Squawk Pod: Davos 2026 – Interview with Amazon CEO Andy Jassy
Episode Date: January 21, 2026
Host/Interviewer: Becky Quick (CNBC)
Main Guest: Andy Jassy (CEO, Amazon)
Recorded At: World Economic Forum, Davos, Switzerland
Episode Overview
This episode features an in-depth, wide-ranging conversation between CNBC's Becky Quick and Amazon CEO Andy Jassy, recorded during Davos 2026. The discussion covers Amazon's scale and growth, leadership priorities, the company’s approach to culture, the AI and semiconductor arms race, infrastructure and power constraints, tariff impacts, jobs and automation, and Amazon's current AI initiatives. Jassy offers candid insights into Amazon's strategy, its efforts to control costs in supply chain turbulence, and the ways the company is balancing rapid technological progress with organizational agility.
Key Discussion Points and Insights
1. Amazon Today: Scale and Business Priorities
(03:41–05:30)
- Amazon's market cap: ~$2.5 trillion, 1.5 million employees, second-largest private employer globally.
- 2024 Revenues: $638 billion. AWS annual run rate: $132 billion (up from $4.8B eleven years ago).
- New business areas: Prime Video, Amazon Pharmacy, space (Amazon Leo satellites), and deep AI projects.
- Jassy’s Time Allocation:
- Existing large business units (retail, AWS, Ads).
- New initiatives and businesses (Prime Video, healthcare, satellites, Zoox, AI across all domains).
- Company culture—keeping Amazon nimble even as it scales.
- Quote (Andy Jassy, 05:30):
“We want to operate like the world’s largest startup…and it takes work and energy.”
2. Culture as a Strategic Imperative
(07:29–09:39)
- Jassy discusses the need to “flatten layers” as Amazon grows, aiming for speed and ownership in decision-making.
- Regular meetings of a “Culture Club” to address challenges and keep the company nimble.
- Identified meeting proliferation (pre-meetings for pre-meetings), slowing agility.
- Result: Mandated a 15% higher ratio of individual contributors to managers across Amazon.
- Quote (Andy Jassy, 09:03):
“If you want to move like the world’s largest startup, you need to be as lean as you can be…as few layers as you can possibly have.”
3. Shifts in AI Leadership and Silicon Strategy
(09:39–12:08)
- Recent leadership changes: Rohit Prashad (AI lead) departs, Peter Desantis steps in, unifying custom silicon and frontier model teams.
- Importance of chip/model co-optimization for future AI.
- Desantis’s deep technical pedigree, history with AWS, and chip/infrastructure focus.
- Quote (Andy Jassy, 11:41):
“We have conviction that over time, the very best frontier models are going to have to work very closely with the chip in which they operate on.”
4. Amazon’s Custom Chips and Nvidia Dynamics
(12:08–14:07)
- Amazon’s chip strategy: Annapurna acquisition → Graviton CPUs (40% better price/performance than other x86 CPUs).
- Trainium AI chips: Now on third generation, each ~40% more price performant than the previous.
- Multi-billion-dollar silicon business; major clients include Anthropic for next-gen Claude model training.
- Still a key Nvidia partner, but custom silicon aids competitive pricing and margins.
- Quote (Andy Jassy, 14:05):
“If you’re building a big inference business like we are and you want to have reasonable margins, if you’re not pursuing your own custom AI silicon, you’re going to be structurally disadvantaged.”
5. The “Circular Deal” Era and Power-Hungry AI
(14:19–17:33)
- Massive demand for compute: AI labs consuming “gobs and gobs” of compute and power, driving unconventional investments.
- Discussion of high-profile partnerships (e.g., OpenAI’s AWS deals), the unprecedented capital and infrastructure buildout.
- Uncertainty over sustainability and exact details of some deals.
- Quote (Andy Jassy, 14:54):
“It is so unprecedented how much compute is being consumed right now...The AI labs believe they have opportunities if they train models very expansively…They need a lot of compute, they need money for the compute.” - Quote (Becky Quick, 16:43):
“OpenAI, for instance...signed deals to build out $1.4 trillion in infrastructure just in the last several months. That’s pretty ambitious.”
6. Infrastructure, Power, and Sustainability in the AI Race
(17:33–20:09)
- Amazon’s aggressive moves to secure resources: copper (Rio Tinto deal), electricity, even unique nuclear investments (SMRs).
- Largest corporate purchaser of renewables for five years. Also pursues significant water replenishment.
- Ongoing challenge: Global power shortages, especially for data centers.
- Quote (Andy Jassy, 17:55):
“Power has been short in the US; around the world, there is a power shortage... while it's better than 18 months ago, we still could be fulfilling more demand if we had more capacity.” - Amazon pays its own way for power infrastructure and invests in community training (electricians, data center operators).
