Three separate funding rounds closed in the past month to build faster, cheaper AI inference infrastructure. In the same window, Anthropic told users of its coding tool Claude Code that it would stop asking permission before most actions. Read together, they describe where AI spending and AI usage are actually heading this year: toward systems that run more, and are checked less.
Inference funding keeps climbing
AI chip startup Etched closed a $300 million Series C on July 23, led by Sequoia Capital with participation from Andreessen Horowitz, Jane Street, Diffusion and SK Hynix. The round values the company at $10.3 billion, up from $5 billion in December, and Etched says it already holds $1 billion in chip pre-orders (TechCrunch, 2026). The company builds transformer-specific inference chips rather than general-purpose GPUs.
Nine days earlier, inference platform Fireworks AI announced a $1.505 billion Series D at a $17.5 billion valuation, led by Atreides Management, Index Ventures and TCV. Fireworks says it now serves more than 40 trillion tokens a day, up from 15 trillion nine months prior, and has crossed $1 billion in annualized revenue. Named customers include Uber, Shopify, Doximity, Elastic, GitLab and MongoDB, along with Cursor and Harvey (Fireworks AI, 2026; Businesswire, 2026).
On August 3, London-based OLIX raised $312 million in Series B funding at a $3.3 billion valuation, backed in part by the UK government's Sovereign AI venture fund, Arm, Hudson River Trading and Netflix co-founder Reed Hastings. OLIX builds photonic inference chips and is targeting delivery of its first processor, the DX-1 decode accelerator, to customers in the second half of 2027 (Data Center Dynamics, 2026; Cooley, 2026).
Three different approaches, chip design, hosted inference, and photonic hardware, all raising large rounds inside a single month. The common bet is the same: the cost of running AI models in production, not training new ones, is where the next multi-billion-dollar market sits.
Anthropic removes a step from the approval chain
On August 9, Anthropic said it would make Claude Code's "auto mode" the default setting for Pro, Max and Team accounts starting August 14, with Enterprise, API and cloud-platform users following within the following month (TechCrunch, 2026; Anthropic, 2026). Auto mode lets the coding agent proceed through a task without asking for approval at each step, unless the action is destructive or difficult to reverse.
Anthropic's own engineering post cites internal testing with 1,053 paid users: auto mode caught 89 percent of harmful actions before they happened, compared with 13.6 percent caught by humans reviewing the same prompts manually. The company also reported that users approve 97 percent of permission prompts regardless, calling manual review a habit that had stopped functioning as a real check (Anthropic, 2026). Users who have already set a different preference get a one-time prompt asking whether to switch; those who pinned a setting keep it.
The Register and Help Net Security both flagged the same tension in their coverage: removing a human step by default, even with a stated 89 percent catch rate, shifts risk onto the agent's own judgment for the 11 percent it misses (The Register, 2026; Help Net Security, 2026).
Why the two trends point at each other
The infrastructure being funded exists to serve exactly the kind of usage Anthropic is now defaulting to: agents that act continuously rather than pausing for a person after every step. Fewer approval pauses means more inference calls per task, which is the demand Fireworks, Etched and OLIX are all pricing their rounds against. None of the three funding announcements mention Claude Code specifically, and this is not a claim that they are connected deals. It is a claim that the capital and the product decision are responding to the same underlying trend: AI systems doing more per human interaction, not less.
For engineering leaders evaluating coding agents, the practical question raised by Anthropic's own numbers is not whether auto mode catches more than manual review on average. It is whether the 11 percent of harmful actions it misses fall inside a team's tolerance for an unreviewed production change.
What to watch next
Anthropic's rollout to Enterprise, API and cloud-platform accounts is due within a month of the August 14 consumer and Team default, which will be the first real test of whether large engineering organizations accept the same default their individual developers now have. Whether inference demand actually grows fast enough to justify three separate billion-dollar-plus valuations built on it will show up first in Fireworks' and Together AI's next disclosed usage numbers, expected with their next funding updates or earnings disclosures.
Sources
- TechCrunch, "Anthropic is turning Claude Code's auto mode on by default" - https://techcrunch.com/2026/08/09/anthropic-is-turning-claude-codes-auto-mode-on-by-default/
- Anthropic, "Claude Code auto mode" (Engineering blog) - https://anthropic.com/engineering/claude-code-auto-mode
- The Register, "Claude Code puts auto mode in the driver's seat" - https://www.theregister.com/ai-and-ml/2026/08/10/claude-code-puts-auto-mode-in-the-drivers-seat/5285326
- Help Net Security, "Anthropic to put AI in charge of reviewing Claude Code actions by default" - https://www.helpnetsecurity.com/2026/08/10/anthropic-claude-code-auto-mode/
- TechCrunch, "AI chip startup Etched defies skeptics, hits $10.3B valuation from big-name investors" - https://techcrunch.com/2026/07/23/ai-chip-startup-etched-defies-skeptics-hits-10-3b-valuation-from-big-name-investors/
- Fireworks AI, "Fireworks Raises a $1.5 Billion Series D" - https://fireworks.ai/blog/series-d-announcement
- Businesswire, "Fireworks Raises a $1.5 Billion Series D to Lead the Specialized Intelligence Revolution" - https://www.businesswire.com/news/home/20260716264405/en/Fireworks-Raises-a-$1.5-Billion-Series-D-to-Lead-the-Specialized-Intelligence-Revolution
- Data Center Dynamics, "Chip startup Olix raises $312m at $3.3bn valuation, backed by UK gov't Sovereign AI venture fund" - https://www.datacenterdynamics.com/en/news/chip-startup-olix-raises-312m-at-33bn-valuation-backed-by-uk-govt-sovereign-ai-venture-fund/
- Cooley, "OLIX Raises $312 Million at a $3.3 Billion Valuation" - https://www.cooley.com/news/coverage/2026/2026-08-03-olix-raises-$312-million-at-a-$3-3-billion-valuation