The May 24 brief aggregates Nvidia's ~$90B deal spree, Barclays' warning that Big Tech AI debt is now testing investment-grade capacity, and BlackRock CIO Wei Li attributing major earnings upgrades to "AI lifting the whole market." The story line for executives: AI capex is increasingly a credit-market signal, not just an equity-market one. Academic Research
Snapshot — May 24, 2026
27 stories
- Alibaba's Qwen 3.7 Max — first shown as a preview on May 20 — is now fully live on OpenRouter and DashScope, completing the rollout in under a week.
- The launch lands as Chinese frontier labs continue compressing the price/performance frontier;
- Qwen 3.7 Max arrives alongside DeepSeek V4-Pro's permanent 75% discount pricing made effective May 22.
- Amazon's 2026 AI infrastructure capex is now estimated at roughly $200B — the largest spend among Microsoft, Google, Meta, and Oracle.
- AWS is growing at its fastest pace in nearly four years, driven by AI training and inference workloads.
- Together the figure positions Amazon as the single biggest capex spender in the current buildout cycle.
Amazon's Bee wearable, an always-listening AI companion device, drew mixed early reviews — intrigue for its conversational summarization capabilities, but renewed privacy concerns over continuous-audio capture. The product positions Amazon directly against Humane, Rabbit, and a fast-growing category of dedicated AI hardware separate from the smartphone.
Reporting today suggests Anthropic will continue supplying models to the NSA despite the Pentagon recently flagging it as a supply chain risk and replacing its $200M DoD contract with awards to eight other vendors. Intelligence agencies are reported to lack access to NVIDIA's latest Grace Blackwell chips, and Anthropic's "Mythos" model is described as filling a specific intelligence-use gap – complicating a cleanly drawn boundary between commercial and national-security AI.
- Researchers from the University of Maryland, Google, Meta, and other institutions used a system called AutoTTS to let a coding agent independently search for control algorithms for AI reasoning.
- The agent surfaced a non-obvious algorithm humans likely would not have designed, reducing compute for test-time scaling by approximately 70%.
- Standard Chartered confirmed AI-driven role reductions and Meta announced reassignment of more than 7,000 employees into AI-focused teams.
- The dual story line — banks and Big Tech simultaneously using AI as a workforce-restructuring lever — is the strongest single signal of accelerating enterprise AI adoption inside the last week.
- Loizos reports that even Google is making AI security decisions in real time as model deployments outpace governance processes.
- The piece sits against the backdrop of the Trump administration's cancelled AI safety executive order earlier in the week — leaving a vacuum that states (California) and the EU AI Act are positioned to fill.
- Within hours of each other, Google DeepMind CEO Demis Hassabis described current progress as the beginning of the singularity, while Meta's Yann LeCun argued today's systems are not genuinely intelligent.
- Gemini co-lead Oriol Vinyals split the difference.
- The exchange has become the weekend's dominant frame for how senior lab leaders disagree on what current capabilities actually represent.
Weekend recaps consolidated Meta's May 20 round of ~8,000 layoffs (≈10% of workforce) and the disclosure that Meta's internal MCI tool had captured engineer keystrokes and screen captures to train replacement AI agents. Additional layoff waves are flagged for August, tied to Meta's ~$135B 2026 AI capex plan.
Mathematician Adam Kucharski showed that Microsoft Copilot, on default model settings, produced fabricated country-level differences when fed identical datasets labeled with different country names. The episode is being used to argue that defaults across Copilot, Gemini, and other assistants are routing analytical work to under-powered models – a reliability and procurement-risk story for enterprises.
- Microsoft Research released Webwright, a terminal-native web-agent framework, scoring 60.1% on the Odysseys long-horizon benchmark versus 33.5% for base GPT-5.4.
- The release is one of the strongest open-sourced web-agent stacks to date and signals continued Microsoft investment in agent infrastructure alongside its model partnerships.
- Nvidia Research published Gated DeltaNet-2, a linear-attention layer that decouples the "erase" and "write" operations inside the delta rule.
- The design targets long-context throughput at sub-softmax cost — relevant for both training efficiency and serving long-context agents at scale.
- Research Breakthroughs HOT RESEARCH
- Nvidia reported $81.6B in quarterly revenue (up 85% YoY), with the data center segment alone at $75.2B (up 92%), and disclosed $43B in startup holdings.
- The print was strong enough for Jensen Huang to claim a "brand new" $200B market for Nvidia, but Michael Burry doubled down on his Substack call comparing Nvidia to Cisco circa 1999 — prompting Nvidia to send sell-side analysts a rebuttal memo, an unusual move.
