ByteDance Builds Own AI Chips as Intel and AMD Prices Soar 2026

ByteDance, the parent company of TikTok, is developing its own processors to support its artificial intelligence infrastructure, ending its long-standing reliance on Intel and AMD for data center chips. The move comes as Intel and AMD have raised prices by 10% to 35% quarter-over-quarter, making external sourcing increasingly unsustainable for the Chinese tech giant.

According to Reuters, ByteDance plans to invest approximately $22.8 billion in AI infrastructure during 2026. The combination of supply shortages and steep price hikes has pushed the company to take chip development in-house, a strategic shift that could reshape how major tech firms approach AI hardware.

Why ByteDance Is Ditching Intel and AMD

The decision is not solely about cost. The AI industry is undergoing a structural change as it moves from the era of massive model training—dominated by Nvidia GPUs—into the age of inference. This workload places far greater demands on CPUs, which work in tandem with GPUs and have become the new bottleneck in AI processing.

Inference requires continuous, low-latency processing of AI models in production, and CPUs handle much of the orchestration and data movement. As a result, demand for high-performance CPUs has surged, creating a shortage that Intel and AMD have struggled to alleviate.

ByteDance's custom chips aim to address both the supply constraint and the cost pressure. By designing its own processors, the company can optimize them for its specific AI workloads, potentially achieving better performance and efficiency than off-the-shelf solutions.

The Price Hikes: A Breaking Point for Tech Giants

Intel and AMD's repeated price increases have hit large-scale buyers hard. For a company like ByteDance, which operates massive data centers to support TikTok's recommendation algorithms and other AI-driven services, even a 10% increase translates into hundreds of millions in additional costs.

The table below summarizes the key factors driving ByteDance's decision:

These increases are not marginal. They reflect a broader trend of rising semiconductor costs driven by manufacturing constraints and soaring demand for AI-capable hardware.

What This Means for the AI Chip Market

ByteDance's move is part of a larger wave of tech companies designing their own silicon. Google, Amazon, and Microsoft have all invested in custom AI chips to reduce dependency on Nvidia and Intel. If ByteDance succeeds, it could inspire other large-scale AI consumers to follow suit.

For Intel and AMD, the loss of a major customer like ByteDance is a significant blow. Both companies are already facing intense competition from Nvidia in the AI space and from ARM-based designs in the CPU market. Losing volume from a high-profile client could further pressure their data center revenues.

However, the custom chip route is not without challenges. Designing processors requires deep expertise, significant R&D investment, and access to advanced manufacturing—areas where ByteDance has limited experience. The company will need to partner with foundries like TSMC or Samsung to produce its designs.

Key Takeaways for Industry Watchers

• Cost pressure is real: Intel and AMD's price hikes are forcing major buyers to reconsider their supply chains.
• Inference changes the game: The shift from AI training to inference increases CPU demand, creating shortages.
• Custom silicon is trending: ByteDance joins other tech giants in developing in-house chips to reduce reliance on external vendors.
• Supply chain resilience: Companies are prioritizing control over their hardware stack to avoid future disruptions.
• Competitive landscape shifts: Intel and AMD may need to adjust pricing and allocation strategies to retain large customers.

What to Watch Next

ByteDance's chip development is still in its early stages, and details about performance and manufacturing partners remain scarce. The company has not announced a timeline for deployment, but given its $22.8 billion AI investment plan for 2026, progress is likely to accelerate. The background to this story is set out in AMD Set to Overtake Intel in Revenue for the First Time. A related thread runs through India Plans Domestic GPU in 24 Months to Boost AI Chips. Earlier coverage in Ferrero's Responsible Sourcing: Tech-Driven Sustainability at Scale shows how the argument developed.

Investors should monitor Intel and AMD's data center revenue trends for signs of customer attrition. Any indication that other large buyers are following ByteDance's lead could trigger a reevaluation of growth prospects for both chipmakers. A closer look at the same subject appears in TOTO's $495M Bet on 1nm Chips: The Hidden AI Giant. Anyone following the wider picture can start with Australian Industry Gas Reliance to Continue for Foreseeable Future. The same theme surfaces in Anthropic Builds Custom AI Chips to Power Claude and Cut Costs, which takes a different angle.

For the broader tech industry, ByteDance's decision underscores a fundamental shift: as AI becomes central to business operations, controlling the underlying hardware is becoming a strategic imperative. Companies that can design custom chips tailored to their workloads will have a competitive edge in cost, performance, and supply chain security. Context on how this played out before sits in How Ferrero is rethinking responsible sourcing at scale. A companion piece, University of Bristol's £500m Temple Quarter Campus Opens, approaches the same subject from another angle.