Amazon Web Services Launches Quantum-Classical Hybrid Service Braket Direct
Access to IonQ's 200-qubit processor via standard API calls, priced at $0.05 per task.
The Commodity Trap: How AWS Sells Smoke and Makes Money from It
Author: Independent analyst specializing in cloud computing and quantum technologies.
[The Gist]: What's Really Happening
Official version: Amazon Web Services launched Braket Direct—a service providing access to IonQ's 200-qubit processor via standard APIs at $0.05 per task. Sounds like democratizing quantum computing. A sensation? No. Marketing.
The $0.05 per task figure is a classic advertising trick. AWS documentation clearly states: IonQ Aria requires a minimum of 2500 shots per task when using error suppression. Multiply: 2500 × $0.03 per shot. That's $75 per task, not 5 cents. $75 is still cheap compared to real quantum computing, but it's not 5 cents.
Moreover, reserving dedicated access via Braket Direct costs between $2,500 and $7,000 per hour. Seven thousand dollars. Per hour. For IonQ Forte—$7,000 per hour. This isn't democratization. It's an exclusive club with a "corporations only" sign.
What's everyone keeping quiet about? A 200-qubit processor is raw qubits without error correction. The effective number of logical qubits is around 10-15. Such hardware can't break RSA, can't simulate a complex molecule, can't do any of the things depicted in sci-fi movies. $0.05 is the price for a try, not for a result. The task will most likely output noise instead of an answer.
Timeline and Context
Braket Direct isn't new technology. It's a repackaging of old functionality.
- 2020—Amazon Braket launch: AWS was the first among cloud giants to offer access to quantum computers from IonQ, Rigetti, and D-Wave. Back then, it was a breakthrough. Now, it's standard.
- Mid-2025—Braket Direct announcement: An AWS blog post announces a new program allowing users to reserve dedicated devices. The key word is "reserve." AWS realized that on-demand access to quantum computers is useless for real researchers: queues, limitations, instability.
- 2026—Current state: Braket Direct provides access to IonQ Forte (30 qubits, but they claim 200?), QuEra Aquila, Rigetti Aspen-M-3. Plus "expert consultations" and "experimental capabilities."
Key point missed by journalists: IonQ Forte has 30 qubits. Where did 200 come from? Marketing. Some providers count not physical qubits but "effective" or "algorithmic" ones. It's an old trick. Like mobile carriers advertising "unlimited" with tons of restrictions.
Who Wins and Who Loses
Winners:
- AWS (obviously): They monetize the hype. Developers rush to test the "quantum API," burn through free credits, then pay $75 per task or $7,000 per hour for dedicated access. Braket Direct isn't a technological breakthrough; it's a business model. AWS sells not a solution but hope. Hope that the quantum computer will do something. You can charge any amount for hope.
- IonQ: They get a steady stream of orders from AWS without needing to build their own sales team. Market share grows even if actual usage is minimal.
- Well-funded research labs: Universities and corporations (Google, IBM, pharma giants) can afford $7,000 per hour. They don't need cheap access; they need predictable and prioritized access. Braket Direct gives them an exclusive window with no interruptions.
Losers:
- Small startups and independent researchers: They can't pay $7,000 per hour. They're forced to use the on-demand queue where results are unpredictable. For them, quantum computing remains a toy, not a tool. The market consolidates around big players—just like with AI.
- AWS competitors (Google Cloud, Microsoft Azure): Google offers up to 1000 qubits via its Quantum AI service, but without the aggressive marketing packaging of Braket Direct. AWS wins not through technology but through ecosystem. A developer already using SageMaker, S3, and Lambda is more likely to stay with AWS for quantum tasks.
- The quantum computing industry itself, long-term: When thousands of developers try Braket Direct and see that 200 qubits don't solve their problems, they'll be disappointed. The hype will fade. Investors will ask tough questions. Braket Direct might become not a driver but a killer of interest in quantum.
What the Media Isn't Saying
Insight you won't find in any blog: A $0.05 task is a task with the minimum number of shots, which most likely won't yield a useful result.
Why? Because quantum computers are noisy. To get a reliable answer, you need to run the task thousands of times and average the results. More shots mean more accuracy. 2500 shots for IonQ Aria is the technical minimum, not a guarantee of accuracy.
A real workflow for a serious task: 100,000 shots. That's $3,000. Plus time to set up the task. Plus result processing.
$0.05 is a bait. Like a car ad for $99 a month (with a $5,000 down payment).
What else is hidden:
- Braket Direct doesn't deliver quantum supremacy. Even with dedicated access to 200 qubits, a classical GPU solves the same task faster and cheaper. The rare tasks where quantum truly wins (factoring large numbers, simulating quantum systems) require hundreds of logical qubits. They don't exist.
- Expert consultations are extra. On AWS Marketplace, you can order consultations from specialists at IonQ, Oxford Quantum Circuits, QuEra. No price listed, meaning "expensive, inquire individually."
- Experimental capabilities are a marketing gimmick. Access to new devices with limited availability is beta testing that you pay for. You test a raw product, give feedback, pay for it, and then the company sells the finished solution to your competitors.
Forecast: Next 30 Days and 90 Days
In 30 days (end of June 2026):
- First wave of disappointment: Posts on Reddit and Hacker News from developers who spent $500 on Braket Direct and got no useful result. Analyses will show that "$0.05 per task" is a lie. AWS will release a clarifying FAQ "refining" pricing details.
- Increased competition: Google Cloud will announce a free 500-qubit quantum simulator. Free. Forever. AWS will respond by raising the free credit for new users from $500 to $2,000.
- IonQ stock rises 15%: Investors see news about "integration with AWS" and rush to buy. IonQ's actual revenue from Braket Direct remains modest—a few million per quarter, a drop in the ocean for Amazon.
In 90 days (August 2026):
- Startup acquisition: AWS will buy one of the smaller quantum providers (likely Rigetti or QuEra) for $200–300 million. Rumors have been circulating for a while. Braket Direct becomes a pretext for vertical integration.
- Quantum "engine" for Bedrock: AWS will announce that quantum computing is available through Bedrock (their generative AI platform). Marketing: "Your AI can use quantum acceleration." Reality: no one can explain why that's needed.
- Mass exodus from quantum startups: Venture funds will realize Braket Direct isn't a breakthrough but standardization. Investing in yet another "quantum software" is pointless because AWS already built the platform. Startups that haven't sold will shut down or enter "zombie mode" (alive but not growing).
Bottom line: Braket Direct is a pretty wrapper for a raw product. AWS isn't selling business solutions; it's selling access to an expensive corporate hobby. $0.05 per task isn't the price of quantum computing. It's the price of an illusion that you're doing something cutting-edge. The real price is $7,000 per hour of dedicated time and months of work to get a result that a classical computer delivers in a second. The technology isn't ready. But AWS isn't waiting for readiness—AWS is waiting for you to open your wallet. And you will. Because "quantum" sounds cool. And cool always costs a lot.
— Editorial Team
No comments yet.