- OpenAI shipped a beta ChatGPT add-in for Microsoft PowerPoint that lets free and paid users build and edit slides from a sidebar directly inside the app.
- The integration plants ChatGPT inside one of Microsoft's flagship surfaces just as Copilot tries to defend the same workflow — read as escalation in the OpenAI-Microsoft frenemy dynamic.
Publishers are increasingly contending with AI-generated, unlicensed audiobook versions of copyrighted titles surfacing on YouTube. Takedowns remain cumbersome, and several publishers are hiring third-party detection firms to monitor and remove the content – an emerging IP-enforcement vector that may shape future platform liability debates.
Sources surveyed: Bloomberg, Tech Times, Invezz, Yahoo Finance, TechCrunch, VentureBeat, MarkTechPost, Ars Technica, USA Today, The Next Web, Analytics Insight, Mashable, Decrypt, Google DeepMind Blog, Apple ML Research, Stanford HAI, Carnegie Mellon, The Batch (DeepLearning.AI), Cerebras IR, codersera, and the AI Track.
- Stanford's flagship benchmark report finds industry produced over 90% of notable frontier models in 2025, with SWE-bench Verified rising from 60% to near-100% in a single year and organizational AI adoption reaching 88%.
- Several models now meet or exceed human baselines on PhD-level science, multimodal reasoning, and competition mathematics — strong validation that the frontier is still moving, not converging.
Stanford's HAI dataset, updated in tracking reports through May 24, shows 233 AI-related regulatory actions passed globally in 2024 — a 56.4% increase year over year and the fastest growth rate since the institute began tracking. The OECD AI Policy Observatory now counts more than 1,000 AI policy instruments across 69+ countries, providing the most comprehensive academic dataset for cross-jurisdictional compliance analysis.
- StepFun shipped StepAudio 2.5 Realtime, an end-to-end voice model with roleplay-specific RLHF and paralinguistic comprehension.
- The release pushes the China voice-AI stack toward parity with OpenAI's Realtime API and reflects a wider 2026 trend of voice-first agentic interfaces.
- 2.
- Products & Tools
- Hurbean (West University of Timișoara), Necula (Alexandru Ioan Cuza University), and Stepan published a peer-reviewed systematic review consolidating the literature on how AI is being embedded into ERP platforms — covering trends, deployment patterns, and forward-looking research directions.
- As one of the highest-revenue enterprise AI categories with relatively thin academic synthesis to date, the review maps the practitioner-research gap and offers a useful waypoint for tracking applied AI adoption literature.
- Steven Rosenbaum's book The Future of Truth has come under scrutiny after at least six quotes were identified as likely AI fabrications.
- The author initially accepted "full responsibility" before further details emerged.
- The incident adds to a growing list of trust-erosion events involving generative AI in published nonfiction, intensifying calls for editorial fact-verification standards as AI-assisted writing tools proliferate across mainstream publishing workflows.
- Berkeley Law adopted a stricter policy limiting student use of LLMs in core legal coursework — including conceptualizing, outlining, drafting, revising, editing, translating, and any exam use — while preserving instructor discretion and AI-focused courses.
- The policy takes effect summer 2026 and is notable as one of the first elite law schools to put a definitive boundary in writing.
- A new VentureBeat analysis flags an emerging category of incidents enterprises aren't tracking: agent-initiated actions that are technically correct given incomplete context, but cascade through downstream infrastructure.
- With 79% of organizations now running agents in production and Gartner projecting 33% of enterprise software will be agentic by 2028, the lack of a unified postmortem framework is becoming a measurable risk.
- AI economist Oren Etzioni's analysis catalogs 12 AI labs that have collectively raised more than $29 billion at a combined valuation approaching $130 billion — without shipping a single customer-purchasable product.
- Top of the list: Project Prometheus ($38B, Bezos/Bajaj), Safe Superintelligence ($32B, Sutskever), Thinking Machines Lab ($12B, Murati), and Reflection AI ($8B).
xAI launched Grok Build, a coding agent CLI aimed at developers and crypto-focused builders, running on Grok 4.3 beta with a 2M-token context and up to 8 parallel subagents. The release positions xAI alongside Anthropic's Claude Code and OpenAI's Codex CLI as the third major terminal-native coding agent, reframing the developer-tools race as a three-way contest rather than the Cursor/Copilot duopoly of 2025.
- xAI today expanded Grok Build — its terminal coding agent positioned as the company's answer to Claude Code and OpenAI Codex CLI — from the $300/month SuperGrok Heavy tier down to standard SuperGrok ($30/mo) and X Premium+ ($40/mo).
- The expansion ships alongside v0.1.218 (Linux image-paste fix, Windows shortcut remap, long-session crash prevention).