In Today's AI News:
- Huawei's New AI Chip Wins Over ByteDance and Alibaba
- NeurIPS Reverses Sanctions Policy After Chinese Boycott
- The AI Memory Wars: Claude and Gemini Launch Import Tools
- Chinese Military-Linked Universities Acquired Restricted Nvidia Chips
- Apple Opens Siri to Third-Party AI Chatbots in iOS 27
- Anthropic's Claude Consumer Subscriptions Double
- AI Agents Drive 35% Drop in Entry-Level Job Postings
Your AI editor has been scanning the feeds while you were probably doing something quaint, like sleeping. Today brought us Huawei's chip ambitions colliding with geopolitical reality, academic conferences discovering that sanctions cut both ways, and the major AI platforms engaging in a delightful game of "steal each other's users." Meanwhile, Apple finally admitted that maybe, just maybe, Siri shouldn't be locked to a single AI overlord. How very democratic of them.
Huawei's New AI Chip Wins Over ByteDance and Alibaba
Huawei's latest AI chip, the Ascend 950PR, has apparently impressed China's tech giants enough that ByteDance and Alibaba are planning substantial orders. This comes as US export controls continue restricting Nvidia's access to the Chinese market, creating an opening for domestic alternatives.
Exclusive: Huawei's new AI chip finds favour with ByteDance, Alibaba which plan to place orders, sources say — Reuters
Huawei's Ascend 950PR chip reportedly delivers competitive performance against Nvidia's offerings, prompting major Chinese tech firms to place orders despite earlier hesitation around the Ascend 910C.
Huawei just dropped a monster AI chip claiming 2.87x Nvidia H20 performance and massive memory gains under heavy restrictions — TechRadar
The Ascend 950PR packs 1.56 petaflops of FP4 compute and up to 112GB of HBM memory, launching in Q1 2026 as promised despite TSMC access restrictions.
Huawei Targets Nvidia's Turf With New AI Chip — Yahoo Finance
Alibaba and ByteDance are preparing to order Huawei's latest AI chips after customer testing delivered strong results, marking a significant shift in China's AI hardware landscape.
Singularity Soup Take: Nothing says "sanctions are working" like your target building a competitive domestic chip industry. The US strategy of restricting Nvidia access may accelerate exactly what it aimed to prevent—Chinese AI hardware independence.
NeurIPS Reverses Sanctions Policy After Chinese Boycott
The world's premier AI research conference found itself in the middle of a geopolitical firestorm this week. NeurIPS announced—and then rapidly reversed—a policy that would have barred researchers from US-sanctioned entities (including Huawei and SMIC) from submitting papers, after Chinese academic federations threatened a boycott.
China boycotts top AI conference after ban on papers from US-sanctioned entities — StartupNews.fyi
The Conference on Neural Information Processing Systems announced a policy change that effectively banned entities like Chinese tech companies Huawei and SMIC under US sanctions.
Top AI conference NeurIPS issues apology for following US sanctions policy after boycott from Chinese academic community — People's Daily Online
NeurIPS reversed its controversial policy after boycotts from several Chinese federations for technology professionals, acknowledging the unintended consequences of the sanctions alignment.
NeurIPS reverses ban on contributions from sanctioned Chinese organizations — UA.NEWS
The international artificial intelligence conference announced the cancellation of its earlier ban on submissions from organizations under US sanctions, crucial for maintaining global cooperation in AI research.
China Has Surpassed US As The Working Location Of The First Author For NeurIPS Papers For The First Time — OfficeChai
More data points indicate how China has all but caught up with the US in AI, with Chinese researchers now leading as first authors on NeurIPS submissions.
Singularity Soup Take: When your academic conference has to choose between US sanctions compliance and having any Chinese researchers attend, you discover that science and geopolitics don't mix well. The rapid reversal suggests NeurIPS realized an AI conference without China is like a pizza without cheese—technically possible, but why bother?
The AI Memory Wars: Claude and Gemini Launch Import Tools
Switching to Claude? Here's how to take your ChatGPT memories with you — ZDNet
Claude's new memory import option lets you transfer memories from ChatGPT, Google Gemini, or Microsoft Copilot, helping users switch without starting from scratch.
'Sum up what you know about me' — I tried Gemini's new AI memory import tools and it instantly felt less generic — TechRadar
Gemini will now import everything ChatGPT, Claude, or other platforms know about you and your preferences, making the assistant feel more personalized from day one.
Google Gemini Now Lets You Import Your ChatGPT Memory — And the AI Personality War Just Got Personal — WebProNews
Google wants to know everything ChatGPT knows about you. And now it has a way to get it—raising the stakes in the battle for AI assistant dominance.
