Amazon CEO Andy Jassy: Young People Must Use AI Extensively but Retain Independent Thinking
Amazon CEO Andy Jassy stated in the a16z podcast that young people must extensively use AI but should never abandon critical reasoning and independent thinking. Universities should not oppose AI, as this trend is irreversible and cannot be avoided or expected to disappear.
He has two children, aged 25 and 22, who are not developers but have created applications using AI and have made AI a core part of their daily work and tools. He emphasized that AI will have a huge impact on life and every work experience, enabling unprecedented ways of invention and creation, thus requiring proactive learning, extensive use, and attempts to step outside of comfort zones.
Jassy cited business school cases, noting that the need to analyze three cases daily, read extensively, take notes, and distill key issues will significantly decrease or even disappear in the AI era. AI can help everyone reach a higher starting point faster, but connecting information and clues, formulating strategies and plans, and developing new products and customer experiences will still largely require human effort.
He pointed out that human judgment, originality, innovation, experimental spirit, and the ability to combine various parts remain indispensable. Tools will provide significant assistance, but critical reasoning and thinking will be crucial in all work, as AI-generated content often can be identified as "not accurately pieced together."
In the same podcast episode, Jassy also discussed organizational changes: AI will reduce team sizes, with models capable of completing about 80% of basic tasks, allowing cross-functional small teams to build from scratch faster rather than through incremental improvements. Amazon itself is rewriting many applications from scratch and views cybersecurity as an underestimated risk in the AI era.
Funding is flowing massively into AI infrastructure and model laboratories: Amazon plans to invest about $220 billion in capital expenditures by 2026, mainly for data centers and custom chips. AWS's AI business has reached about $25 billion in annualized revenue within three years, far exceeding early AWS growth rates. Beneficiaries include computing power providers, chip suppliers, and model companies receiving investments, while companies and positions still relying heavily on specialized human labor and failing to reduce team sizes with AI are under pressure.
Amazon is developing self-researched chips like Graviton and Trainium through Annapurna Labs and has made significant investments in Anthropic and OpenAI to secure demand. The cost of inference is still seen as needing to be reduced by an order of magnitude to fully penetrate existing enterprise workloads.
Source: Public Information
ABAB AI Insight
Andy Jassy joined Amazon in 1997, starting as an assistant product manager, leading the establishment and long-term leadership of AWS, transforming it from early annual revenues of tens of millions to a cloud computing leader. In 2015, he led the acquisition of Annapurna Labs, initiating a path for self-researched chips to reduce reliance on high-priced hardware like Intel. After becoming CEO in 2021, he implemented large-scale layoffs and cost controls to bring the company back to efficiency after over-expansion during the pandemic.
On the capital front, he pushed Amazon's capital expenditures for 2026 to about $220 billion, mainly directed towards data centers, networks, and self-researched AI chips, while also increasing investments in Anthropic and signing large cloud service agreements with OpenAI to bind front-end model demand back to AWS computing power sales. The motivation is to avoid repeating Microsoft's early monopoly on OpenAI cloud contracts by locking in long-term revenue through "infrastructure + model equity" dual lines, and to leverage self-researched Trainium to achieve a 30%-40% price-performance advantage to lower inference costs.
This is analogous to how AWS disrupted traditional IT outsourcing and local data centers in the 2000s with self-built infrastructure, echoing the strategy of acquiring Annapurna to counter Intel. The current phase involves parallel expansion and control: rapidly increasing supply on the computing power side, compressing team sizes with AI on the organizational side, forcing applications to be rewritten from scratch, while locking in large downstream customers through investments.
Essentially, this represents a combination of technological substitution and industrial chain reconstruction: AI replaces the foundational analysis, coding, and documentation work that previously required a lot of specialized human labor, raising the starting point, but leaving higher-level judgment and integration capabilities in human hands. The mechanism at play is that declining computing costs and improved model capabilities have lowered the barriers to rewriting applications, forcing companies to shift capital from human redundancy to infrastructure and custom silicon, or risk falling behind competitors that have completed their restructuring in pricing power and delivery speed.
ABAB News · Cognitive Laws
- Tools raise the starting point, judgment determines the endpoint.
- Irreversible technology; opposition equals self-elimination.
- Scale dividends come from compressing labor, not piling on labor.