- Hark closed a $700M+ Series A at a $6B post-money valuation, two months out of stealth with no shipped product.
- Nvidia, AMD, Intel, and Qualcomm all wrote checks — a rare alignment of rival silicon vendors on a single cap table.
- Founder Brett Adcock (Figure, Archer) is betting that integrating models, software, and native hardware from day one beats single-layer plays.
- The category that ate Humane AI Pin and Rabbit R1 is now the most expensive bet in consumer AI — and the supply allocation problem may be partly solved.
A 70-person startup with 70 GPUs and zero customers just raised more capital in a Series A than most companies see across their entire lifecycle. Hark, founded by Figure and Archer entrepreneur Brett Adcock, banked $700 million at a $6 billion post-money valuation this week — and the most interesting part isn’t the number. It’s who signed the check.
The Round That Shouldn’t Make Sense
A Series A bigger than most growth rounds
Parkway Venture Capital led the round, but the rest of the syndicate is what changes the read. Nvidia, AMD Ventures, Intel Capital, and Qualcomm Ventures all participated — four silicon vendors who normally compete for the same data center allocation budget, now aligned behind the same consumer AI device bet. ARK Invest, Brookfield, Greycroft, Prime Movers Lab, Salesforce Ventures, Align Ventures, and Tamarack Global filled the rest of the cap table.
By Bloomberg’s reporting, Hark closed roughly two months after emerging from stealth. The company has 70 employees, runs a private data center on Nvidia B200 GPUs, and has not disclosed headcount targets, form factor, target price, launch market, or a customer pipeline. The Series A buys silence.
Business Insight — When Nvidia, AMD, Intel, and Qualcomm all show up on the same round, the investment is partly equity and partly supply allocation insurance. The binding constraint on AI hardware in 2026 isn’t capital — it’s getting B200s in the building when you need them. Hark may have just solved that problem before solving the product problem.
What Hark Is Actually Building
A “universal interface” between humans and machines
Hark describes itself as developing a personal AI platform that pairs in-house foundation models, software, and native hardware. The pitch from director of design Abidur Chowdhury, a former Apple product executive, is that the existing AI market is optimized for software builders, not for normal users — and that the way to fix that is to own every layer from model to silicon.
The company says it will ship four capabilities: natural speech and vision interaction, persistent memory that compounds across sessions, deep personalization, and proactive operation that anticipates needs. First multimodal models are scheduled for release this summer, with the native hardware following at an undisclosed date.
Founder résumé as collateral
Adcock has shipped hardware before. He co-founded Vettery (sold to Adecco for $100M), founded electric aircraft maker Archer (SPAC-listed in 2021), still runs humanoid robotics firm Figure, and founded school-security company Cover. He seeded Hark with $100 million of his own money in late 2025. For an early-stage AI hardware play, that operating history is the closest thing to a moat the syndicate could underwrite.
Business Insight — The competitive read here is sharp: Anthropic and OpenAI are both prioritizing coding and developer-facing tools as their commercial wedge. Hark is one of the few well-funded teams explicitly aiming at the mass-market consumer interface layer. That positioning is either visionary or a $700 million blind spot — there is no middle outcome.
The Graveyard Hark Has to Avoid
Humane, Rabbit, and the cost of being early
Hark enters a category that has burned capital faster than it has produced revenue. Humane’s AI Pin became the highest-profile failure of 2024. The Rabbit R1 followed a similar arc — strong launch, weak retention, quiet retrenchment. Even Apple, with the most refined hardware distribution machine on the planet, has spent the past year visibly struggling to define what on-device AI should be.
The structural problem the entire category shares: pushing rich personal context into an always-on assistant without violating the privacy of everyone standing near the user. Meta’s existing smart glasses have not solved this. Forthcoming Android spectacles do not appear to have solved it either. When TechCrunch pressed Chowdhury on the question, his answer was a smile and the line: “Sounds like that would make a great product.”
What $700M actually buys
The fresh capital is earmarked for top hardware, design, and AI research talent, plus compute and components. What it does not buy is product-market fit — and by Hark’s own timeline, the device that turns the models into a business is “still further out.” That is the gap operators should watch most closely over the next twelve months.
Business Insight — For CEOs evaluating consumer AI partnerships in the next 12 months, the practical question is not whether Hark will ship — it is whether enough of the major model labs continue to deprioritize the consumer interface layer. If they do, Hark inherits a category by default. If even one major lab pivots to native hardware, the $6B valuation gets stress-tested very fast.
Related
- Forget ChatGPT — SAP Just Bet $1B on Tables
- Anthropic Just Overtook OpenAI in the Enterprise
- Sierra’s $950M Round and the Fortune 50 Agent Race
- Amazon Tested 40 Voice AI Startups. This One Took 100%.
- AI Just Found 10,000 Bugs Nobody Can Patch Yet
Sources
- TechCrunch — Hark raises $700M Series A for its secretive ‘universal’ AI interface
- The Next Web — Brett Adcock’s AI hardware startup Hark raises $700m at $6bn valuation
- Bloomberg — AI Hardware Startup Hark Valued at $6 Billion in New Funding Round
- BusinessWire — Hark Raises $700M Series A at a $6B Valuation
AI Biz Insider · AI Business EN · aibizinsider.com
댓글 남기기