Long-termism + Customer Obsession + Cultural Resilience: Reflections on Reading "Amazon's Letter to Shareholders"

I recently finished reading Andy Jassy's 2025 letter to shareholders, which was just released on April 9th. It took half an hour to read and was packed with substance. The style continues the pragmatic, data-driven approach of the Bezos era, but Jassy adds a lot of personal reflection and a strong sense of urgency regarding AI as a "once-in-a-generation opportunity."

1. "Squiggly Lines" Are the Real Path; Don't Believe in Straight-Line Success

Jassy begins by talking about how he wanted to be a sports commentator after college but ended up taking a winding path through production, coaching, retail, product management, a failed startup, sales, and investment banking before finally joining Amazon in 1997. He uses the band The Beths' album "Straight Line Was a Lie" to make his point: careers and corporate development are never straight lines; they are full of stagnation, restarts, and detours.

It was the same in the early days of AWS. The initial vision included storage, computing, payments, and even "human intelligence." Many ideas died, but the survivors, S3 and EC2, changed the world.

True long-termism isn't about constant, uninterrupted progress; it's about tolerating ambiguity, experimenting rapidly, and correcting course in time. Amazon's culture must have "truth-tellers," the ability to withstand criticism, a flat organization, and the agility of a startup team. This is the underlying logic of how Amazon has continuously reinvented itself from e-commerce to cloud to AI. Many companies, including crypto projects, get arrogant when things go smoothly, but when they hit a detour, they engage in infighting or give up, destined not to go far.

2. Customer Obsession + Bold Bets on AI Are the Real Moat

The data in the letter is still quite solid: 2025 total revenue of $717 billion, up 12%; AWS revenue of $129 billion, up +20%; Q4 run rate of $142 billion; operating profit of $80 billion. However, free cash flow dropped from $38 billion to $11 billion, mainly due to ~$200 billion in capital expenditures in 2026, most of which is invested in AI infrastructure.

Regarding AI, AWS AI revenue run rate for Q1 2026 has already exceeded $15 billion; in-house chip business run rate exceeds $20 billion, which could reach $50 billion if sold independently to third parties. Trainium3/4 offer 30-40% better price-performance than GPUs. Bedrock inference primarily runs on their own Trainium chips, saving tens of billions in annual capex and improving profit margins. Jassy states bluntly: "We are not investing $200 billion on a hunch; we have substantial customer commitments (OpenAI alone is over $100 billion) and will be able to monetize in 2027-2028."

This is not burning money; it's a high-certainty, long-term investment.

Amazon sees AI as a "once-in-a-generation technology inflection point," just as AWS disrupted IT back then. Now, it aims to reshape retail, logistics, advertising, video, and even AWS itself.

The key is not chasing trends but reconstructing products from first principles, all centered around "making customers happier, cheaper, and faster." This is the sustainable moat.

$Amazon(AMZN.US) 

3. Culture and Execution Determine Everything

Jassy repeatedly emphasizes "tiny teams," "ownership mentality," and "invent and simplify." Amazon's ability to withstand criticism and restart quickly, like the Bedrock inference engine being rebuilt by 6 engineers in 76 days, relies on this culture.

Facing AI, they proactively build chips, expand 3.9GW of power capacity, and place parallel bets on robots, drones, and satellites. The letter is essentially Jassy "preaching" to Wall Street and employees: the short-term drop in free cash flow doesn't matter; long-term operating income and free cash flow will be much larger because of AI.

Just one day before Jassy released his letter, CZ's memoir came out, and the OKX CEO directly attacked him on X, dredging up old accounts from OKCoin in 2014-2015 and calling CZ a "habitual liar." CZ immediately counterattacked, even throwing out a "$100 million bet" to prove he was already divorced. The two went back and forth, posting chat logs and old contracts, with the entire crypto community watching.

This forms a stark contrast with Jassy's letter. Amazon hones its internal skills amidst "detours," makes heavy bets, develops its own chips, secures customer commitments, and grows the AWS AI run rate to over $15 billion and the chip business run rate to over $20 billion—all focused on long-term value and customers.

The two giants of the crypto world, however, spend their energy on personal grudges, contract disputes, and attacks on private lives from over a decade ago. Publicly throwing around a $100 million bet—this is not competition; it's emotional infighting.

And the result? Users see "exchange bosses are still doing this," and trust is further eroded.

What the crypto industry lacks most is precisely what Jassy repeatedly emphasizes: long-termism + customer obsession + cultural resilience. Over 28 years, Amazon has survived the dotcom bubble and been written off multiple times, but each time it survived and grew stronger by relying on "squiggly lines" and putting customers first.

The crypto world is now obsessed with "narratives" and "founder personas," but those who truly survive will ultimately have to return to product, compliance, security, and user experience.

If the leaders of BN and OKX spent more time making AI/on-chain settlement/institutional-grade risk control better instead of battling on X, the industry could mature faster.

Jassy says at the end of the letter: "We are very optimistic about the future." I feel the same way, not about any particular coin or project, but about those companies that truly turn "detours" into moats.

At a technology inflection point, conservatism is the greatest risk; but emotional infighting is the real waste.