טכנולוגיה סלולרית

Shockwaves in 5G: SpaceX’s $17 Billion Spectrum Coup, EU Nixes ‘Big Tech Tax’, Trump’s 5G Surprise
 ·  ·  · 

גלי הלם ב-5G: מהלך הספקטרום של SpaceX בשווי 17 מיליארד דולר, האיחוד האירופי דוחה את "מס הטכנולוגיה הגדולה", ההפתעה של טראמפ ב-5G

עובדות מרכזיות האצת פריסת תדרים ורשתות שחרור ספקטרום מסיבי: ממשלות פותחות באגרסיביות ספקטרום חדש כדי להאיץ את התרחבות ה-4G/5G. בבריטניה, הרגולטור Ofcom יזם את המכירה הפומבית הגדולה ביותר אי פעם של ספקטרום סלולרי, ומעמיד לרשות הציבור 5.4 גיגה-הרץ של mmWave באזורים עירוניים בצפיפות
ספטמבר 22, 2025

Technology News

  • Tesla sued over 2023 Model 3 crash as Amazon layoffs hit engineers; markets edge higher
    November 23, 2025, 3:18 AM EST. US stocks rose on rate-cut optimism as Fed officials signal near-term relief. In tech business news, Tesla is being sued in Washington over a 2023 fatal Model 3 crash, with plaintiffs alleging a defective, unique door handle design left doors inoperable and slowed rescues. Tesla has not commented. Separately, Amazon's mass layoff round hit engineers hardest, with about 40% of roughly 4,700 cuts across several states, plus ripple effects in video games, advertising, and AI search. The updates highlight ongoing risks and shifts in the tech and mobility sectors.
  • Tesla FSD v14.2: Neural Network Vision Upgrade, Gestures and New Self-Driving Stats
    November 23, 2025, 3:16 AM EST. Tesla's FSD update v14.2 goes out to early-access customers and brings a neural network vision upgrade and a new Self-Driving Stats menu. The upgraded vision encoder uses higher-resolution imagery to improve dealing with emergency vehicles, obstacles, and even human gestures such as a stop or a wave. Tesla cautions this isn't a small update, hinting at expanded parameters or training behind the scenes. A new Autopilot menu feature, Self-Driving Stats, tracks total miles driven with FSD and miles driven with FSD enabled, showing a non-retroactive percentage since v14.2. Stats are per-vehicle and cannot be reset by driver; resets may appear later as FSD evolves. The next major release, v14.3, is expected weeks after v14.2, with Musk teasing a potential sense of sentience.
  • Connecting in the AI Era: Rediscovering Latent Skills for an AI-Driven Career
    November 23, 2025, 3:12 AM EST. An Op-Ed on how the GenAI era demands more than prompts. Barton College research shows many companies lack in-house AI skills to move beyond experimentation. The challenge isn't only upskilling or reskilling but learning how to apply existing capabilities across AI-enabled roles. The piece urges rediscovering latent skills-the lifelong competencies that make us unique-and applying them to new contexts, from Business Analyst to AI Auditor or Data Scientist. Even frontline experiences, like fast-food work, build empathy, problem-solving, and critical thinking that today's jobs require. The author argues that cultivating a high-performance culture means integrating past strengths with new tools, turning connections into career leverage in the age of AI.
  • Generative AI meets the genome: Evo completes genes and probes novel toxin design
    November 23, 2025, 3:08 AM EST. Researchers trained Evo on bacterial genomes and show that prompting a large DNA sequence lets the model behave like an LLM for genomic context. When given 30% of a known gene, Evo completes about 85% of the rest; with 80% prompts, it recovers the full sequence and can spot missing genes in a cluster. The system also highlights the most critical protein regions, reflecting learned evolutionary constraints on allowable changes. To probe novelty, researchers supplied a toxin sequence only loosely related to known toxins and filtered outputs that resembled existing antitoxin motifs. The work demonstrates how generative AI can infer functional genomic context and explore novel protein outputs, while underscoring biosafety considerations.
  • Google: Double AI infrastructure capacity every six months to meet demand
    November 23, 2025, 3:06 AM EST. During an all-hands, Google's AI infrastructure head Amin Vahdat said Google must double serving capacity every six months to meet AI demand, aiming for a 1000x growth in 4-5 years. The plan targets growth in compute, storage and networking while keeping cost and energy roughly level. He stressed success will require collaboration and co-design, not just spending. The challenge mirrors rivals like OpenAI, which plans six new data centers under the Stargate partnership with SoftBank and Oracle and has pledged over $400B to reach nearly 7 gigawatts of capacity. With 800 million weekly ChatGPT users and paid subscribers hitting limits for features like video synthesis, the AI infrastructure race remains the most critical and expensive frontier, where Google seeks to be more reliable, performant and scalable.