Guilherme Breda Rezende’s Post

😂 September 2025 Tech Forecast 🚀 Happy new month This month, AI isn’t just advancing… it’s sprinting like it’s late for a Monday meeting! 🍏 Apple is about to drop “Apple Intelligence” with the iPhone 17 — basically giving Siri a PhD. 🪟 Windows 11 is rolling out new AI features — soon even File Explorer might know where you hid that “Final_Final_2” document. 💪 Nvidia is flexing again, preparing a new AI chip. At this rate, their GPUs will need a gym membership. 🏠 IFA Berlin is showing off AI in homes — one step closer to your fridge judging you for late-night snacking. We’re watching these shifts closely. Because whether it’s phones, laptops, or fridges with attitudes… AI is shaping our future in ways we can’t ignore — but maybe we can laugh about. 😅 👉 Which of these September AI updates excites (or scares) you most? #AISisters #AIForAfrica #TechHumor #SeptemberUpdates

🚀 This morning, Dr. Vinesh Sukumar, VP of Product Management for AI at Qualcomm took the stage for his keynote: “The Future of Gen AI Innovations on Snapdragon Platforms.” He unveiled how Snapdragon is driving a new era of on-device generative AI - with breakthrough speed, ultra-low latency, and unmatched efficiency. From smartphones to PCs and beyond, these innovations are set to transform how we experience AI every day. The future of Gen AI is here - and it’s powered by Snapdragon. ✨ See who will be taking to the stage next https://lnkd.in/esFQxnz8 #AIInfraSummit #GenerativeAI #Qualcomm #Snapdragon

Apple just 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲𝗱 𝘁𝘄𝗼 𝗶𝗺𝗽𝗿𝗲𝘀𝘀𝗶𝘃𝗲 𝗻𝗲𝘄 𝘃𝗶𝘀𝗶𝗼𝗻-𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 on Hugging Face: FastVLM and MobileCLIP2! 𝗙𝗮𝘀𝘁𝗩𝗟𝗠:  • Built around FastViTHD, a 𝗵𝘆𝗯𝗿𝗶𝗱 𝘃𝗶𝘀𝗶𝗼𝗻 𝗲𝗻𝗰𝗼𝗱𝗲𝗿 that dramatically reduces visual tokens while handling high-resolution images streamlining both speed and accuracy in VLM workflows  • Offers up to 𝟴𝟱× 𝗳𝗮𝘀𝘁𝗲𝗿 𝘁𝗶𝗺𝗲 to first token than previous systems (e.g., LLaVA-OneVision-0.5B), and is 3.4× smaller, lowering memory and serving costs  • Supports 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝘃𝗶𝗱𝗲𝗼 𝗰𝗮𝗽𝘁𝗶𝗼𝗻𝗶𝗻𝗴 directly in-browser, powered by WebGPU 𝗠𝗼𝗯𝗶𝗹𝗲𝗖𝗟𝗜𝗣𝟮:  • A mobile-optimized image–text model with state-of-the-art zero-shot capabilities, leveraging training on a large dataset (DFNDR-2B)  • Matches FastVLM in efficiency: up to 𝟴𝟱× 𝗳𝗮𝘀𝘁𝗲𝗿 𝗮𝗻𝗱 𝟯.𝟰× 𝘀𝗺𝗮𝗹𝗹𝗲𝗿 than its predecessor  • Designed to run entirely on-device no cloud dependency needed making it ideal for iPhones, iPads, and Macs If you think Apple isn’t making strides in AI, think again these releases underscore quiet, powerful advances in on-device intelligence that go far beyond the LLM headline noise

So, the new capabilities and hardware on the iPhone 17 are to let LLMs run locally. Makes sense, as the future will be “Small LLMs.” Apple tucked a neural accelerator straight into the GPU, basically turning your phone into a pocket-sized AI workstation. You don’t need to run GPT5 on your commute, but now it’s about lightweight, specialised models that actually make sense on-device. Still, there was no major AI feature or capability announced for or with this new iPhone. #Apple #iPhone #LLM #AI

Apple just dropped the iPhone 17 lineup yesterday — and yes, the new orange 17 Pro looks like bubble tea. 🧋 But behind the aesthetics is a serious leap in on-device AI compute powered by the new A19 Pro chip: 👉 What’s new under the hood: • 6-core CPU (2 performance + 4 efficiency) with smarter branch prediction + 50% bigger cache. • 6-core GPU (5 on Air) — each with Neural Accelerators for AI workloads. • 16-core Neural Engine: Apple claims MacBook Pro-level AI capability on your wrist-sized device. • Dynamic Caching 2.0 + doubled 16-bit FP math → faster ML + graphics. • Vapor chamber cooling (yes, in a phone) with deionized water → up to 40% better sustained performance vs. A18 Pro. In plain English: the iPhone just turned into a pocket-sized AI workstation. So whether you’re training models, crunching images, or just scrolling TikTok… you’re basically holding a cybernetic bubble tea in your hand. #apple #iphone17 #machinelearning #ai #tech

The iPhone 17 just raised Apple’s AI game. This isn’t just a camera or chip upgrade. It’s a clear move toward 2026, when Apple’s AI vision needs to deliver. Here’s what stood out: 1. A19 chips built for AI 2. Smarter Siri, live translation 3. On-device processing for privacy 4. A new iPhone Air that’s ultra thin, but AI-ready 2026 is the real test. #iPhone17 #AppleIntelligence #AI2026 #OnDeviceAI #AppleEvent #MobileAI #A19Chip #TechTrends

