One year after Chinese startup DeepSeek rattled the AI industry, the fallout looks less like a collapse and more like a reset that ultimately fueled the next leg of the boom. What initially erased nearly $1 trillion in market value ended up reshaping how leading AI companies build and deploy models.
The $6 million model that changed the playbook
DeepSeek turned heads by training its DeepSeek-R1 model on a modest cluster of older Nvidia $NVDA ( ▲ 3.08% ) H800 GPUs, reportedly spending just a fraction of what US tech giants were pouring into massive AI training runs. Instead of brute force scaling, it leaned on a “mixture of experts” approach, where smaller specialized models work together, and still managed to outperform some frontier models in key benchmarks.
That raised uncomfortable questions across Silicon Valley. If DeepSeek could compete without tens of thousands of top-tier GPUs and giant data centers, were companies overspending on infrastructure?
Markets reacted violently. Nvidia, Broadcom $AVGO ( ▼ 1.74% ) , and Google $GOOGL ( ▼ 2.08% )
all saw sharp selloffs as investors questioned whether AI hardware demand might slow.
From panic to pivot
Rather than pull back, most big players adapted. Microsoft $MSFT leaned into the idea that cheaper AI could drive more usage, not less. OpenAI doubled down on reasoning-focused models and later moved toward more open-weight releases. Google reoriented Gemini around similar reasoning capabilities, and Elon Musk’s xAI followed suit with its own reasoning push.
At Meta, the shock reportedly sparked internal urgency to rethink its model roadmap. The broader industry began embracing more efficient architectures and smaller, specialized models alongside giant frontier systems.
Markets recovered and then some
The selloff did not last. Tech stocks rebounded quickly and then surged to new highs. Since the DeepSeek scare, Nvidia has climbed sharply, Broadcom has posted strong gains, and Google has moved well above its pre-shock levels. OpenAI’s valuation has also ballooned as investor appetite for AI exposure intensified.
In hindsight, DeepSeek’s breakthrough did not derail the AI trade. It forced a strategic rethink, accelerated innovation, and reminded the industry that smarter models are not just about throwing more hardware at the problem.