Meta–Midjourney Deal, Apple Eyes Gemini for Siri, Google’s AI Footprint, Nvidia’s FugakuNEXT

Important: If you're reading this in the Promotions tab, please drag it to your Inbox and click ‘Yes’ when your email provider asks to do it for future emails.

Meta is partnering with Midjourney and will license its technology for ‘future models and products’
(Click 👆️ to read more)

• Meta will license Midjourney’s aesthetic technology for future AI models and products, aiming to bring richer visual quality to its platforms.

• Midjourney’s founder, David Holz, emphasized the lab remains independent, community-backed, and investor-free, even with the partnership.

• Specific terms—including financials, implementation timeline, and whether this affects Midjourney’s planned Enterprise API—weren’t disclosed.

Apple in talks to use Google's Gemini AI to power revamped Siri
(Click 👆️ to read more)

• Apple is in early talks with Alphabet to possibly use Google’s Gemini AI to revamp Siri, with a decision expected later, potentially next year.

• Both Apple’s and Google’s stocks rose following the report, although no formal agreement or comment was issued by either company.

• The move follows delays in Apple’s Siri overhaul, initially due last spring but postponed by a year, as Apple weighs external AI solutions—including earlier talks with Anthropic and OpenAI.

How much energy does Google’s AI use? We did the math
(Click 👆️ to read more)

• Google estimated that the median text-generation Gemini prompt consumes 0.24 Wh of energy, emits 0.03 g CO₂e, and uses 0.26 mL of water—equivalent to watching TV for under nine seconds.

• Over the past year, Gemini’s energy and carbon footprint per prompt dropped by 33× and 44× respectively, thanks to full-stack optimizations across hardware, algorithms, and data-center efficiency.

• Google detailed its measurement approach, factoring in idle power, cooling, CPU/RAM overhead, and infrastructure, aiming to drive industry-wide consistency around AI environmental footprints.

Nvidia joins development of successor to Japan's Fugaku supercomputer
(Click 👆️ to read more)

• Nvidia, Fujitsu, and Japan’s Riken institute will collaborate on "FugakuNEXT," the successor to the Fugaku supercomputer, expected around 2030.

• FugakuNEXT will combine Fujitsu CPUs and Nvidia GPUs, targeting 5-to-10× the performance of Fugaku, with exascale-level capabilities of 1 sextillion calculations per second.

• Based in Kobe, FugakuNEXT will support advanced AI workloads, aiding research in drug discovery, automotive design, and climate change mitigation.

Bluesky blocks service in Mississippi over age assurance law
(Click 👆️ to read more)

• Bluesky has blocked access in Mississippi rather than comply with a new law requiring age verification and parental consent for minors.

• The compliance burden, including collecting sensitive data and building new infrastructure, was deemed too resource-intensive for the small platform.

• Bluesky warned that the law threatens free speech, disadvantages smaller players, and could entrench larger tech companies.