Published today, sourced from stories within the last 24 hours
🔥 Pentagon Signs Deals with 7 AI Companies for Classified Military AI — Anthropic Holds Out
The U.S. Department of Defense announced on May 1 that it has reached agreements with seven major tech companies — SpaceX, OpenAI, Google, NVIDIA, Reflection AI, Microsoft, and Amazon Web Services — to deploy their AI capabilities on classified military networks. The deals include a “lawful operational use” clause that grants the military broad deployment authority.
Notably absent: Anthropic, which publicly rejected the “lawful use” standard in a high-profile dispute with the Pentagon last month. The company’s stance has drawn both praise and criticism, highlighting the growing tension between AI safety commitments and defense contracting. The Pentagon described the move as accelerating its transformation into an “AI-first fighting force.”
Sources: The Guardian, NYT, Reuters, Bloomberg, Washington Post, BBC, Al Jazeera
☁️ OpenAI Models Land on AWS Bedrock as Microsoft-OpenAI Exclusivity Ends
After Microsoft and OpenAI officially ended their cloud exclusivity deal on April 27, the ripple effects are now fully visible. AWS announced immediate availability of OpenAI models on Amazon Bedrock, including GPT-5.5 (listed as coming soon), GPT-4.1, and the Codex coding agent. Even more significant: Bedrock Managed Agents powered by OpenAI — a new product allowing enterprises to build AI agents using OpenAI frontier models entirely within AWS infrastructure.
Key details: Microsoft retains a non-exclusive IP license on OpenAI models that now has a hard 2032 expiration. Microsoft no longer pays a revenue share to OpenAI for certain Azure deployments, but gains the ability to use OpenAI models in its own products freely. Satya Nadella defended the deal as strategically favorable despite losing exclusivity.
Sources: AWS, The New Stack, CIO Dive, Motley Fool, Microsoft 10-Q
🇨🇳 Huawei Targets $12B AI Chip Revenue in 2026 as Nvidia Stalls in China
Huawei expects its AI chip revenue to surge at least 60% to approximately $12 billion in 2026, up from $7.5B in 2025. The growth is driven by mass orders for the Ascend 950PR chip, which entered mass production in March, and the upcoming 950DT variant. Chinese tech firms are rapidly shifting to domestic processors as U.S. export controls keep Nvidia’s most advanced chips out of China.
Huawei is projected to capture roughly 60% of China’s AI chip market this year. The FT reports that DeepSeek V4’s launch has further accelerated orders for Huawei silicon. Nvidia’s H200 shipments remain stalled in regulatory limbo, leaving the door wide open for Huawei’s domestic dominance.
Sources: Financial Times, Tom’s Hardware, Android Headlines, WCCFtech, Seeking Alpha
📈 Alphabet Nears $5 Trillion Market Cap, on Verge of Overtaking Nvidia
Following a blowout earnings report, Alphabet’s market cap jumped ~$240 billion to $4.65 trillion, putting it within striking distance of Nvidia as the world’s most valuable company. GOOGL shares reached a high of $386.05 on May 1. A key catalyst: Alphabet announced it will sell its custom TPU AI accelerators directly to select customers for installation in their own data centers — a direct challenge to Nvidia’s chip dominance.
Meanwhile, Nvidia stock slipped as investors weighed rising competition from both Google and Amazon, which noted its in-house chip business is booming. The AI capex theme remains strong — analysts estimate combined Big Tech AI spending could approach $700 billion in 2026.
Sources: CNBC, Yahoo Finance, IndexBox, StockTwits, Schwab
⚖️ Meta Threatens to Pull Facebook & Instagram from New Mexico Ahead of Trial
A New Mexico bench trial starting May 4 could force sweeping changes to how Meta’s platforms operate. Following a $375 million jury verdict in March finding Meta harmed children’s mental health and violated state law, the remedies phase begins Monday. Meta has warned it may withdraw Facebook, Instagram, and WhatsApp from New Mexico entirely rather than comply with state-ordered child safety reforms it calls impractical.
New Mexico AG Raúl Torrez called Meta’s threat a “PR stunt.” This is the first of more than 40 state attorney general cases against Meta to reach trial, making it a potential bellwether for the broader legal landscape around AI-powered social platforms and youth safety.
Sources: Reuters, AP News, The Hill, Fox Business
🚀 Moonshot AI Open-Sources FlashKDA for Kimi Delta Attention
Moonshot AI (the team behind Kimi.ai) released FlashKDA, a high-performance CUTLASS-based kernel implementation of Kimi Delta Attention (KDA). The open-source release plugs directly into the flash-linear-attention ecosystem and benchmarks show it is meaningfully faster than existing implementations. The release supports variable-length batching and includes H20 GPU benchmarks.
This comes as Moonshot AI’s Kimi K2.6 model continues to climb leaderboards, ranking #12 overall and #6 on verified leaderboards — matching the performance of Qwen3.6 Max and DeepSeek V4 among open models.
Sources: MarkTechPost, llm-stats.com, DeepLearning.ai, BenchLM
📊 Also Making Waves
- AI capex continues to surge: Big Tech AI capital spending could approach $700B in 2026, reinforcing the infrastructure buildout thesis across hyperscalers.
- Micron emerging as AI chip contender: Analysts predict Micron could be the next AI chip stock to reach a $1 trillion valuation, driven by explosive memory demand for AI workloads.
- Microsoft stock under pressure: MSFT shares declined below $400 as analysts weigh the loss of OpenAI exclusivity, Copilot growth concerns, and elevated capex guidance.
Published May 2, 2026. All stories sourced from reports published within the last 24 hours. This digest is for informational purposes only and does not constitute financial or investment advice.
Leave a Reply