The Money Moved: Why AI Investors Are Betting on “Picks and Shovels”

AI Agents

The Most Important AI Story This Week Isn’t a Model — It’s Where Agents Actually Run

Here’s the signal hiding in this week’s funding news: the next phase of AI won’t be decided by whose model answers best, but by whether an agent can safely carry out work. Daytona, a startup building cloud infrastructure purpose-built for AI agents, just raised a $60 million Series B led by FirstMark Growth at a reported $1 billion valuation — bringing its total to $91 million.

The pitch is worth your attention because it names a real problem. Today’s cloud was designed for long-lived services that sit there waiting for a request. AI agents are the opposite: they spin up thousands of short, stateful jobs nobody planned in advance. An agent needs to open files, run software, keep its context, use a specific operating system, and finish a task over time — all without sharing a messy environment with other agents. Daytona’s answer is “every agent its own computer”: sandboxed Linux, Windows, and macOS boxes with GPU access on tap. The workspace is built around the agent, rather than bolting the agent into an existing app.

Why it matters: We’re watching the “picks and shovels” trade play out in real time. The people who got rich in the gold rush were rarely the miners. As businesses move from testing agents to running them in production, the bottleneck shifts from “can the model reason” to “can we give it a safe, isolated place to work.” Investor money — and the list of backers here includes Stripe’s Patrick Collison, Shopify’s Tobi Lütke, and Perplexity’s Aravind Srinivas — is now chasing that second question.

What’s next: Watch for the same shift in how you evaluate your own agent stack. The question is no longer just “which model,” but “where does it run, and can I isolate and audit what it does.” If you’re building agents, assume the infrastructure layer — not the model layer — is where the next wave of differentiation happens.

Source: WeRSM

Oracle Puts $1B Behind Partners Building AI Agents

Oracle committed $1 billion over four years to its partner ecosystem, explicitly to help partners build “specialized applications, integrations and AI agents” on Oracle Cloud. It’s the enterprise mirror of the same bet: the winners at this layer will be whoever can help customers deliver agents, not just host models. Source: The AI Insider

Fireworks AI Eyes a $40B Valuation — Right After a Security Incident

Fireworks AI is in early talks that could lift its valuation to $30–40 billion, nearly double its July figure — driven by $1B+ in annualized revenue and 40 trillion tokens processed daily. Notably, the talks come weeks after it disclosed a credential breach exposing some customer data, a reminder that infrastructure trust is now a deal-making variable. Source: Jawlah / Business Insider

AI News

DeepSeek Raises $15B to Keep Pace With the US Giants

China’s DeepSeek raised roughly 100 billion yuan ($15 billion), backed by Tencent and CATL, at a valuation near 500 billion yuan — with its fundraising target repeatedly raised as demand swelled. It’s the clearest sign yet that the model race has become a capital race, with OpenAI and Anthropic each already past $100 billion. Source: The Daily Star / Reuters

The AI Graveyard Grows — and 42% of Corporate Projects Die Quietly

Character.ai filed for bankruptcy (Disney is the opening bidder at $20M), while TechCrunch’s running “AI graveyard” now spans Humane’s $230M pin to Sora and Recall. The sobering stat: S&P Global says roughly 42% of AI initiatives are abandoned by their parents — a useful caution against mistaking funding headlines for durable products. Source: TechCrunch

AI Is Now a Factor in Nigeria’s 2027 Election

At the AI and Democracy Forum in Abuja, officials acknowledged AI can “distort the choice” ahead of Nigeria’s 2027 vote, while insisting the final decision stays with voters. It’s an early signal of how AI-generated misinformation is becoming a formal election-integrity concern in the world’s largest democracy-by-population. Source: TheCable