AI News Roundup: GPT-5.6 Rumors Intensify, Anthropic Eyes Custom Chips, South Korea’s Massive Semiconductor Push, and AI in Drug Discovery

The AI industry is buzzing with anticipation this week as OpenAI appears poised to drop a major model update, while hardware and biotech ambitions take center stage across the sector.

GPT-5.6 Release Speculation Builds

OpenAI is widely expected to release GPT-5.6 very soon, with many observers betting on a launch between July 7–8. The timing could prove strategic: Anthropic is reportedly winding down access to its Fable model in subscriptions, potentially freeing up users for OpenAI to attract with fresh capabilities and improved performance.

OpenAI has traditionally launched new models on Thursdays, though exceptions do occur. Adding fuel to the speculation, Tibo (head of Codex & ChatGPT at OpenAI) posted a cryptic tweet featuring a frog image captioned “iykyk” (if you know, you know). In AI circles, this is widely interpreted as a subtle hint pointing toward a Wednesday release.

Users are advised to prepare for temporary rate-limit resets once the model goes live.

---

Anthropic in Early Talks for Custom AI Chips

Anthropic has begun preliminary discussions with Samsung about manufacturing custom AI chips. The talks remain in the very early stages, highlighting how seriously the company is taking hardware independence.

This move comes as Anthropic trails behind OpenAI, which has already showcased prototypes of its own custom chips and is preparing for production and large-scale server deployment later this year.

---

South Korea Commits $518 Billion to Memory Chip Production

On June 29, 2026, South Korea announced a landmark 10-year investment plan led by Samsung Electronics and SK Hynix.

The two companies will together invest approximately $518 billion (800 trillion won) to build four new fabrication plants focused on memory chips, primarily to meet surging AI demand.

Memory — especially high-bandwidth memory (HBM) — has become one of the most critical bottlenecks for scaling AI infrastructure.

In 2025, AI chips consumed over 90% of global HBM production, with demand significantly outpacing supply and pushing prices higher into 2026.

While production capacity for GPUs can theoretically be reallocated, memory supply is largely locked in for years ahead, making expansions like South Korea’s strategically vital for the entire AI ecosystem.

Also read:

To Catch a Hacker, Think Like One: The Nuance in Modern Cybersecurity

Ludomania in a Tie: How Prediction Markets Are Making Gambling Look Respectable

Netflix Stock Plunges 45% from Record High – The Streaming Giant Needs a Fresh Growth Catalyst

Why Is Fertility in South Korea So Low?

---

Anthropic Plans to Develop Its Own Drugs with AI

Anthropic has signaled its intention to use its models to discover and develop new medicines, with a focus on treatments for “neglected” diseases. Eric Kauderer-Abrams, Head of Life Sciences at the company, confirmed the strategic direction.

The company has also launched Claude Science, an AI workbench that integrates tools and datasets to help scientists accelerate research. While no specific drug candidates or disease targets have been named yet, the move marks another step by frontier AI labs into real-world scientific applications beyond chat and coding.

Bringing any new drug to market will still require extensive lab work, clinical trials, and regulatory approval — a process that typically takes over a decade — but AI companies are increasingly positioning themselves to influence and potentially accelerate parts of this pipeline.

These developments reflect the maturing AI landscape: model releases, custom hardware, supply-chain investments, and vertical applications in science are all advancing in parallel. The coming weeks are likely to bring more concrete updates across all these fronts.