Amazon lost $170 million on a phone nobody wanted. Now it’s trying again—this time betting $200 billion on AI infrastructure to prove the Fire Phone disaster was just bad timing, not bad strategy. Reuters reported March 20, 2026 that Amazon’s ZeroOne team is developing a new smartphone codenamed “Transformer,” centered on Alexa+ AI capabilities. The project’s led by J Allard, the ex-Microsoft executive who helped create Xbox—the same “visionary hardware” playbook that gave us a $640 phone that ended up selling for 99 cents.
Here’s why this matters right now: Amazon’s AI genuinely works. Alexa+ users make three times more purchases than classic Alexa users. But that success is exactly why the phone will fail—smartphones aren’t shopping terminals, and Amazon’s never proven it understands what people actually want from consumer hardware.
Amazon’s last phone cost $640 and ended up in a landfill
The Fire Phone launched in July 2014 at $649 unlocked, or $199 with a two-year AT&T contract. Competitors sold for $99 on contract. Jeff Bezos micromanaged every detail—Dynamic Perspective, Firefly shopping scanner, exclusive AT&T deal. None of it mattered.
Fourteen months later, Amazon discontinued the device. The price had dropped to 99 cents. Amazon announced a $170 million charge in Q3 2014 for inventory write-downs and supplier commitments—unsold phones destroyed, contracts paid out, ambition incinerated. That’s 0.5% of one quarter’s revenue, but the reputational damage lasted a decade.
The Transformer phone reportedly takes the opposite approach: minimalist design, no app store, AI-first personalization. It’s early-stage and could be canceled. But the strategy reveals the same fundamental misunderstanding—Amazon thinks its infrastructure advantage (Alexa, AWS, shopping data) translates to consumer desire for Amazon-branded hardware.
It doesn’t.
Alexa+ proves Amazon’s AI works—for groceries, not gadgets
Amazon showcased Alexa+ at CES 2026’s AI hardware announcements, and the numbers are real: users make three times more purchases and request recipes five times more frequently than classic Alexa users. The service scaled to tens of millions of customers in nine months. Engagement doubled. This isn’t vaporware—it’s a working product that monetizes shopping behavior.
But compare that scale to the smartphone market. Apple ships 230 million iPhones annually. Samsung ships 270 million devices. “Tens of millions” sounds impressive until you realize it’s 5% of one quarter’s iPhone sales. And those Alexa+ users aren’t asking for a phone—they’re asking for recipes and reordering paper towels.
The real revenue model is the $19.99/month Alexa+ subscription, free initially for Prime members. That’s recurring revenue tied to shopping frequency, not hardware margins. The Alexa+ data supports the broader trend of voice AI replacing traditional interfaces, but that doesn’t translate to smartphone sales. Amazon’s AI works because it makes shopping frictionless. A phone optimized for shopping is a phone nobody asked for.
The $200 billion bet that ignores what people actually buy
Amazon’s projected 2026 AI capital expenditures—$200 billion across chips, robotics, and infrastructure—dwarf the entire smartphone industry’s R&D budgets. Apple spends roughly $35 billion annually on research and development. Samsung spends about $26 billion. Amazon’s betting eight times Apple’s R&D on AI, and a fraction of that could fund a phone launch.
But money doesn’t solve the core problem: Amazon doesn’t understand consumer hardware outside its own ecosystem. Fire tablets work because they’re $50 shopping portals. Kindles work because they’re single-purpose reading devices. Echo speakers work because they’re stationary shopping assistants. Meanwhile, Apple’s 2026 hardware roadmap includes foldables and AR glasses—categories where Amazon has zero credibility.
The Fire Phone failed because Amazon built a shopping device and called it a smartphone. The Transformer phone will fail for the same reason, even if the AI is revolutionary. Alexa+ users buy more groceries, not more phones. And $200 billion in AI spending won’t change the fact that nobody wakes up wanting an Amazon phone—they wake up wanting an iPhone that happens to have Alexa on it.
The question isn’t whether Amazon can build an AI phone. It’s whether anyone wants to buy one from a company that turned its last attempt into $170 million of e-waste.
