Author: Stephanie Palazzolo
Here are my core predictions for the artificial intelligence industry in 2026: ✅ Google will acquire Thinking Machines Lab. This year has seen the emergence of several so-called new-generation AI labs—small to medium-sized laboratories attempting to delve into AI research paths overlooked by major players. In 2026, such labs are likely to experience a wave of acquisitions. I predict that Thinking Machines Lab, a high-profile startup co-founded by former OpenAI CTO Mira Murati, will become one of the acquisition targets. The company has recently been in talks with investors for a funding round at a valuation of $50 billion, but the deal has not yet been finalized. The stalled financing process may indicate growing investor concern over high-valuation bets on AI labs; coupled with the fact that the company has only launched one product since its official debut in February, these signs point to a high probability of a merger or acquisition. Google could provide a stable "home" for this startup while also leveraging its top-tier AI R&D capabilities and core competitive advantages in the AI race to enhance the company's industry prestige. ✅ OpenAI will launch an automated AI research intern system in September 2026, but it will fall short of expectations. In October of this year, OpenAI's Chief Scientist, Jacob Pachocki, stated during a live stream that the company expects to launch an AI self-research system with the capabilities of a research intern by September of next year, and a more advanced AI R&D system by March 2028. I predict that OpenAI may launch a product branded as an "Automated AI Research Intern" next September, but the actual performance of this technology is another matter. It is noteworthy that OpenAI has set the first milestone for automated AI R&D as "research intern level," primarily because this definition is inherently vague. After all, human interns are allowed to make mistakes and require guidance from senior researchers, which also leaves considerable room for error for an AI technology bearing the same name. ✅ Leading AI labs will launch agent products with monthly fees exceeding $1,000. Remember when OpenAI reportedly planned to launch a doctoral-level agent for $20,000 per month? The developer of ChatGPT may have been too hasty. In 2025, the actual performance of these AI agents, designed to autonomously complete tasks on user devices, fell short of expectations. Several companies have since scaled back their projections for such products, and while pricing varies, the cost is far below $20,000. As AI technology improves in reliability for handling long-cycle tasks, the price of AI agent products from leading labs is expected to rise in 2026. The first high-priced product is anticipated to set the monthly fee threshold at $1,000. ✅ Google Gemini's weekly active users will catch up to OpenAI's ChatGPT. This is a bold prediction! But the prerequisite is crucial: If Google continues its conservative strategy, keeping core AI features like the Gemini chatbot and AI mode separate from its standard Google Search service, the challenge of catching up to ChatGPT's nearly 900 million weekly active users will be significantly greater (Gemini currently has 650 million monthly active users). Conversely, if Google decisively increases its efforts to aggressively promote AI mode and Gemini features to its Google Search user base, its user numbers could easily surpass those of ChatGPT. ✅ A breakthrough in continuous learning technology will trigger a sharp drop in Nvidia's stock price. This year, several top AI researchers have criticized the current mainstream methods of AI model training. Researchers, including former OpenAI Chief Scientist Ilya Sutskever, argue that existing technology cannot achieve superintelligent AI, with the core issue being the inability to learn in real-time from real-world scenarios like humans do (i.e., continuous learning technology). Current model development still requires massive amounts of data and computing resources, with training processes being time-consuming. Major labs are all working on breakthroughs in continuous learning technology. Once a significant advance is made in this area, it will be negative news for Nvidia. Continuous learning technology requires far less data and computing power than existing training models, which would directly lead to a sharp decline in market demand for Nvidia's blockbuster chips. Other Industry Highlights ✅ SoftBank has completed an additional $22.5 billion investment in OpenAI, thereby fully delivering on its $40 billion investment commitment. It now holds approximately 11% of OpenAI's equity. ✅ MiniMax, a leading Chinese large model startup, announced on Wednesday its plan to raise up to HK$4.19 billion (approximately $538 million) through an initial public offering. ✅ Elon Musk confirmed a report by The Information stating that his company xAI has purchased land outside Memphis, Tennessee, USA, to build a data center aimed at deploying 1 million chips to provide computing power for its AI models.
