Alibaba Falls 4% as Reported Beijing AI Probe Hits DeepSeek and Moonshot AI

Crusoe and Thinking Machines Lab have announced a $65 million annual agreement to power inference for the lab's open models on Crusoe Cloud. Under the deal, Thinking Machines Lab will run a range of production workloads, including its Inkling models, GLM 5.2 and 5.3, and its own fine-tuned variants, through Crusoe Managed Inference on a dedicated Tailored Deployment of NVIDIA HGX B200 systems connected with NVIDIA Quantum-2 InfiniBand networking. Crusoe will run and support the cluster directly as a Tailored Deployment, a dedicated, benchmarked, SLA-backed endpoint, giving Thinking Machines Lab the price-performance and headroom to scale without managing its own inference stack. The companies are also looking to expand into batch inference for large-scale synthetic data generation. Thinking Machines Lab joins a roster that has taken Crusoe Managed Inference past $100 million in contracted ARR less than a year from launch.

Alibaba is expanding across the AI stack, from its Qwen models and proprietary chips to cloud infrastructure and overseas markets, as it prepares for a larger role in the global AI race. The company is already working on Qwen 4, while future versions like Qwen 5 are designed to scale up to 10 trillion parameters, and the Qwen 3.8-Max model has gone through 33 rounds of recursive self-improvement. Alibaba unveiled its Zhenwu V900 accelerator, calling it China's most powerful AI accelerator, with commercial use expected in 2027. Alibaba Cloud is expanding internationally with new data centers planned in Turkey, Finland and the Netherlands, plus added capacity in Malaysia, Germany, the UAE, France and Hong Kong, building on its current 107 availability zones across 31 regions. According to Gartner, BABA was the world's fourth-largest cloud IaaS provider by revenues in 2025, behind Amazon, Microsoft and Google, and the largest in Asia-Pacific with a 22.5% regional market share, while AI-related products have maintained triple-digit year-over-year growth for 12 straight quarters and now account for more than one-third of Alibaba Cloud's external revenues. Alibaba is working through a RMB380 billion, three-year investment plan, with about RMB190 billion spent by the June quarter, and the Zacks Consensus Estimate for BABA's fiscal 2027 and 2028 EPS implies year-over-year growth of roughly 67% and 40%, respectively, though the stock remains a Zacks Rank #3 (Hold).

Meta Platforms has clawed back nearly 30% over the past month, outperforming the broader market ahead of its annual Connect event on September 23rd, after a summer in which the stock fell 13% in 2026 while the S&P 500 gained more than 10%. The second-quarter report showed revenue growing 28% to $60.8 billion, but earnings of $6.18 per share missed the roughly $7.10 consensus as capital expenditures hit $31.1 billion, up 57% from the first quarter and consuming about 98% of the company's $31.86 billion in operating cash flow, leaving free cash flow at $784 million. Management narrowed its 2026 capex guidance to a range of $130 billion to $145 billion, nearly double the $72.2 billion spent in 2025, took on another $24.9 billion of debt to bring long-term debt to $83.7 billion, and repurchased zero shares for the second straight quarter after buying back $26.3 billion in 2025. Earlier this month Meta launched Muse, a personal AI agent powered by its Muse Spark family of models, which reached as high as No. 3 in the U.S. App Store on its second day, and JPMorgan analyst Doug Anmuth wrote that Meta's Superintelligence Lab essentially delivered on its goal of reaching the frontier within a year with Muse Spark 1.3. JPMorgan upgraded the stock to Overweight from Neutral and raised its price target to $820 from $640, while advertising revenue grew 27% to $59.4 billion on 14% growth in ad impressions and a 12% increase in average price per ad, and CEO Mark Zuckerberg noted on the second-quarter call that Meta is receiving offers for its computing capacity at a significant premium to what the company paid for it.