AI Startup Funding Rounds Shift Toward Divergent Strategies
AI startup funding rounds are increasingly split between hardware interfaces, real-time AI systems, and infrastructure startups, as investors search for scalable, defensible platforms beyond generic chatbot software and pure model development. This shift pairs large Series A funding valuation deals with smaller but strategic seed rounds, showing that capital is available for both ambitious platform plays and focused infrastructure tools. In this cycle, three companies stand out: Hark, which blends AI-native hardware with multimodal models; Decart, which builds real-time video and world-model systems; and Flexprice, which targets AI billing infrastructure. Together, they signal that investors are backing both foundational layers of the stack and specialized applications that sit close to users or revenue. The result is a more diversified AI ecosystem in which universal interfaces, high-speed inference, and real-time billing pipelines all compete for attention and budget.
Hark Bets on AI Hardware Interface and Universal Assistants
Hark raised more than USD 700 million (approx. RM3.22 billion) in an oversubscribed Series A round at a USD 6 billion (approx. RM27.6 billion) valuation to build AI systems and AI-native hardware designed as a universal interface between humans and machines. The round, led by Parkway Venture Capital with investors including Nvidia, AMD Ventures, Intel Capital, Qualcomm Ventures, and Salesforce Ventures, reflects confidence in AI hardware interface products rather than software alone. Hark is developing an integrated platform that combines agentic, multimodal models with personalized devices that can remember users, retain context, and operate across existing services. The company has grown to about 70 employees and secured a new NVIDIA B200 data center to train its upcoming models, which it plans to release later this summer. Its strategy targets real-time AI systems that feel more persistent and personal than current chatbot-based experiences.
Decart Targets Real-Time Video and World Models for Physical and Digital Worlds
Decart, a vertically integrated AI research lab focused on real-time video and world models, has secured USD 300 million (approx. RM1.32 billion) in funding led by Radical Ventures, with new investors such as NVIDIA, Adobe Ventures, Toyota Ventures, and eBay Ventures. The company’s core infrastructure product, DOS 2.0, delivers over 1,600 tokens per second for agentic inference and supports full-HD video at up to 100 frames per second across major cloud hardware. Built on this stack, Decart’s Lucy model targets immersive real-time AI experiences in gaming, e-commerce, and advertising, with sub-30-millisecond response times, while Oasis focuses on physical AI applications like robotics and autonomous vehicles via real-time simulation. According to Decart, Amazon is a strategic customer, and individual backers include OpenAI co-founder Andrej Karpathy and former Disney CEO Michael Eisner. The company positions itself at the intersection of real-time AI systems and next-generation inference infrastructure.
Flexprice Builds AI Billing Infrastructure for Usage-Based Monetization
Flexprice, an open-source billing infrastructure startup, has closed a USD 1.5 million (approx. RM6.9 million) seed round led by Shastra VC, with participation from Anupam Mittal and TDV Partners. Unlike AI hardware or model labs, Flexprice focuses on AI billing infrastructure for AI-native and API-first enterprises that rely on usage-based monetization. Its platform processes over 20 billion events per month and supports complex pricing tied to token consumption, API calls, GPU usage, and other real-time compute workloads. The company recorded 6X revenue growth in the last quarter and a 20-fold increase in event volume over the past year, and plans to expand across the US and Europe while building AI-native finance products for metering, revenue recognition, and financial reporting. Flexible pricing models include pay-as-you-go, prepaid credits, volume-based tiers, seat-based subscriptions, and hybrids, all integrated with payment providers such as Stripe, Adyen, and Razorpay.
What These Deals Reveal About AI Infrastructure and Vertical Bets
Taken together, Hark, Decart, and Flexprice show how AI startup funding rounds are diverging across the stack. Large Series A funding valuation levels, such as Hark’s USD 6 billion (approx. RM27.6 billion) mark and Decart’s USD 300 million (approx. RM1.32 billion) raise, indicate that investors continue to support ambitious platform plays in AI hardware interface devices and real-time AI systems despite broader consolidation concerns. At the same time, Flexprice’s USD 1.5 million (approx. RM6.9 million) seed round highlights growing demand for AI infrastructure startups that handle billing, pricing, and revenue automation for generative AI businesses. One quotable insight from Flexprice’s leadership is that “billing is the hardest layer to get right, and the most consequential when you get it wrong.” Capital is therefore flowing not only to headline-grabbing models and hardware, but also to critical, often overlooked infrastructure that keeps AI businesses commercially viable.