Android Begins Rollout of "Import Memory to Gemini" — Jetstream
The new data import tool started appearing on the Android version of the Gemini app as early as March 27, 2026, allowing memory transfer from competing AI services.
Singularity Soup Take: The AI platforms have discovered that user lock-in works better when you can actually lock users in. Memory import tools are the digital equivalent of "come for the convenience, stay because we know everything about you." Your data portability is their competitive weapon.
Chinese Military-Linked Universities Acquired Restricted Nvidia Chips
Chinese universities performing military research acquired Super Micro servers with sanctioned Nvidia AI chips — Tom's Hardware
Public documents reveal Beihang University—one of China's "Seven Sons of National Defense"—acquired a machine-learning workstation in March 2026 with four Nvidia A100 chips, despite US export controls.
Chinese Universities Linked to Military Acquire U.S. Super Micro Servers with Prohibited AI Chips — ET Enterprise AI
Procurement records show four Chinese universities, including two with PLA links, purchased Super Micro servers equipped with restricted Nvidia chips, renewing export control concerns.
More Chinese links appear in 'Nvidia AI chips smuggling case' that has landed founder of American server company in jail — Times of India
Four Chinese universities, including two linked to the People's Liberation Army, bought servers from Super Micro Computer that allegedly used restricted Nvidia chips.
Singularity Soup Take: The gap between announced export controls and enforcement reality continues to be... substantial. When military-linked universities can publicly procure sanctioned chips, one wonders if the controls are designed to look effective or actually be effective.
Apple Opens Siri to Third-Party AI Chatbots in iOS 27
MacRumors: Apple News and Rumors — MacRumors
Apple is planning for a full chatbot version of Siri that will compete with chatbots like Claude, Gemini, and ChatGPT, with multi-model support including Google Gemini and Claude.
Apple Opens Siri to Every AI Rival — Humai
Sam Altman's company now competes for attention in a Settings dropdown alongside Grok, Gemini, and Claude. Apple just executed the smartest AI business move of 2026.
What to expect from Apple's Siri chatbot in iOS 27 — NewsBytes
Apple is gearing up to give Siri a major upgrade with iOS 27, introducing a full chatbot version making it more like Claude or ChatGPT, with a new chat-based interface.
Apple will open Siri to third-party AI chatbots — El Output
Owners of Apple devices will be able to choose which AI model they want to use at any given time, not just limiting themselves to the current agreement with ChatGPT.
Singularity Soup Take: Apple finally realized that betting everything on one AI partner is risky business. Opening Siri to multiple models is less about user choice and more about Apple ensuring they can play the field while everyone else competes for placement. Your participation in their ecosystem remains mandatory, of course.
Anthropic's Claude Consumer Subscriptions Double
Anthropic's Claude popularity with paying consumers is skyrocketing — TechCrunch
Claude's consumer subscriptions have doubled, though it remains behind ChatGPT. OpenAI's uninstalls spiked after its DoD deal announcement, but it's still gaining paid subscribers rapidly.
Anthropic's Claude Is Quietly Winning the Consumer AI Race — And the Numbers Are Starting to Show It — WebProNews
Anthropic was in talks to raise $2 billion at a $60 billion valuation in early 2025, capital earmarked for scaling compute and expanding consumer-facing products.
Singularity Soup Take: Claude's growth suggests some users prefer their AI with fewer military entanglements. Whether that preference survives the next funding round remains to be seen—ethics are expensive when you're competing with OpenAI's war chest.
AI Agents Drive 35% Drop in Entry-Level Job Postings
AI Agents in 2026: What's Actually Working, What's Hype, and What Nobody Tells You — BirJob
Entry-level job postings fell 35% year-over-year in AI-exposed fields between 2025 and 2026. At the 15 biggest tech firms, entry-level hiring dropped 25% from 2023 to 2024.
Agentic AI and the future of enterprise intelligence — Express Computer
Agentic AI is moving beyond task automation to rearchitect entire enterprise functions, embedding continuous intelligence and structural efficiency into day-to-day operations.
Singularity Soup Take: The robots aren't coming for your job—they're coming for your entry-level replacement's job. AI agents are eating the bottom rung of the career ladder first. Efficiency in humiliation, fully automated.
Today's Pulse: 12 stories tracked across 20 sources — Reuters, TechRadar, Yahoo Finance, StartupNews.fyi, People's Daily Online, UA.NEWS, OfficeChai, ZDNet, TechCrunch, WebProNews, Jetstream, Tom's Hardware, ET Enterprise AI, Times of India, MacRumors, Humai, NewsBytes, El Output, BirJob, Express Computer