 2030.
Can a small business use AI agents affordably?
Yes. No-code automation and AI-enabled workspace tools make it possible to build simple agent workflows for a modest monthly cost. The best approach is to automate one clear, repetitive task first, then expand once you see the time savings.
The bottom line
This week showed both sides of the same coin: record money flowing into the infrastructure that powers AI agents, and a credible warning that those agents will reshape how software gets sold. For investors, the opportunity is the plumbing. For businesses of every size, the smart move is to start experimenting now — before the software you rely on quietly turns into something an agent uses on your behalf.
Sources: Gartner (agentic AI SaaS risk), Business Wire (Baseten), CNBC (Qualcomm / Modular).