← Back to Blog AI News EN

AI News Weekly Summary — April 27 – May 3, 2026 | Big Tech Q1, Manus Blocked, Pentagon

Microsoft, Alphabet, Meta, Amazon, and Apple reported earnings; combined 2026 CapEx tracking ~$720B. China blocks Meta's Manus acquisition. OpenAI reportedly missed growth targets. Pentagon signs deals with 8 AI firms — Anthropic kept off the list.

Pęknięcia w lustrze AI: Big Tech, OpenAI, Pentagon, Manus

Big Tech reports a record AI quarter with combined capital expenditures (CapEx) tracking around $720B for 2026. China blocks Meta’s $2B+ Manus acquisition. WSJ reports OpenAI missed growth targets. The Pentagon signs deals with eight AI firms — Anthropic kept off the list.

This is the AI news weekly recap for April 27 – May 3, 2026.

1. Microsoft Q3 — AI Business at a $37B Annual Run Rate, CapEx Jumps to $190B

On April 29, Microsoft reported fiscal Q3 2026 (quarter ended March 31): revenue $82.9B (+18% YoY), operating income $38.4B (+20%), EPS $4.27 (+23%). Azure and other cloud services +40% YoY. Strongest signal: Microsoft’s AI business is now running at a $37B annual revenue rate (+123% YoY).

Second signal: 2026 CapEx rises to roughly $190B — driven mainly by ballooning memory costs for GPU and data center buildout. Stock essentially flat after-hours — the market has priced in aggressive spend.

AI is generating real revenue, but $190B of CapEx in one year is unprecedented scale. The recurring question for the coming quarters: when does margin start feeling the investment pace.

2. Alphabet Q1 — Google Cloud +63% to $20B Quarterly

On April 29, Alphabet reported $110B revenue in Q1 2026 (+22% YoY). The engine: Google Cloud revenue grew 63% YoY to $20B in the quarter — more than doubling the prior year’s growth rate. Alphabet raised 2026 CapEx guidance to $180–190B (from $175–185B).

Alphabet stock +7% after hours — the strongest market reaction across the Magnificent 5 this week. Context: Anthropic just sealed up to $40B and 5 GW of TPU capacity from Google/Broadcom (last week).

Google’s edge is now distribution, in-house TPUs, and infrastructure partnerships. Cloud is growing faster than Azure, Anthropic is paying billions for TPU capacity, and Gemini is reportedly powering the new Siri.

3. China Blocks Meta’s $2B+ Manus Acquisition

On April 27, China’s National Development and Reform Commission (NDRC) announced it is blocking Meta’s $2B+ Manus acquisition — a Singapore-based AI startup with Chinese roots (deal announced December 2025). NDRC ordered both parties to unwind the transaction. Practical issue: Manus employees have already joined Meta’s AI teams; investors (Tencent, HongShan) have already received their cuts.

Bloomberg, April 29: the Manus model is effectively “dead” — Meta can’t develop it in China, and China won’t allow its export.

The first time China has hard-blocked a strategic AI acquisition by a US firm. Signal to the market: Chinese AI tech is under Beijing’s control even when the company is formally Singaporean. Future deals of this kind will be much harder.

4. Pentagon Signs Deals With 8 AI Firms — Anthropic Kept Off the List

On May 1, the US Department of Defense, also using the Department of War name, signed agreements with eight AI companies to deploy advanced models on classified military networks: SpaceX, OpenAI, Google, NVIDIA, Reflection AI, Microsoft, AWS, and Oracle. The initial communiqué covered seven firms; Oracle was added later the same day (Breaking Defense). Google will provide Gemini with adjusted safety settings.

Anthropic is not on the list — per Federal News Network, the company refused to grant “unrestricted access” to Claude for all lawful military uses without restrictions. Relations with the Pentagon have been strained recently (the US government ban from February — see our Feb 23 – Mar 1 recap).

The Pentagon goes for maximum model access, but the firm that most strongly emphasizes its own safety policies stays off the list. The price: no billion-dollar government contracts.

