ByteDance, the parent company of TikTok, has increased its planned AI infrastructure spending for 2026 by 25%, bringing the total budget to over 200 billion yuan ($29.4 billion), up from a preliminary plan of 160 billion yuan discussed late last year. The revision reflects both the company’s expanding AI ambitions and the impact of surging memory chip costs that have rippled across the global tech industry.
Rising Costs Drive Budget Overhaul
The budget increase comes as memory chip prices have skyrocketed industrywide. TrendForce reported DRAM contract prices rising roughly 95% quarter over quarter in Q1 2026, with a further 58% to 63% increase projected for Q2. ByteDance is not alone in facing these headwinds — Microsoft attributed $25 billion of its record $190 billion capex budget to higher memory and component costs, while Meta raised its full-year capex range to $125 billion to $145 billion, citing similar pressures. Collectively, major U.S. tech companies announced $725 billion in AI capital expenditure for 2026, up 77% from the prior year.
ByteDance had initially outlined the 160 billion yuan figure in December 2025, according to a Financial Times report at the time, with roughly half earmarked for advanced semiconductors and approximately 85 billion yuan specifically for AI processors. The company also planned to spend about 100 billion yuan on Nvidia chips alone, contingent on U.S. export approvals.
Domestic Chips Take Center Stage
A notable element of the revised budget is a proportionally larger allocation toward domestic AI chips, a strategic shift aimed at reducing ByteDance’s vulnerability to U.S. export controls and aligning with Beijing’s push for semiconductor self-sufficiency. In February, Reuters reported that ByteDance was developing its own AI inference chip in talks with Samsung for manufacturing, aiming to produce at least 100,000 units this year with plans to scale to 350,000.
ByteDance has also pursued offshore computing arrangements to access restricted chips. According to The Wall Street Journal, the company assembled over $2.5 billion in Nvidia Blackwell AI computing systems in Malaysia through a partnership with a Southeast Asian cloud provider, deploying approximately 36,000 B200 chips for AI research outside China.
Geopolitical Calculus
The dual strategy — investing in domestic silicon while building offshore capacity for cutting-edge foreign chips — underscores how U.S. export restrictions continue to reshape the AI supply chain. ByteDance’s chip design division now employs about 1,000 people and has developed a processor that reportedly rivals the efficiency of Nvidia’s China-market H20 chip at lower cost. As memory costs climb and geopolitical tensions persist, ByteDance’s willingness to spend nearly $30 billion signals a bet that AI infrastructure will remain central to its competitive position across short video, e-commerce, and enterprise cloud services.
