AI dominance

Anthropic’s Super Bowl ad, criticizing OpenAI’s ChatGPT ads, appears to have boosted Anthropic’s brand significantly. Post-game analysis shows Anthropic experienced a 6.5% surge in website traffic and an 11% jump in daily active users, propelling Claude into the App Store’s top 10. This ad battle intensifies the competition between Anthropic and OpenAI as both companies reportedly prepare for potential IPOs.

Tech giants are heavily investing in social media influencers to promote AI services, driving a significant increase in creator marketing. Companies are offering substantial payouts for sponsored content showcasing AI tools, fueling an advertising war for AI adoption. While many creators embrace these lucrative partnerships, some decline due to ethical concerns about AI’s impact on livelihoods, particularly regarding generative image and video tools.

DeepSeek’s initial emergence last year significantly disrupted AI markets, causing temporary stock drops for major Western tech firms by challenging U.S. dominance. While DeepSeek has since released model updates, they haven’t matched the initial shockwave. Factors like incremental improvements, compute limitations due to geopolitical influences, and continued rapid advancements from Western AI labs have softened the market’s reaction. However, ongoing research from DeepSeek suggests potential for future disruptive moments in the AI race.

The US-China AI chip battle escalates as Nvidia’s restricted H200 and B200 GPUs enter China via shadow networks during a 90-day tariff reprieve, fueling a volatile black market. Cloud giants face acute scarcity, while emerging hybrid supply chains disguise GPUs as industrial goods. Structural contradictions emerge: despite speculative bubbles and unviable projects, specialized AI adoption grows, exposing systemic bottlenecks in technical innovation, data readiness, and vertical integration. Government subsidies clash with industry demands for foundational ecosystem reforms as companies pivot to VC-driven compute models amid shifting demand from pre-training to inference workloads.