Urgent.News

What's breaking now, across thousands of outlets.

AI

Top tech and startup stories to start your day with

Happy Monday! Menlo Ventures, an early backer of Anthropic, says that the real constraint in AI development is compute power, not a slowdown in frontier models. Menlo's partner Matt Murphy, who led the firm's Anthropic investments, described the slowdown as a "red herring." The frontier AI models are already intelligent enough to serve the current market and present an enormous opportunity for all AI companies, Murphy told ET.

Menlo Ventures also recently invested heavily in Anthropic, leading a $750 million funding round that valued the company at $18.4 billion. This substantial investment increased Menlo's stake to nearly $14 billion by June, resulting in the largest cheque in the firm's history. Menlo has since raised $3 billion across two new funds, the largest corpus in its 50-year history.

AI portfolio: Menlo's other AI investments include Suno, Lovable, Modal, and the work assistant Town. The firm also backed OpenRouter, which was acquired by Stripe for approximately $7.5 billion. Menlo believes that Anthropic has paved the way for enterprise and consumer-focused AI assistants, similar to how Anthropic broke into enterprise markets and OpenAI captured the consumer segment.

On the other hand, Menlo's partners noted that the market is experiencing a mix of growth and high valuations. Some companies are achieving rapid revenue growth, going from zero to $1 billion in just 18 months, while smaller startups command billion-dollar valuations. Bhavish Aggarwal, founder and CEO of Ola Electric, pledged a 4.32% stake in the company to fund his participation in a proposed rights issue.

Written by urgent.news from The Economic Times's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at economictimes.indiatimes.com →

More in AI

Stop Instagram and TikTok labelling your Photoshop and Premiere exports as AI

Quick fix: the label comes from Content Credentials (C2PA) that Photoshop and Premiere Pro write into your export after a generative edit, and export settings don't switch them off.

  • Adobe adds Content Credentials metadata to exports after generative edits.
  • C2PA markers appear as APP11 segments in JPEGs and uuid boxes in MP4/MOVs.
  • MetaClean removes AI metadata from exported files without re-encoding.

Quiz Master: AI Quiz Learning Companion

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend What I Built I built Quiz Master, an AI-powered quiz application designed to make learning more interactive, engaging…

  • Quiz Master is an AI-powered quiz application.
  • Created by Divya Hodke for interactive learning.
  • Open innovation model makes AI accessible.

Running 100B+ MoE Models on a Single RTX 4090: A Practical Guide to Expert Offloading with llama.cpp

Tuần này trên Hacker News có một bài hơn 600 điểm: chạy một model MoE 125B tham số trên một con RTX 4090 mà vẫn đạt tốc độ sinh token rất đáng nể.

  • A 125-billion parameter MoE model runs on a 24GB VRAM RTX 4090 using llama.cpp
  • llama.cpp offers two expert offloading methods: --n-cpu-moe and -ot flags
  • Quantizing KV caches to 8-bit reduces VRAM usage significantly

More from Monday 5 October →