Major model developers navigated a complex week of commercial releases and government oversight. Anthropic restored global access to its Fable 5 and Mythos 5 models after a temporary suspension, introducing a new safety classifier to address a jailbreak technique. The company also launched Claude Sonnet 5, a more cost-effective model aimed at agentic tasks, now the default for its free and pro users. In contrast, OpenAI took a more cautious approach, launching its GPT-5.6 series in a limited release to just 20 government-vetted partners. This staggered rollout came at the request of the U.S. government ahead of an expected announcement on voluntary AI standards.
The push for vertical integration and greater control over the hardware stack also accelerated. OpenAI unveiled its first custom-designed AI inference chip, dubbed “Jalapeño,” developed in partnership with Broadcom in just nine months. The chip is designed to offer better performance per watt for inference workloads, reducing the company's reliance on third-party GPUs. In a related move, reports surfaced that Meta is developing a cloud infrastructure business to sell AI compute power directly to enterprises, a strategy that would position it as a new competitor to established players like Amazon Web Services, Microsoft Azure, and Google Cloud.
On a national scale, South Korea announced a monumental $974 billion industrial plan to bolster its dominance in the AI and semiconductor sectors. The investment includes a $260 billion commitment from Samsung for new fabrication plants and $36 billion for High Bandwidth Memory facilities. Not to be outdone, SK hynix outlined approximately $714 billion in long-term investments for new front-end, NAND, and advanced-packaging capacity. This strategic national initiative is set to significantly impact the global semiconductor supply chain, from manufacturing capacity to chip availability and pricing.
Venture capital continued to flow into the AI ecosystem at a historic pace, with June becoming the largest AI funding month on record. Among the most significant deals, AI infrastructure developer Together AI raised an $800 million Series C at an $8.3 billion valuation, led by Aramco Ventures. Energy infrastructure provider Joulent secured $1.75 billion from National Grid Ventures to support AI data centers. In specialized AI, video intelligence firm Twelve Labs raised $100 million in a Series B co-led by New Enterprise Associates and Naver Ventures, while Argentum AI received a $100 million note from Super Micro Computer and OXMIQ Labs raised $35 million involving Samsung Catalyst Fund.
Highlighting the immense valuations being placed on specialized AI applications, London-based voice technology firm ElevenLabs is reportedly in discussions for a secondary share sale that would value the company at around $22 billion. The potential valuation is supported by strong commercial traction, with the company's annual recurring revenue reaching $500 million in early 2026, up from $350 million at the end of 2025. With over $800 million in total funding, the company's growth underscores intense investor confidence in generative AI companies that have achieved significant product-market fit in high-value niche markets.