TechCapsules

NVIDIA unveiled GR00T 2, an upgraded world‑action model, and Isaac Lab 3.0, the release of its robot‑learning framework, at the Conference on Robot Learning in Austin. GR00T 2 was trained on 100,000 hours of egocentric robot and human data and can complete 40 of 50 benchmark tasks without task‑specific demonstrations, achieving 71 % success with a single demonstration and 95 % with five. The model ranks in the top three across vision‑language‑action benchmarks and supports an API for sequencing skills and error [...]

Google is preparing to roll out two re‑engineered TPU accelerators—Sunfish TPU 8t and Zebrafish TPU 8i—later in 2026, with broader deployment slated for Q1 2027 and a multi‑region ramp through 2028. The 8t, fabricated by Broadcom and TSMC, targets AI training, featuring a 3‑D torus interconnect linking 9,600 chips and delivering 121 exaflops at FP4 precision, plus 216 GB HBM3E per device. A derivative, the 8tx, swaps the torus for a lower‑latency “boardfly” fabric and replaces Google’s Axion ARM CPU with an AMD EPYC [...]

HTEC’s latest research shows that semiconductor firms are ahead of most industries in AI adoption, with 48% labeling themselves as AI‑native or scaling AI versus 44% overall. Investment growth mirrors the broader market, as 78% report increased AI spending. The sector distinguishes itself by prioritizing strategic alignment and demanding formal ROI models before project approval—45% and 36% respectively. However, once AI initiatives launch, measurement fragments: 38% rely on case‑by‑case reporting and only 28% use enterprise dashboards, compared [...]

Apple has filed a second regulatory request in China to extend its Apple Intelligence suite beyond the iPhone to iPad, MacBook and desktop Macs. The move follows the iOS 27.3 developer beta that introduced the AI features on Chinese iPhones after a two‑year delay compared with global rollouts. The earlier July registration covered only smartphones, so a new filing was required for tablets and computers. Apple is collaborating with Alibaba to train a China‑specific large language model and [...]

Meta announced on Monday that it is rolling out its AI‑driven age‑estimation system to Instagram users worldwide, aiming to flag accounts that appear to belong to teenagers even when an adult birthdate is entered. The technology scans profiles for contextual clues—such as birthday posts, school‑grade references, and visual cues like height or bone structure—across text, images and video. Identified teen accounts will be shifted into Meta’s “Teen Account” mode, which restricts direct messages to followers, hides potentially [...]

A wave of agentic AI coding tools has sparked a surge of game mash‑ups, browser ports and reverse‑engineered remakes, from Halo in a web browser to Spider‑Man traversing Arkham City. Industry veterans, including Nightdive Studios founder Stephen Kick, acknowledge the technology’s ability to decode opaque game binaries and accelerate modding, though Nightdive has not yet adopted it. Kick warns that publishers may respond with aggressive legal actions and a shift toward streaming‑only distribution to protect assets. Former [...]

Apple announced its first paid “intelligence usage packs,” a $49.99 one‑time purchase that unlocks additional AI processing for features in Logic Pro. The option appears in the Logic Pro menu under Intelligence Features → Show Usage → Buy More Usage. Until now Apple Intelligence tools have been free, but the company warned that expanded access would eventually cost users. The new model mirrors common AI services that limit free usage before requiring payment, though Apple chooses a single purchase rather than [...]

Robotics firms have bet heavily on massive data sets to teach robots human‑level tasks, but the approach has stalled as companies cannot determine the volume or type of data needed and the resulting models remain opaque. At the 2026 Conference on Robot Learning, researchers from UC Berkeley and NVIDIA unveiled “agentic robotics,” a method that replaces data‑driven imitation learning with large‑language‑model‑generated, human‑readable code. The technique decomposes a target task into sub‑tasks and prompts an LLM to produce [...]