Tech News August 2026: Microsoft Maia 300, DLSS 5 on RTX 50, and Unlimited Free ChatGPT
The week of August 10-16, 2026 is all about AI hardware: Microsoft is ramping up production of its Maia 300 chip, NVIDIA detailed the RTX 50-exclusive DLSS 5 at SIGGRAPH, and OpenAI made text chats unlimited and free with GPT-5.6 Luna.
Microsoft Ramps Up Production of Its Maia 300 AI Chip
A September Reveal and 300,000+ Units with TSMC
Microsoft plans to significantly boost production of its next-generation AI accelerator, the Maia 300, with a public unveiling expected as soon as September. According to The Information, the company is negotiating capacity with TSMC for more than 300,000 units with delivery in 2027, as part of its push to reduce reliance on NVIDIA for AI compute.
Azure AI Foundry and Copilot: Less Dependence on NVIDIA
Microsoft's plan is to expand internal use of the chip through Azure AI Foundry and Copilot, and later offer it to its largest external cloud customers. It's the clearest move yet from a hyperscaler that wants to stop paying NVIDIA's margin on every GPU its infrastructure consumes.
DLSS 5: NVIDIA's Neural Rendering Arrives This Fall, RTX 50 Only
Lighting and Materials Reconstructed with Diffusion Models
NVIDIA presented DLSS 5 at SIGGRAPH 2026, a real-time neural rendering model that reconstructs photorealistic lighting and materials with diffusion models. The company calls it the most significant advance in computer graphics since real-time ray tracing arrived in 2018.
Read also
Three AI Models and a Single GPU: RTX 40 Cards Are Left Out
DLSS 5 will run on a single GPU, but it will be exclusive to GeForce RTX 50 cards and arrive this fall. RTX 40 owners are left out, a key fact for anyone deciding which card to buy this year for gaming or 3D work.
OpenAI: GPT-5.6 Luna Free and Unlimited in ChatGPT
Unlimited Text Chats for Free and Go Users
Since August 6, GPT-5.6 Luna is the default model for Free and Go users of ChatGPT, with unlimited text chats and a new Think button for hard questions. OpenAI also improved GPT-5.6 Sol in accuracy and consistency, and ChatGPT recently passed 1 billion weekly users.
Hot Chips 2026: Murata Brings Agentic AI Power to the Last Millimeter
Murata announced its participation in Hot Chips 2026 with a clear message: while AI compute generates intelligence, its compact components and clean power delivery let agentic AI move from compute to interacting with the physical world. In its own words for the conference, powering AI "at the last millimeter".
The Rest of the Week: Pricier GPUs Due to Memory and the Farewell to the DALL·E GPT
The rising cost of memory chips keeps hitting graphics hardware: AMD has told its partners about an approximate 10% price increase on GPU die and GDDR memory bundles, effective since July and the second memory-driven hike in six months. And OpenAI will retire the DALL·E GPT inside ChatGPT on August 30, recommending users download their generated images beforehand.
Why These Stories Matter for Developers and 3D Artists
Competition between Microsoft and NVIDIA in AI chips points to cheaper compute in the medium term, good news for anyone training or serving models. The RTX 50-exclusive DLSS 5 affects GPU purchases: if you do 3D work and game, RTX 40 cards miss out on the headline technology. Free, unlimited AI in ChatGPT changes daily workflows, and memory-driven price hikes affect both new builds and hardware refreshes.
Conclusion
The week confirms three trends: hyperscalers want their own chips, AI graphics become a GPU buying argument, and consumer AI turns free and unlimited. Keep reading the blog so you don't miss the next weekly tech roundup.
