ByteDance’s CEO Liang Rubo opened 2026 with a crisp new directive: the company must “climb the highest peaks” of the AI era. He set a short-term summit — the company’s consumer assistant, Dola (豆包/Dola) — as the immediate focus and ordered a sustained, global hiring and compensation push to ensure model capability ranks at the industry front.
Liang framed AI as a tectonic shift on the scale of the PC and internet revolutions, but warned that the time window to dominate a new platform peak is shrinking. The company’s calculus is explicit: superior models plus integrated assistant experiences could elevate an already leading content and distribution engine into a new platform layer, but speed and scale matter far more than in past transitions.
On the product front Liang argued that model capability must lead, while the assistant stitches together ByteDance’s existing services. He also flagged business-to-business opportunities via MaaS (model-as-a-service), arguing that only by winning in both consumer-facing assistants and enterprise model services can a firm push the intelligence ceiling higher.
To back that ambition, ByteDance is increasing pay and incentives globally. Internal measures announced in late 2025 expanded bonus and equity pools, raised budgeted salary adjustments by roughly 1.5x over the prior cycle, and lifted pay-band floors and ceilings — moves designed to keep compensation “ahead of peers” in major markets.
Rivalries are already playing out up and down the stack. Tencent has begun seeding AI into social with its “Yuanbao” project and plans a deep-pocketed red-envelope promotion during the Lunar New Year. Alibaba is pushing hard on both software and silicon: its large model effort and the consumer-facing Qianwen app have registered rapid take-up, while its chip unit Pingtouge has unveiled the “Zhenwu 810E” PPU and an AI “golden triangle” with Tongyi Lab and Alibaba Cloud.
The competition extends into infrastructure: Alibaba says Zhenwu is an all‑in‑one training and inference chip with HBM2e memory and high‑bandwidth inter‑chip links, already deployed at scale inside Alibaba Cloud. Baidu’s Kunlun chip business has taken the next step toward a public listing, touting tens of thousands of card deployments. ByteDance, meanwhile, is exploring hardware entry points: a second‑generation Dola phone is expected in mid‑2026 and its Volcano Engine business has been named an exclusive AI cloud partner for China’s Spring Festival Gala.
What is unfolding is not merely an application war for user attention but a systemic contest over model quality, compute architecture, hardware supply chains, and public trust. The winners will control the inexpensive, high‑utility interfaces through which billions access AI and will capture disproportionate returns from downstream monetisation and enterprise contracts.
For global observers, the message is clear: China’s largest tech firms are racing simultaneously on product, compute and chip fronts, compressing a decade‑long sequence of platform consolidation into months. The strategic window is narrowing, and firms that fail to synchronise model innovation, hardware scale and regulatory trust will cede outsized ground to those that do.