Anthropic Posts First Quarterly Profit of $559M, DeepSeek Raises ¥70B, Valued at $45B

This week the AI sector displayed a clear pattern of simultaneous technical breakthroughs and commercial maturation, as evidenced by Google I/O's focus on the agent era, OpenAI extending ChatGPT into high‑sensitivity personal finance, and Anthropic becoming the first AI company to achieve quarterly profitability.

1. Qualcomm & OpenAI (May 16): Reported collaboration on an AI‑driven smartphone chip aimed at powering new AI consumer devices, with mass production targeted for 2028 (source: Reuters).

2. OpenAI (May 16): Launched a preview of ChatGPT Personal Finance for U.S. ChatGPT Pro users, allowing connection via Plaid to over 12,000 financial institutions for spending analysis, budgeting, and long‑term planning (source: TNW).

3. Turing Award winner Yoshua Bengio and hundreds of experts (May 16): Jointly released the "2026 International AI Safety Report" supported by more than 30 countries to provide scientific evidence for policymakers on AI risks and governance (source: TechOrange).

4. Google (May 19): At I/O 2026 unveiled Gemini 3.5 Flash, a large model that outperforms Gemini 3.1 Pro with four‑times the speed and over 50% cost reduction; also introduced Gemini Omni multimodal world model and Gemini Spark personal agent, with Gemini monthly active users surpassing 900 million (source: Google blog).

5. 2026 China Cyber‑Civilization Conference (May 19): Released the "AI Application Ethics and Safety Guidelines 1.0" to set safety baselines and value directions for AI development and services (source: China News Service/Guangxi Science & Technology Department).

6. Google (May 20): Announced over 100 AI‑related products and technologies at I/O, including Gemini 3.5 Flash, Gemini Omni, Gemini Spark, Google Antigravity 2.0 agent platform, AI Search with over 1 billion MAU, a $100/month Google AI Ultra subscription, and a free AI Pro tier with YouTube Premium Lite (source: Google blog).

7. The Information (May 21): Reported OpenAI Q1 revenue of $5.7 billion driven by Codex coding assistant, enterprise sales, and ChatGPT ad testing; Anthropic revenue reached $4.8 billion with annualized revenue near $45 billion, surpassing OpenAI's previously announced $25 billion annualized figure (source: PYMNTS.com).

8. Microsoft (May 21): Launched its self‑developed AI chip Maia 200, optimized for large‑model inference, reducing per‑unit compute cost by 15% versus Nvidia A100 and 10% versus Nvidia H100, marking a step toward AI hardware autonomy for cloud providers (source: Ti Media).

9. Anthropic (May 22): Told investors Q2 revenue is expected to hit $10.9 billion, a 130% quarter‑over‑quarter increase, and to generate $55.9 million operating profit—the first quarterly profit since its founding, achieved two years ahead of schedule, driven largely by enterprise adoption of Claude Code (source: 36 Kr).

10. Anthropic (May 22): Negotiating with Microsoft to use the second‑generation Maia 200 chip to further cut inference costs; employs a multi‑chip strategy using Nvidia GPUs, Amazon Trainium, Google TPUs, and SpaceX Colossus clusters, and has pledged at least $330 billion in compute spend to Amazon, Google, and Microsoft, with per‑dollar compute cost projected to drop from $0.71 to $0.56 (source: 36 Kr).

11. DeepSeek (May 22): Reported by Bloomberg to be raising roughly ¥70 billion in financing, with a pre‑money valuation of about $45 billion; founder Liang Wenfeng pledged to keep focusing on open‑source AI models rather than short‑term monetization (source: IT Home).

12. Alibaba (May 22): Its voice large model Fun‑Realtime‑ASR and Fun‑Realtime‑AudioChat topped the global AI evaluation platform Artificial Analysis, surpassing GPT‑Realtime‑2 on word error rate, speech inference, and dialogue fluency (source: Phoenix Net).

13. Zhipu AI (May 22): Released GLM‑5.1 High‑Speed model, achieving a global fastest inference speed of 400 tokens/s, marking a breakthrough for domestic models in high‑performance inference (source: China Steel Network/Gold Ten Data).

These developments illustrate the AI industry's shift from pure cash‑burn growth to scalable commercialization, with increasing emphasis on cost‑effective hardware, multimodal capabilities, and enterprise‑grade services.