AI Startup Hark Completes $700 Million Series A Financing, Post-Money Valuation of $6 Billion, Develops Next-Generation AI Hardware
This round was led by Parkway Venture Capital, with follow-on investments from NVIDIA, AMD Ventures, ARK Invest, Brookfield, Intel Capital, Qualcomm Ventures, and Salesforce Ventures.
Hark plans to expand its team from about 70 to 200 engineers, focusing on expanding GPU infrastructure, accelerating large model development, and developing next-generation AI hardware.
Source: Public Information
ABAB AI Insight
Hark was founded by former core engineers from OpenAI and DeepMind, and has previously showcased early prototypes of personal intelligent agents on a small scale. This substantial Series A financing continues the trend of high valuations in the AI agent sector leading up to 2026, with earlier projects like Cognition and Adept also receiving significant investments.
In terms of capital strategy, Hark will direct funds towards its own GPU clusters and hardware development, with chip giants like NVIDIA, AMD, and Intel participating directly, motivated by the goal of capturing the "personal AGI" entry point. They aim to build highly personalized systems through voice, vision, and persistent memory, while also accumulating early training data and user feedback for hardware development.
Similar to the high valuation expansions of vertical AI tools like Perplexity and ElevenLabs, and OpenAI's personal agent strategic layout, the current AI startup industry is transitioning from general models to highly personalized agents. Early hardware and model vertical integration teams are accelerating their pursuit through substantial financing.
Essentially, this represents capital concentration: the $6 billion valuation shifts pricing power from traditional cloud GPU leasing to self-developed personal intelligent hardware platforms. The mechanism is that personalized memory and multimodal interaction require dedicated computing power and data loops, with chip giants' participation forming supply chain ties, accelerating the structural shift from shared models to personal exclusive AGI.
ABAB News · Law of Cognition
When chip giants lead investments, personal intelligent agents are no longer just software, but a new hardware track. A team of 70 people valued at $6 billion means that the speed of cash burn will determine who gets AI into everyone's daily lives first. The deeper the personalized memory, the higher the platform stickiness, and pricing power shifts from models to personal data sovereignty.