🧠 𝗶𝗣𝗵𝗼𝗻𝗲 𝟭𝟳 𝗣𝗿𝗼’𝘀 𝗔𝗜 𝗰𝗵𝗶𝗽 𝗶𝘀 𝗽𝗼𝘄𝗲𝗿𝗳𝘂𝗹, 𝗯𝘂𝘁 𝗳𝗼𝗿 𝘄𝗵𝗼, 𝗔𝗽𝗽𝗹𝗲 𝗼𝗿 𝘆𝗼𝘂? Apple just unveiled the A19 Pro chip, and it’s not just a spec bump, it’s a statement. 40% better sustained performance. Neural accelerators in every GPU core. A redesigned thermal system so power doesn’t throttle when you actually push the phone hard. This is Apple doubling down on making the iPhone the primary hub for AI. But here’s the part worth pausing on: Are these gains unlocking new value for users , or simply giving Apple more control and justification to raise prices? 𝗙𝗼𝗿 𝗽𝗼𝘄𝗲𝗿 𝘂𝘀𝗲𝗿𝘀, 𝘁𝗵𝗲𝗿𝗲’𝘀 𝗻𝗼 𝗱𝗲𝗻𝘆𝗶𝗻𝗴 𝘁𝗵𝗲 𝗹𝗲𝗮𝗽: ✅ Faster on-device AI – Apple Intelligence, Siri offline, generative image tools all run locally, keeping data private. ✅ Pro workflows that hold up – 4K video editing, AR apps, gaming 40% sustained performance gains mean fewer slowdowns under pressure. ✅ Battery efficiency that surprises – despite the added muscle, A19 Pro promises longer all day battery life. 𝗕𝘂𝘁 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗮𝘃𝗲𝗿𝗮𝗴𝗲 𝘂𝘀𝗲𝗿? 𝗦𝗼𝗰𝗶𝗮𝗹 media scrolling and casual video capture were already smooth on the iPhone 15 and 16. The difference you’ll feel might be smaller than Apple’s keynote makes it sound. That’s why I find this launch so interesting. Apple is building the most advanced consumer AI platform on the planet , but the business model still relies on nudging millions of users into an annual or bi-annual upgrade cycle. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻 𝗶𝘀𝗻’𝘁 𝗵𝗼𝘄 𝗳𝗮𝘀𝘁 𝘁𝗵𝗶𝘀 𝗰𝗵𝗶𝗽 𝗶𝘀. 𝗜𝘁’𝘀 𝘄𝗵𝗲𝘁𝗵𝗲𝗿 𝗶𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗲𝘅𝗽𝗮𝗻𝗱𝘀 𝘄𝗵𝗮𝘁 𝘄𝗲 𝗰𝗮𝗻 𝗱𝗼, 𝗼𝗿 𝗷𝘂𝘀𝘁 𝗲𝘅𝗽𝗮𝗻𝗱𝘀 𝗔𝗽𝗽𝗹𝗲’𝘀 𝗴𝗿𝗶𝗽 𝗼𝗻 𝗼𝘂𝗿 𝘄𝗮𝗹𝗹𝗲𝘁𝘀. 𝗪𝗵𝗮𝘁 𝗱𝗼 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸 , is the iPhone 17 Pro a creative unlock, or just another very expensive flex? 👇 #Innovation #Technology #DigitalTransformatio #iphone #AppleTechnology

Yesterday I had the privilege of attending a special session reviewing Apple’s latest devices as an AI specialist and shared my insights on their hardware capabilities for on-device processing. Here is my take on how Apple is carefully building the infrastructure for intelligent experiences without making a show of it. Apple’s September 9, 2025 session did not feature flashy demonstrations but the introduction of the A19 Pro chip with neural accelerators integrated across GPU cores signals a deliberate shift toward powerful and efficient on-device processing. The strategy is clear: improve daily experiences like Live Translation, health monitoring, and gaming performance while maintaining user privacy. What struck me most was Apple’s layered approach. The iPhone Air supports seamless tasks for everyday users, while the iPhone 17 Pro with its six-core GPU, vapor-chamber cooling, and advanced thermal design is built for sustained and complex workloads. This conservative yet potent direction contrasts with competitors pushing cloud-based experiences and underscores Apple’s commitment to performance, efficiency, and data protection.

Google's Pixel 10 launch just proved something nobody wants to admit. Hardware wars are officially over. While Samsung and Apple fight over megapixels and battery life, Google said "screw it" and went all-in on AI. The new Pixel lineup barely mentions specs. Instead? Magic Cue reads your texts and surfaces restaurant info before you ask. Gemini Nano runs locally without internet. Real-time translation happens on-device. Real talk: I've been testing similar AI features for months. The shift is massive. Users don't care about your chip's nanometer process. They want their phone to think ahead of them. Google's betting the entire smartphone game changes when your device becomes genuinely intelligent instead of just powerful. Bold move. Especially when their hardware has always been... okay at best. Are you ready for phones that predict what you need, or still chasing benchmark scores? 📱 #GooglePixel #AIFirst #SmartphoneEvolution #TensorG5 #GeminiAI