Summary: Brett Adcock, the serial entrepreneur behind Figure AI (humanoid robotics) and Archer Aviation (electric air taxis), has raised $700 million for his third company: Hark. The raise, announced May 21, 2026, values Hark at $6 billion before any product has shipped. The investor list — Nvidia, AMD, Intel Capital, Qualcomm Ventures — signals that the chip industry is betting on Hark’s hardware ambitions. Adcock’s pitch: a proactive AI that learns who you are, operates through speech and vision, and eventually runs on purpose-built hardware. It is one of the largest early-stage bets in AI hardware since the category emerged as a serious investment thesis.

What Is Hark

Adcock describes Hark as a “universal interface” with the digital world.

The specific design is still largely private, but the public framing is consistent: Hark aims to build an AI that is proactive rather than reactive, personalized rather than generic, and multimodal (speech, vision, memory) rather than text-only. Rather than a standalone chatbot, Adcock envisions it sitting on top of — and interacting with — existing digital products and services. You do not replace your apps; you acquire an AI layer that interacts with all of them on your behalf.

The company plans to release its first multimodal models this summer (2026), which will power a personal AI platform built on top of existing products. Custom hardware follows: Hark will design devices specifically optimized for these AI systems, not repurpose existing consumer hardware.

The company remains secretive about engineering details, team composition, and its specific hardware roadmap. What Adcock has described publicly is the vision layer, not the architecture.

Why This Round Is Unusual

$700 million at $6 billion post-money, pre-product, is a large number even in 2026’s AI funding environment.

What distinguishes this round is not the dollar amount but the investor composition. The lead is Parkway Venture Capital. But the strategic participants — Nvidia, AMD Ventures, Intel Capital, Qualcomm Ventures — are chip manufacturers with a direct financial interest in what hardware Hark builds and which silicon it runs on. Getting all four to co-invest in the same pre-product company is a meaningful signal. Each of these firms typically backs a small number of AI hardware startups per year, and they tend to compete rather than co-invest.

Brookfield and Salesforce Ventures round out the cap table, adding infrastructure capital and enterprise software validation.

The reading: major chip companies believe Hark’s hardware ambitions are credible enough to anchor early.

Brett Adcock’s Track Record

Adcock is not a first-time founder making large promises. He has executed two capital-intensive ventures to meaningful scale:

Archer Aviation: Founded 2018. Electric air taxi startup. Went public via SPAC in 2021. As of 2026, Archer has completed FAA-certified test flights and entered pre-commercial operations. The company built physical aircraft that fly. This required managing FAA certification, complex supply chains, and multi-year hardware development timelines.

Figure AI: Founded 2022. Built a bipedal humanoid robot — Figure 01 — from zero to commercial deployment. Secured a partnership with BMW for automotive manufacturing. The company raised over $675 million before Adcock stepped back from day-to-day operations to found Hark.

Both companies involved: designing novel hardware, navigating regulatory environments, and managing the gap between vision and production reality. The timelines were long. The capital requirements were large. Both produced physical products.

This context matters for evaluating Hark’s prospects. Adcock’s stated goal — custom AI hardware — is not a new aspiration for him. The question is whether AI consumer hardware is a more tractable problem than robotics and aviation. There are reasons to believe it is (smaller form factor, faster iteration, no FAA), and reasons to be cautious (competitive landscape, AI Humane Pin precedents, platform distribution).

The Market Context

Hark is entering a personal AI hardware market that has already produced high-profile failures.

The Humane AI Pin (2024) raised over $230 million, launched to terrible reviews, and failed to gain adoption. Rabbit R1 similarly failed to demonstrate value beyond novelty. Both companies bet on standalone AI hardware as a category before the underlying AI was capable enough to justify the form factor.

What has changed: frontier AI models in 2026 are substantially more capable than those that powered Humane and Rabbit. Voice interaction quality has improved dramatically. On-device inference is increasingly viable with Qualcomm and Apple silicon. The gap between “AI is impressive in demos” and “AI is useful enough to wear daily” has narrowed.

Whether Hark’s timing is right — or whether it is repeating the category’s mistakes with better capital — is the central question for this company.

The Nvidia/AMD/Intel co-investment suggests the chip industry believes the category’s time has come. Their investment is also, in part, a hedge: if any company breaks through in personal AI hardware, they want to supply the silicon.

What to Watch

  • Summer 2026 model release: The first concrete deliverable. Quality and reception of Hark’s multimodal models will signal whether the software foundation is real.
  • Hardware reveal: When and how Hark shows its device design. Consumer AI hardware has a design-credibility problem. The first look will matter.
  • Platform relationships: A “universal interface” for existing apps requires platform cooperation (or workarounds). How Hark negotiates or bypasses iOS/Android/desktop gatekeeping is a key structural challenge.
  • Team disclosure: Hark has been opaque about who is building this. As the company moves toward launch, its engineering leadership will become more visible.

Rating

Hark has no public product as of this writing. This rating reflects potential based on founder track record, investor signal, and market timing — not demonstrated performance.

Potential: 3.8/5 — Adcock has built hardware companies before and delivered physical products on difficult timelines. The investor composition (all four major chip manufacturers) is unusual and credibility-raising. The personal AI hardware category has failed before, and the failure modes are well-documented. If the summer 2026 model launch is strong, the rating climbs. If it mirrors Humane’s trajectory, the category thesis fails again regardless of capital.