24-hour AI News Summary: OpenAI's self-developed inference chip releases real-world test data, further extending AI competition to chips, industry implementation, and commercialization
In the past 24 hours (August 25 to August 26, 2026), the AI industry's focus has shifted from simply competing in model capabilities to inference chips, enterprise agents, vertical industry applications, and commercial revenue. The semi-annual reports of Chinese AI companies have also sent clear signals: model call and enterprise service revenue is growing rapidly, and AI commercialization is entering a stage where revenue, cost, and implementation efficiency are being emphasized.
1. OpenAI released the first batch of test results for the Jalapeño inference chip
OpenAI claims its first self-developed inference chip, Jalapeño, achieves higher power throughput and lower latency in multiple tests, covering models such as DeepSeek R1 and Kimi K2.5. The company plans to deploy it to its own computing infrastructure by the end of the year, further clarifying the trend of AI giants developing their own chips.
2. MiniMax's revenue in the first half of the year increased by 283.1% year-on-year.
MiniMax announced revenue of $116.6 million for the first half of 2026, with open platform and enterprise AI services revenue up 703.1% year-on-year, accounting for 63.4% of total revenue. As of July, its token consumption had reached 20 times that of January, with enterprise-level model calls becoming a key growth driver.
3. Insilico Technologies' revenue surpassed $100 million in the first half of the year
AI pharmaceutical company Insilico Intelligence generated $106.3 million in revenue in the first half of the year, up 287.2% year-on-year, with net profit of $35.54 million, and operating cash flow turning positive. AI drug R&D is moving from the technology validation stage to large-scale commercial revenue.
4. The Ministry of Industry and Information Technology emphasizes promoting high-value scenarios of "artificial intelligence + manufacturing."
The Ministry of Industry and Information Technology stated that it will distill high-value AI application scenarios by industry and cultivate AI application service providers, launching lightweight, low-cost, and easy-to-deploy industry solutions. Policy focus is shifting from "large model capabilities" to the actual productivity of manufacturing.
5. Google launched Gemini Enterprise for the legal industry
Google has expanded Gemini Enterprise to launch AI capabilities specifically for the legal industry, which can be used for contract review, regulatory tracking, document analysis, and other tasks. Legal services are becoming a key vertical scenario for OpenAI, Anthropic, and Google competing in the enterprise AI market.
6. Claude connects chat and Cowork memory
Anthropic adds shared memory capabilities across chat and cowork for Claude, so users no longer need to repeatedly explain project backgrounds to AI. AI assistants are shifting from single-session Q&A tools to long-term collaborative agents that continuously understand work contexts.
7. The head of OpenAI's data center resigned
Chris Malone, head of OpenAI's data centers, has left the company, while the company is restructuring its infrastructure organization. As AI companies rapidly expand their investment in computing power, data center construction, chip procurement, and infrastructure management have become core competitive elements.
8. Stability AI completed a $76 million financing round
Stability AI announced the completion of a $76 million Series B funding round, with Sony Music, Universal Music, Warner Music, and EA participating. The relationship between generative AI companies and the traditional entertainment industry is gradually expanding from copyright conflicts to capital and product collaborations.
9. AI Agent startup Runable raised $21 million
Runable secured a new round of financing, extending its AI Agent capabilities from "helping users create websites and applications" to subsequent operations and growth. Agent's entrepreneurial direction is expanding from automated programming to more complete enterprise business processes.
10. AI startup Primero secured $12 million in seed round funding
Mexican AI company Primero completed a $12 million seed round aimed at providing AI solutions to large Latin American enterprises. Enterprise-level AI investment is further expanding from core markets such as the US and China to Latin America and other regions.
Frequently Asked Questions (Q&A)
Q: What are the most noteworthy technological changes in the past 24 hours?
A: OpenAI's self-developed Jalapeño chip has released test data, indicating that leading model companies are further mastering the complete technology stack from model to chip.
Q: What signals are emerging from the commercialization of AI in China?
A: MiniMax and Insilico Intelligent have both achieved rapid revenue growth, with enterprise APIs, industry solutions, and AI pharmaceuticals becoming measurable sources of revenue.
Q: What will be the next phase of enterprise AI competition?
A: The focus will shift from chatbots to Agents, long-term memory, as well as vertical business processes such as legal, manufacturing, and R&D.
Q: What trends do developers need to pay the most attention to?
A: The importance of inference costs and agent execution efficiency continues to rise. Beyond model capability, inference chips, contextual memory, and tool call capabilities will directly impact product competitiveness.