5. OpenAI Misses Growth Targets — WSJ Hits the Valuation

On April 28, the Wall Street Journal published a report based on internal OpenAI data: the company missed its end-of-2025 targets, including 1B weekly active users for ChatGPT and revenue goals. ChatGPT growth slowed in late 2025. According to the WSJ/Reuters report, CFO Sarah Friar reportedly voiced concerns about covering massive compute contracts. The board has questioned Sam Altman’s aggressive compute push.

Market reaction: Oracle, NVIDIA, and AMD stocks declined. Altman and Friar issued a joint statement: “We are totally aligned on buying as much compute as we can.” OpenAI called the report “clickbait,” but the market didn’t ignore it.

ChatGPT is losing consumer share to Gemini and ground in coding/enterprise to Claude. Meanwhile OpenAI is planning an IPO. The market will test either the valuation or the spending pace.

6. Apple Q2 — $111B Revenue, R&D +33%, A Different Bet on AI

On April 30, Apple reported fiscal Q2 2026 (quarter ended March 28): revenue $111.2B (+17% YoY), EPS $2.01 (+22%), gross margin 49.3%. Services hit a record $30.98B, iPhone $57B (+22%) — a March-quarter record.

Unlike the rest of the Magnificent 5: Apple isn’t aggressively scaling AI infrastructure CapEx. Instead: R&D +33% YoY to $11.42B. Tim Cook: “Collaboration with Google is going well” — per Bloomberg/Reuters reports, Gemini is set to power the new Siri in iOS 27. $100B buyback added.

7. Meta Q1 — CapEx Raised to $145B, Assured Robot Intelligence

On April 29, Meta reported revenue of $47B (+33% YoY), but the market focused on raised 2026 CapEx guidance: $125–145B (prior $115–135B). Stock down 6% after hours — the worst megacap reaction this week.

Separately, on May 1 Meta announced the acquisition of Assured Robot Intelligence — an AI robotics startup. Goal: strengthen Meta’s robotics and humanoid efforts.

8. Amazon Q1 — AWS Fastest Growth in 15 Quarters

On April 29, Amazon reported: AWS grew at its fastest rate in 15 quarters. Total revenue $159B (+13% YoY), AWS $30.4B (+22% YoY). 2026 CapEx held at ~$200B. AWS is closing the gap with Azure and Google Cloud — helped by the Anthropic deal (up to $25B from Amazon, $100B+ from Anthropic into AWS).

9. Microsoft Agent 365 — GA, Control Over AI Agent Sprawl

On May 1, Microsoft made Agent 365 generally available. Pricing: $15 per user per month (or bundled in Microsoft 365 E7). Three product pillars: observe, govern, secure — visibility, policies, and protection for an organization’s AI agent fleet.

Key feature: public preview of registry sync with AWS Bedrock and Google Cloud. IT can discover and inventory agents, and where the platform supports it, perform first governance actions, starting with agent deletion. Policy-based controls and runtime blocking — via Intune and Defender — in preview from June.

10. Tesla — Robotaxi Android App, but 5 Cities Pushed Back

On April 24, Tesla shipped its Robotaxi app on Android — a year after the iOS version (Electrek covered it on April 27). The app covers unsupervised service in Austin, Dallas, and Houston. Concurrently, Electrek reported on April 22 that Tesla is pushing back the Robotaxi launch in 5 additional cities (beyond the active 3) to H2 2026.

In the background: Waymo is now operating in 11 cities plus 20 announced (including two outside the US); it has been in Dallas and Houston since late February — about 7 weeks before Tesla.

11. OpenAI and Anthropic — Closed-Door Cyber Briefings on Capitol Hill

On April 28, Axios: OpenAI and Anthropic met behind closed doors with the House Homeland Security Committee on cyber threats. Topics: agentic attacks (CyberStrikeAI from March, the MCP RCE from last week), models capable of autonomously finding and exploiting vulnerabilities (Mythos and GPT-5.4-Cyber).

The briefings may suggest that federal cyber-AI regulation is gaining momentum in Congress — something states have been waiting for since 2024.

12. Big Tech Combined 2026 CapEx Reaches ~$720B

Post-earnings: combined 2026 CapEx guidance for the Magnificent 5 is roughly $720B ($705–735B depending on framing). Microsoft: $190B; Amazon: ~$200B; Alphabet: $180–190B; Meta: $125–145B; Apple: ~$10B. For context: the US FY2025 defense budget was on the order of $830–850B, depending on framing (DoD request, NDAA, total national defense outlays).

