Science has achieved what was once confined to the realm of fantasy: the direwolf, extinct for over 10,000 years and popularized by the Game of Thrones series, has been brought back to life thanks to an ambitious de-extinction project led by the biotechnology company Colossal Biosciences.

The Texas-based company announced on Monday the birth of three direwolf pups, created through a combination of cloning and genetic editing. The pups include two males, Romulus and Remus, both six months old, and a three-month-old female named Khaleesi, in honor of the character from the HBO series.

The process that led to these wolves began with the extraction of ancient DNA from a 13,000-year-old tooth and a 72,000-year-old skull. Using this genetic material, scientists edited 20 genes in the cells of a modern gray wolf, its closest relative. These cells were then implanted into domestic dog eggs for gestation.

“The birth of these pups is a huge milestone,” said Ben Lamm, CEO of Colossal, who emphasized that this is the world’s first successful de-extinction. “This achievement shows that our technology works and marks the beginning of a new era for species conservation,” he said.

NASA announced another delay for its Artemis 3 mission, pushing the first manned lunar landing since 1972 to mid-2027. Technical problems with the Orion spacecraft’s heat shield were cited as the primary reason for the postponement.

“Safety is our top priority, and we won’t proceed until we are ready,” said NASA Administrator Bill Nelson. He emphasized that the delay will still keep the U.S. ahead of China’s plan to send a crewed mission to the Moon by 2030.

Artemis 1 successfully sent the uncrewed Orion capsule around the Moon in 2022, but subsequent missions have faced repeated setbacks.

The development of the current generation of AI chatbots has relied heavily on specialized microprocessors, pioneered by Nvidia, which has dominated the market and become a symbol of the artificial intelligence boom.

However, the same qualities that make these GPU (graphics processing unit) chips so effective for building powerful AI systems from scratch render them less efficient for running AI products on a daily basis. This has created opportunities in the AI chip industry for rivals aiming to compete with Nvidia by offering so-called AI inference chips. These chips are specifically designed to handle the operational needs of AI tools while cutting down on the significant costs associated with generative AI.

“These companies see opportunities for this kind of specialized hardware,” said Jacob Feldgoise, an analyst at Georgetown University’s Center for Security and Emerging Technology. “As the adoption of these models grows, the need for inference calculations will increase, driving demand for inference chips.”