China has launched a comprehensive open-source AI strategy that directly challenges the U.S. closed-source model dominance. On July 28, 2026, Xi Jinping outlined a three-pillar vision at the World AI Conference in Shanghai, emphasizing open-weight, open-source models, multilateral governance frameworks, and massive capacity-building in the Global South. According to reports, Chinese AI models now cost roughly one-tenth of U.S. models per million output tokens, making them the default affordable option for cash-constrained developing countries. The strategy leverages China's hardware and compute constraints to build an ecosystem rather than racing on frontier closed models, with China's World AI Cooperation Organization (WACO) now having 29 members including Brazil, Russia, Indonesia, and South Africa.
Google has fundamentally shifted its AI strategy away from competing directly with OpenAI and Anthropic in the same race for raw computational power. According to reports, Google's newest release, Gemini 3.6 Flash, landed 10th on one independent ranking when measured against other major AI labs. The company is now focusing on world models that understand and simulate the real world rather than AI that improves itself. This strategic pivot is evident in Google's recent product launches, with the Gemini 3.6 Flash model producing 17% fewer tokens than its predecessor, making it cheaper to deploy while maintaining competitive performance.
Google's DeepMind division is actively developing world models that learn how physical things behave, including gravity, motion, and cause-and-effect relationships. As reported, DeepMind's own website categorizes Genie 3 and Gemini Robotics under a heading for world models and embodied AI, meaning software that controls machines. The lab extended Project Genie to Street View in May and released SIMA 2, an agent that learns by playing inside virtual 3D worlds. This approach differs significantly from OpenAI and Anthropic's focus on recursive self-improvement (RSI), where AI builds the next generation of itself.
Despite Google's strategic pivot, investors remain concerned about the company's AI direction. According to reports, Alphabet shares fell 6% in June after two senior researchers left for rivals. However, Google maintains its market presence with Gemini app having 950 million monthly users. The company has also begun its biggest training run yet for Gemini 4 and continues testing Gemini 3.5 Pro with partners. Google's approach appears to be paying off in some areas, with the company leading the ML-Bench test that measures AI research skill, where a Gemini 3 model scored 64.4% in February.
Google's AI investments are creating significant financial strain on the company's cash flow. As reported, Alphabet's revenue reached $119.8 billion in the June quarter, up 24%, with search alone bringing in $63.3 billion. However, the company's capital expenditure on data centers and equipment reached $44.9 billion in three months, roughly double the previous year. This resulted in negative free cash flow of $5.86 billion, compared to positive $10.1 billion in March. The company covered this gap by selling $49.6 billion of new shares in June and borrowing $20.3 billion, with long-term debt doubling from $46.5 billion to $98.2 billion in six months.
