What's New in AI, April 10, 2026
Key numbers from today's stories, April 10, 2026
Q1 2026 AI Funding Hit $300 Billion
Global venture funding reached $300 billion in Q1 2026, up 150% year over year. AI startups captured $242 billion of that, roughly 80% of all capital deployed. The single largest round: OpenAI closed $122 billion on March 31 at an $852 billion valuation. [1]
The concentration is significant. A small group of frontier labs is absorbing the majority of available capital. For founders outside that tier, the practical implication is that compute costs, talent, and partnership leverage all tilt further toward the incumbents.
Building lean, shipping fast, and differentiating on vertical expertise matters more now than at any point in the past two years.
Chinese Open-Source Models Now Lead Global Adoption
The ATOM Report released this week confirms that Chinese open language models have overtaken US and European models in cumulative downloads, inference token share, and model derivatives. The numbers: 1.15 billion downloads, 73% of inference tokens running through Chinese models. Alibaba's Qwen family leads the ecosystem, outperforming the next eight major labs combined on derivative model count. [2]
If you are building on open-source foundations, this is the ecosystem you are building on. Model selection and fine-tuning strategy should account for where the most active development is happening.
White House Proposes a Federal AI Framework That Favors Builders
The White House released a national AI framework proposing to preempt restrictive state laws like California's SB 53 and Colorado's AI Act. The framework shields developers from liability for third-party misuse and protects a "Right to Compute." It also refuses to ban AI training on copyrighted data and proposes streamlining federal data center permitting. [3]
One restriction worth noting: tools accessed by minors would face strict age-assurance requirements, including parental attestation. [3]
If this passes, it replaces a patchwork of state regulations with a single federal standard. For anyone shipping AI products across state lines, that is a real reduction in compliance overhead.
Sources
${sourcesHtml}
Originally published on chento.io