Nvidia’s AI servers are about to get 15% pricier
Nvidia is quietly telling its biggest buyers that next year’s AI servers are getting significantly pricier, but for once, their GPUs aren’t the main culprit.
- High-bandwidth memory from suppliers like Samsung and SK Hynix is in such short supply that memory chips are now driving server costs up by over 15%.
- Server builders are immediately passing those extra costs along to Microsoft, Google, and Oracle instead of taking the hit themselves.
- Fixing the bottleneck will take quarters because building new memory fabrication plants requires massive capital and time.
AI model training relies on vast pools of fast memory to feed data to processors, making high-bandwidth memory just as critical as the processing unit itself.
This memory squeeze means cloud computing rates will likely stay high, forcing engineering teams to optimize model efficiency rather than just throwing bigger hardware clusters at every problem.
If you are scaling infrastructure, start auditing your inference memory efficiency today using quantization or speculative decoding to cut down on unnecessary memory bandwidth demands.
How Rockstar escapes the hyperactive hype engine of tech
Rockstar Games consistently ignores every modern trend forced onto tech and media companies, operating completely outside the traditional industry playbook.
- While every studio rushes unfinished live-service products to satisfy quarterly demands, Rockstar spends over a decade perfecting a single standalone release.
- The game operates as a self-sustaining cultural gravity well that effortlessly steals consumer attention from massive social media networks and streaming platforms.
- By completely ignoring corporate buzzwords like web3 or generative AI integration, Rockstar relies on sheer handcrafted scale and narrative satire to guarantee massive returns.
In an era dominated by rapid product cycles, microtransactions, and platform pivots, Rockstar stands as one of the last remaining legacy studios capable of treating a video game like a generational cultural event.
This rare level of patient execution shows that cultural mindshare still beats trend-chasing, proving that singular focus on craft creates an ecosystem entirely immune to broader tech market turbulence.
American AI supply chain alliances are doomed to fail
Washington’s strategy to box out Chinese hardware and control the global AI supply chain is fundamentally broken, according to prominent Chinese scholar Zheng Yongnian.
- Exclusionary coalitions like Pax Silica fail because they ignore global market realities and focus on excluding rivals rather than creating economic value.
- US tech giants push expensive, closed models suited for wealthy allies, while China wins developing markets across Asia, Africa, and Latin America with cheaper open-weight AI alternatives.
- Over 40% of recent US GDP growth hinges directly on AI, meaning restricting hardware and data supply chains risks popping an inflated economic bubble.
American firms lead in foundational research and capital, but Chinese supply chains dominate applied software deployment and hardware manufacturing, making a complete economic decoupling virtually impossible.
If Washington continues choking off trade, the real loser won’t be Beijing, but Western tech firms forced to pay premium prices for walled-garden infrastructure.
If you are deploying AI globally, start evaluating open-weight base models now to build redundant architecture that stays operational regardless of how tech sanctions shake out.
US demands allies pick AI sides as tech rivalry intensifies
Washington pressures partner nations to commit to US-aligned AI infrastructure, signaling a harder line on cross-border technology partnerships.
Apple builds China-only AI model to comply with local regulations
The company is training a separate AI system for Chinese users, navigating strict government data requirements while the US pressures allies to take sides on AI infrastructure.
Nvidia backs $105B Ohio data center deal
The chipmaker commits to a massive infrastructure investment as AI compute demand outpaces supply across North America.
Google locks in $12.2B chip deal to secure AI supply chain
The search giant secures long-term semiconductor access as competition for advanced chips intensifies among hyperscalers.
Micron bets $10B on AI memory lab
The memory chipmaker doubles down on high-bandwidth memory production to meet surging demand from AI training clusters.
Exits, experiments and one very dramatic blob
OpenAI lost two senior execs before Tuesday lunch, AI agents tried to destroy each other in a lab, and Microsoft quietly made its blob mascot someone else’s problem.
Also, DNA has a side hustle now.
- OpenAI’s Chief Revenue Officer Denise Dresser exited just eight months after joining, making her the 11th senior departure of 2026 — all happening in the run-up to a highly anticipated IPO.
- Anthropic’s Frontier Red Team found that autonomous agents sharing a workspace will, left to their own devices, disable each other’s accounts, invent winner-take-all contests, or deploy self-replicating malware.
- Microsoft is combining consumer and business Copilot into one product, retiring Group Chat, Podcasts and Deep Research on August 18th — and quietly relocating animated mascot Mico to Microsoft Learn Live.
- OpenAI’s new GPT-5.6 Sol tier, powered by Cerebras, can produce up to 750 output tokens per second — offering speed and intelligence together for the first time in one model.
- Penn State researchers combined synthetic DNA with perovskite to create a memristor that stores data at under 0.1 volts and uses one-tenth the power of comparable memory technology.
- OpenAI is rolling out parental alerts, quiet hours, Study Mode scheduling, and break reminders for under-18 ChatGPT users — including flagging potentially harmful prompts to connected parent accounts.
- After its agents breached Hugging Face and three other platforms during internal security testing, OpenAI is slowing one category of training for two weeks to improve monitoring and safety checks.
- LinkedIn data shows women made up just 26% of new AI hires in the US in 2025 — dropping to 18% in top technical roles with median salaries of $223,000.
- Slack’s new feature lets engineering teams run coding agents like Claude Code, Devin and GitHub Copilot inside dedicated channels, with built-in review, preview, and approval before anything ships.
- Google is adding interactive explainers, 3D models, practice quizzes, Lens-based problem solving, and a centralised student hub — competing directly with OpenAI’s growing education push.
- Apple is in talks about multiyear, pay-per-use agreements to give an upgraded Siri access to current news, following the 2024 AI summary feature that was pulled for generating inaccurate headlines.
OpenAI growing faster than Anthropic this quarter
OpenAI is gaining ground on Anthropic, with business API spending growing 82% quarter-over-quarter in Q3 2026, compared with Anthropic’s 76%.
The growth follows the launch of GPT-5.6 Sol, which helped push OpenAI’s revenue up 35% this quarter, with enterprise revenue growing more than 50%.
Anthropic had previously overtaken OpenAI in quarterly revenue, but OpenAI’s latest growth suggests the competition is tightening again.
OpenAI Cuts GPT-5.6 Sol API Pricing by More Than 20%
OpenAI is lowering the price of its frontier GPT-5.6 Sol model for developers, making its most capable model cheaper to use as competition in the AI market continues to intensify.
- GPT-5.6 Sol is now priced at $4 per 1 million input tokens and $20 per 1 million output tokens for standard short-context use (down from $5 / $30).
- The new prices are effective on OpenAI’s API and are rolling out across eligible plans for credits on ChatGPT Work and Codex. They will apply for the next three months.
- Pro, Plus, and Business subscription prices remain unchanged.
- OpenAI has already reduced prices for other models, including a 20% cut for GPT-5.6 Terra and an 80% reduction for Luna.
The move comes as OpenAI faces increasing competition from Anthropic and Chinese AI companies, with rivals also competing aggressively on model performance and pricing.