7. Tariffs, Trade, and Consumer Prices
(20:09–22:39)
- Amazon’s response to 2025 tariffs: heavy pre-buying, forward-staging by Amazon and 3rd party sellers to delay price increases.
- By late 2025, stockpiled supply ran out, tariffs started affecting prices, but strategies differ by seller (pass to consumer or absorb).
- Amazon’s business scale means millions of items, 2M sellers, and diverse strategies.
- Quote (Andy Jassy, 20:36):
“Consumers have been pretty resilient...they are trying to trade down in price. They are looking for bargains wherever they can find them.” - Retail operating margins are thin; limited options for permanent cushioning of future tariff shocks.
8. Jobs, AI, and the Shape of Amazon’s Workforce
(23:23–25:44)
- Jassy revisits previous comments about AI’s likely net reduction in jobs (contrasting Jamie Dimon's optimism), but frames current layoffs as culture/speed-driven, not AI-driven.
- Predicts the nature of jobs will change; rote work will diminish, new roles (e.g., in AI, robotics, space, healthcare) will emerge.
- Historical context: jobs like “cloud architect” didn’t exist 15 years ago—future holds similar role creation.
- Quote (Andy Jassy, 25:17):
“I don’t think [AI] wipes out all those jobs...but I do think you won’t need as many people in all those jobs we’ve thrown humans at for the last 20 years.”
9. Amazon’s Real-world AI: From Shopping to Agents to Cloud
(25:44–29:48)
- Over 1,000 AI projects or applications active or in development.
- For consumers:
- “Rufus” AI shopping assistant now much improved (“actually quite good” per Jassy and Quick).
- Smarter Alexa, AI-driven advertising creative, improved Prime Video features.
- For AWS/business customers:
- Bedrock offers a broad suite of foundational models.
- “Novaforge” tool for early integration of customers’ proprietary data into models.
- Growth of AI agents: coding (Curo), autonomous agents for source code management and bug fixing.
- AI’s major impact on “cost of inference” (linked back to custom silicon story).
10. Davos: Goals and Early Impressions
(29:48–30:39)
- What Jassy hopes for at Davos: Insight gathering, government relationships, listening to leaders’ needs (business and geopolitical), and “stay warm.”
- Early learning at the Forum: “It’s cold outside.”
Notable Quotes & Memorable Moments (with Timestamps)
-
On Operating Like a Startup
“We want to operate like the world’s largest startup…and it takes work and energy.”
— Andy Jassy (05:30) -
On Flattening Organization Structure
“If you want to move like the world’s largest startup, you need to be as lean as you can be…as few layers as you can possibly have.”
— Andy Jassy (09:03) -
On AI’s Demand for Power
“Power has been short in the US; around the world, there is a power shortage... while it's better than 18 months ago, we still could be fulfilling more demand if we had more capacity.”
— Andy Jassy (17:55) -
On Custom AI Silicon
“If you’re building a big inference business like we are and you want to have reasonable margins, if you’re not pursuing your own custom AI silicon, you’re going to be structurally disadvantaged.”
— Andy Jassy (14:05) -
On AI’s Impact on Jobs
“I don’t think [AI] wipes out all those jobs...but I do think you won’t need as many people in all those jobs we’ve thrown humans at for the last 20 years.”
— Andy Jassy (25:17) -
On Rufus AI Shopping Assistant
“Rufus has gotten much, much better over the last several months. It’s actually quite good.”
— Andy Jassy (26:48)
Important Timestamps by Segment
| Segment | Time | |-----------------------------------------|----------| | Amazon scale & Jassy’s priorities | 03:41–07:29 | | Company culture & “Culture Club” | 07:29–09:39 | | AI leadership shifts & chip strategy | 09:39–12:08 | | Custom silicon vs. Nvidia | 12:08–14:19 | | Circular deals & AI compute boom | 14:19–17:33 | | Power shortages & infrastructure | 17:33–20:09 | | Tariffs: impact & consumer prices | 20:09–22:39 | | Jobs and AI's workforce impact | 23:23–25:44 | | AI applications (Rufus, agents, Bedrock)| 25:44–29:48 | | Davos goals & final thoughts | 29:48–30:46 |
Summary for the Uninitiated
This episode is an essential listen for anyone interested in the intersection of big tech, AI, geopolitics, and organizational agility at massive scale. Andy Jassy provides rare clarity on how Amazon is building for the future: flattening org structures, betting big on custom chips, racing to keep up with AI’s infrastructure demands, preparing for changing job dynamics, and deploying AI deep into retail and cloud. Quick’s questions are sharp, extracting plenty of unscripted color on headline topics from tariffs to OpenAI’s wild infrastructure bets.
Tone:
Engaged, candid, and analytic, with both interviewer (Becky Quick) and guest (Andy Jassy) focused on substance, scale, and the operational nuances of leading a global megacorporation through unprecedented times.