OpenAI CFO Sarah Friar publicly signaled difficulty covering compute contracts at the current growth pace — investors are asking the same question about the entire industry.

13. AI Chip Stocks Drop on the WSJ OpenAI Report

On April 28, hours after the WSJ report: NVIDIA, AMD, and Oracle all opened materially lower. Forward P/E: NVIDIA 25x, Oracle 22x, AMD 48x — the highest valuations in the hardware sector. Bear case: if OpenAI is missing annual targets, the market may question the scale of projected industry capital expenditures around $720B.

Bull case: it’s a correction, not a reversal. Microsoft and Alphabet are still buying everything Nvidia can produce, and TSMC raised guidance.

14. Funding Rounds and New AI Startups

April 30 – May 2 highlights:

  • Netomi — California-based startup automating customer service via generative AI — $110M Series C led by Accenture Ventures.
  • Legora (Stockholm, AI for legal) — $50M Series D extension with participation from Atlassian and NVIDIA NVentures, total round $600M at $5.5B+ valuation.
  • Parallel Web Systems (founded by Parag Agrawal, ex-Twitter CEO) — $100M from Sequoia, $230M total.
  • Shapes — exits stealth with an $8M seed round led by Lightspeed. App that drops labeled AI characters into ordinary group chats (WhatsApp, Discord, Telegram). 400K monthly active users already before the public announcement.

Signal: a consumer agentic product built mostly on entertainment and social interaction is scaling faster than B2B agent products at the same stage.

15. Amazon “Join the Chat” in Hear the Highlights

On April 28, Amazon launched “Join the chat” — a feature within Hear the highlights that lets users ask text or voice questions while listening to an AI-generated product summary. Available in the US shopping app.

A direct attempt to stem the migration of product search to ChatGPT and Perplexity.

16. Kakao Mobility — Level 4 Autonomous Driving Roadmap

On April 22, Kakao Mobility at the World IT Show unveiled a Level 4 (L4) autonomous driving roadmap built on machine learning, redundant vehicle architecture (safety-system redundancy), and simulation-based validation. The first major Korean entry in L4.

17. In Brief

  • MIT FTTE — a new framework that accelerates federated learning by 81% while cutting memory overhead 80% and communication payload 69% (April 29).
  • Microsoft Defender — new features for detecting attacks on AI agents (preview from June, announced April 30).
  • Tesla Cybercab — first production-line units rolled off in Austin in late April; ultimately destined for the robotaxi fleet.

Weekly Takeaway

The week of April 27 – May 3, 2026 was defined by the battle of numbers. Big Tech reported a record Q1 with AI as the lead growth engine — Azure +40%, Google Cloud +63%, AWS the fastest in 15 quarters. Combined 2026 CapEx is now around $720B — comparable in scale to the US defense budget. Apple chose a different course: instead of CapEx, R&D +33% and per reports Gemini under Siri in iOS 27.

Cracks are appearing in the background. WSJ reported OpenAI is missing growth targets — Oracle, NVIDIA, and AMD shares dropped. OpenAI’s CFO signaled internal tension. The Pentagon signed agreements with eight AI firms — without Anthropic, which refused to allow “unrestricted access” to Claude for all lawful military uses without restrictions. China blocked Meta’s Manus acquisition — Beijing’s first hard intervention in the cross-border AI market.

The third thread is enterprise maturity. Microsoft Agent 365 is GA — IT and security teams finally have a tool to govern an organization’s AI agent fleet, with cross-cloud sync. Closed-door briefings with OpenAI and Anthropic may suggest federal cyber-AI regulation is gaining momentum in Congress. Tesla is trying to scale Robotaxi faster than Waymo’s pace and regulators allow. The era of “AI as an experiment for everyone” is winding down — the era of “AI inside the organization,” with policies and KPIs, is starting.

Źródła / Sources:

MML Studio

Written by

MML Studio

Comments

Leave a comment

Comments are published after admin approval.

← Previous AI Monthly Recap April 2026 — Mythos, GPT-5.5, $65B Anthropic, Manus Blocked