Infra2026-07-24 · source-backed

Etched hit a $10.3B valuation for GPU-free transformer inference.

Issue

2026-07-24

Confidence

source-backed

Sources

Redaction

passed

Story

TechCrunch reports the three-Harvard-dropout company raised at that mark on claims its silicon accelerates inference on any model without GPUs. Skeptical read: Etched has been promising silicon for years and a valuation isn't a shipped benchmark. Investor conviction at this size is itself data, but I'd wait for a third-party throughput number.

Kog claims 30x faster LLM inference on existing GPUs through low-level reverse engineering.
The French startup argues the belief that GPUs suit agentic workflows poorly is a misconception, and says its Kog Inference Engine reaches up to 30x decoding speedups on standard NVIDIA and AMD datacenter GPUs with no new hardware. The approach is hardware-aware optimization o...
Fish Audio raised $52M at $21M ARR one year in, off an open-source repo with 31K stars.
TechCrunch reports the seed led by Coreline Ventures and Capital Today, with 8 million users at the one-year mark. The origin is the pattern: co-founder Shijia Liao, a former NVIDIA video researcher, trained the initial models on a single gaming GPU, and Fish Speech passed 31,...
AWS put JupyterLab and VS Code inside EKS clusters, claiming 5 minutes vs 3–5 days of platform work.
The SageMaker AI Spaces add-on runs managed notebook environments on the same cluster hosting training pipelines, so notebooks keep GPU nodes, shared storage, and IAM identity. AWS claims consolidating interactive and training workloads lifts GPU utilization by up to 30%. SSM...
AMD's Helios rack packs 72 GPUs, Microsoft signs on
AMD unveiled its first rack-scale system to directly contest Nvidia at the rack level, with engineering samples in H2 2026 and mass production targeted Q2 2027. Microsoft joins Meta, OpenAI and Oracle as customers; Meta plans 1 gigawatt of Helios racks by year-end against a lo...
Instinct's AI assistant grants itself a perpetual, irrevocable license to everything it touches.
The private-beta personal assistant, from a small team led by an ex-Sierra research scientist and backed by Kleiner Perkins and Conviction, connects to email, messaging, calendar, device audio, location, screen, keystrokes and cursor data (TechCrunch). Its terms grant a sub-li...
Rillet raised $100M and hit unicorn status in 48 hours
TechCrunch's follow-up details the deal mechanics: led by ICONIQ with Sequoia reinvesting and a16z on the cap table, unplanned and inbound. Rillet rebuilt the general ledger rather than layering automation on an existing ERP, and has about 600 customers, most migrating off Ora...

Source trail

TechCrunch reports More from techcrunch.com

Entities

Etched GPU GPUs Harvard Investor Skeptical TechCrunch

Provenance

Canonical issue: Ramsay Research Agent — July 24, 2026
AI generated: no
Story unit: 2026-07-24-etched-hit-a-10-3b-valuation-for-gpu-free-transformer-inference
Labels: source-backed, canonical briefing excerpt

Fetching from the wire…

Etched hit a $10.3B valuation for GPU-free transformer inference.

Story

Related stories

Kog claims 30x faster LLM inference on existing GPUs through low-level reverse engineering.

Fish Audio raised $52M at $21M ARR one year in, off an open-source repo with 31K stars.

AWS put JupyterLab and VS Code inside EKS clusters, claiming 5 minutes vs 3–5 days of platform work.

AMD's Helios rack packs 72 GPUs, Microsoft signs on

Instinct's AI assistant grants itself a perpetual, irrevocable license to everything it touches.

Rillet raised $100M and hit unicorn status in 48 hours

Source trail

Entities

Provenance

Etched hit a $10.3B valuation for GPU-free transformer inference.

Story

Related stories

Kog claims 30x faster LLM inference on existing GPUs through low-level reverse engineering.

Fish Audio raised $52M at $21M ARR one year in, off an open-source repo with 31K stars.

AWS put JupyterLab and VS Code inside EKS clusters, claiming 5 minutes vs 3–5 days of platform work.

AMD's Helios rack packs 72 GPUs, Microsoft signs on

Instinct's AI assistant grants itself a perpetual, irrevocable license to everything it touches.

Rillet raised $100M and hit unicorn status in 48 hours

Source trail

Entities

Provenance

Story

Related stories

Kog claims 30x faster LLM inference on existing GPUs through low-level reverse engineering.

Fish Audio raised $52M at $21M ARR one year in, off an open-source repo with 31K stars.

AWS put JupyterLab and VS Code inside EKS clusters, claiming 5 minutes vs 3–5 days of platform work.

AMD's Helios rack packs 72 GPUs, Microsoft signs on

Instinct's AI assistant grants itself a perpetual, irrevocable license to everything it touches.

Rillet raised $100M and hit unicorn status in 48 hours

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance

Story

Related stories

Kog claims 30x faster LLM inference on existing GPUs through low-level reverse engineering.

Fish Audio raised $52M at $21M ARR one year in, off an open-source repo with 31K stars.

AWS put JupyterLab and VS Code inside EKS clusters, claiming 5 minutes vs 3–5 days of platform work.

AMD's Helios rack packs 72 GPUs, Microsoft signs on

Instinct's AI assistant grants itself a perpetual, irrevocable license to everything it touches.

Rillet raised $100M and hit unicorn status in 48 hours

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance