01
Bloomberg Technology ↗
Meta is making its biggest AI monetization push yet, and Bloomberg's Kurt Wagner sat down with CEO Mark Zuckerberg for an exclusive interview about it. They discussed why the company is betting on ultra-low API pricing to win developers, how Meta plans to turn its massive AI infrastructure spending into real revenue, and what the transformation means for employees and investors. Zuckerberg also addressed Meta's competitive position against OpenAI and Google in the race for AI dominance.
02
TechCrunch AI ↗
Paris-based AI voice startup Gradium closed a $100 million seed round, adding $30 million from Nvidia to an initial $70 million raised from FirstMark Capital, Eurazeo, DST Global Partners, Eric Schmidt, and Xavier Niel. The company, spun out of French AI lab Kyutai and co-founded by former Google Brain and DeepMind researcher Neil Zeghidour, builds audio AI models that deliver near-instantaneous voice responses for AI agents. Gradium is using the funding to open a Bay Area office to compete for talent, with automaker Renault already among its customers.
03
TechCrunch AI ↗
Meta is developing four new AI chips under its MTIA program, covering model training, ranking and recommendation, general AI workloads, and inference, with production beginning in September. The modular chiplet design, built with Broadcom, manufactured by TSMC, and paired with Samsung memory, is meant to cut Meta's GPU costs and reduce reliance on Nvidia and AMD even as the company keeps spending heavily with both. Meta plans to deploy 7 gigawatts of compute in 2026, doubling that in 2027, with capital expenditures projected at 125 to 145 billion dollars for the year.
04
VentureBeat ↗
A VentureBeat Pulse Research survey of 107 enterprises found that 69 percent run AI agents with shared credentials somewhere in their deployments, meaning a single compromised agent can inherit the access of every other agent using the same API key. That risk has fueled a wave of security acquisitions, including Palo Alto Networks' $21.1 billion purchase of CyberArk, CrowdStrike's $740 million acquisition of SGNL, and Cisco's planned purchase of non-human identity specialist Astrix Security. CrowdStrike has already shipped its first product from the SGNL deal, Continuous Identity for AI Agents, which validates each agent action in real time.
05
TechCrunch AI ↗
Open source AI tool Ollama raised a $65 million Series B led by Theory Ventures, adding to a $15 million Series A led by Benchmark's Peter Fenton for $88 million raised total. The company, which lets developers run open-weight AI models on personal computers, now counts 8.9 million monthly active developers, 176,000 GitHub stars, and roughly 17,000 forks, and says it is used at 85 percent of Fortune 500 companies. Ollama runs with just 14 employees and also offers a paid cloud tier priced by GPU usage rather than token limits.
06
Hacker News ↗
An independent developer built a free public tool that lets users see and edit what an open source language model is thinking mid-generation, based on Anthropic's recent paper on verbalizable representations forming a shared workspace inside language models. The tool uses a technique the developer calls a Jacobian Lens to visualize concepts lighting up in the model's middle layers before it commits to a word, and lets users manually insert or suppress concepts to change the output in real time. Testing across models found the technique worked well on Qwen and Llama architectures but showed little benefit on Pythia models, regardless of size.
07
Bloomberg Technology ↗
OpenAI unveiled ChatGPT Work, a new agent that can independently handle complex, multi-step tasks like building documents, spreadsheets, presentations, and web apps for hours at a time by breaking them into smaller steps. The tool is powered by GPT-5.6, OpenAI's newest model line released in three versions: Sol for maximum capability, Luna for speed, and Terra balancing performance with efficiency. ChatGPT Work began rolling out on web and mobile to Pro, Enterprise, and Edu users, with Plus and Business access following in the days after.
08
Bloomberg Technology ↗
Anthropic appointed former Federal Reserve chair Ben Bernanke to its Long Term Benefit Trust, the independent oversight body meant to hold the company accountable to its public mission. Bernanke, a Nobel laureate in economic sciences who led the Fed from 2006 to 2014 through the 2008 financial crisis, joins existing trustees Neil Buddy Shah, Richard Fontaine, and Mariano-Florentino Cuellar. All trust members are independent of Anthropic's management and investors, hold no equity, and are compensated only for their time.
09
TechMeme ↗
US venture capital deal value hit $412.7 billion in the first half of 2026, up nearly 30 percent from all of 2025. AI startups accounted for 86 percent of that total, or roughly $355.9 billion, underscoring how dominant AI has become in venture investing. The second quarter alone saw seven funding rounds of $1 billion or more.
10
CNBC Technology ↗
Meta publicly launched Muse Spark 1.1, a multistep reasoning AI model built for agentic coding, marking the first time the company has charged businesses for access to one of its models. Trained by Alexandr Wang's Meta Superintelligence Labs, the model is priced at $1.25 per million input tokens and $4.25 per million output tokens, positioning it against Anthropic's Claude Haiku 4.5 and OpenAI's GPT-5.6 Luna. Wang called the pricing very aggressive and attractive, and the launch was significant enough that Zuckerberg posted about it on X for the first time in three years.