‘Genie is out of the bottle’: Inside the world of AI development at Amazon HQ
Saturday, 22 November 2025
It’s like we’re in 1994, on the cusp of the Internet creating the revolution it has. That’s how one executive described the current state of innovation being brought to the fore by artificial intelligence (AI) right now.
Back then people knew there was an amazing technology that would change the world, but they did not know how or to what extent. The same is true for AI - and just how far reaching and unconventional the impacts will be.
The technology, able to perform tasks requiring human intelligence, take action and complete goals autonomously, with the ability to reason and make decisions on its own, is being developed at lightning speed.
On the ground in Seattle - home to some of the world’s largest companies such as Amazon, Microsoft, Starbucks and Boeing - everyone you speak to about the technology seems to be optimistic about its benefits.
And every second or third person walking the streets of downtown Seattle seems to be an Amazon employee, with a front row seat to a technological revolution unfolding.
For a few days, the Post was also in the front row, invited to visit Seattle to learn more about Amazon Web Services (AWS) and how it is developing AI.
“There's no area, there's no work, there's no role or experience that does not change with AI,” says Australian AWS executive Colleen Aubrey, senior vice president of applied AI solutions.
AWS is a subsidiary of Amazon, the world’s fifth largest company, providing the world’s most broadly adopted cloud computing services from data centres dotted around the world. AWS’ revenue has grown by 20% to $33b in the last quarter alone, suggesting it remains competitive with the likes of Microsoft and Google in using and developing AI for widespread use.
AWS has thousands of customers across New Zealand using its products and applications - from small companies to large New Zealand corporations including BNZ and One NZ. And while adoption is generally considered to be lagging here, there’s no question it will widen.
In Seattle, I come across no-one who is not impressed or inspired by what AI can do.
The New Zealand perspective on AI is one of tentative use of the technology in a remote corner of a world seemingly awash with AI mania - companies large and small pouring billions into AI development and data centres to run the tech and keep themselves on the cutting edge of things.
Research firm CB Insights says more 1300 AI startups now have valuations of more than $100 million, with 498 AI “unicorns,” or companies with valuations of $1 billion or more, and while the money keeps flowing, questions are increasingly being asked about whether the marketplace is heading for a correction. As bubble talk increases, some investors are already displaying some renewed caution, even if this week’s Nvidia result calmed some anxiety.
But AI its not an “opt in” choice any of us can make any more, says Aubrey, who oversees several projects including a generative AI-powered version of Amazon's Alexa voice assistant, Alex Plus, which the company says will allow for more natural and conversational interactions - the chatbot can remember previous parts of a conversation, perform tasks like generating responses to queries and shop on your behalf and set up price comparisons on consumer goods.
Aubrey says she is excited to see how Australian and New Zealand firms put AI to work in their workforces to get “humans into the business of being humans” again.
“The human experience is going to change. It will change in how we work. It will be changing in our homes. I'm excited about the potential of Alexa Plus, the way that it will be able to traverse many surfaces to be able to be personal assistants in our lives. I'm excited about how the way that we work will change, the coming together of AI, team mates and people, and the collaboration that will happen. The bottlenecks are changing, the constraints are changing, and I think AI gives us this opportunity to re-explore the potential of what we can do.”
Aubrey says she thinks the biggest constraint for AI right now is us.
“People have varying levels of tolerance, and that might be the bigger constraint than the technology. But actually, I don't think this is an opt in change. Everyone is in.
“The way we navigate the world and the way that we interact with businesses and communities will be impacted by this. Regardless of what you might choose in your personal sphere, any time you're a part of a community that's going to change.”
AI hype?
Matt Garman, chief executive of Amazon Web Services, calls AI nothing short of “the biggest change in technology that has ever been seen”.
He, and all the big technology firms, are tag teaming in the AI race and seemingly all customers of each other, working to cash in on widespread adoption, optimistic about what AI will mean for the future.
“I do think that generative AI is going to completely change every single company, industry and job that we have out there … That kind of transformative technology is an enormous opportunity to change, and it's a disruptive force,” Garman told reporters including The Post as he discussed the future of AI at the company’s headquarters in Seattle earlier this month.
“[AI] is a big opportunity for companies. It's a big opportunity for employees. It's a big opportunity for industries to rethink how they operate. Part of what is most disruptive about this technological change is it is like nothing we've ever seen before, because the capabilities that are possible are enabling discoveries. They're enabling different work flows, enabling completely new ways of doing things, and that technology is moving faster than almost every technology has ever before.
Even tackling cancer comes within the remit for AI, according to Garman.
“Whether it's going to be finding cures for cancer, whether it's going to be discovering new minerals, or [speeding up] our ability to deliver software, it's an incredible opportunity and time to be out there inventing.”
There are many people working on bringing that dream to life. Amazon has more than 15 offices and employs almost 50,000 people in Seattle alone. And so far, the results have been successively promising, with surging demand for AWS services driving record after results record. The 20% growth rate in its third quarter was the fastest since 2022, and comes even as the company suffered a global outage that impacted thousands of websites for a time.
Amazon’s huge investment will continue. The company expects to spend US$125b ($220b) on AI infrastructure in 2025, and plans to invest more as demand for its services continues.
The company increased its year-on-year purchases of property and equipment by $50.9b, and added 3.8 gigawatts of power capacity to support AI infrastructure, and launched a massive computing cluster with almost 500,000 custom AI chips.
Garman believes AWS will be able to continue its double-digit growth.
“We're a US$135b business growing 20% and that requires more computers, more storage, it requires more data centres and more power, and so that's a massive investment for us, and we have to make that investment well ahead of where the demand comes.”
Agentic AI
Next year is quipped to be “the era of agentic AI”, one that will see the rate of transformation pick up pace, powered by multiple AI agents that work together to identify problems and tasks and resolve them for organisations.
The types of jobs people will do as a result of this, and “the shape” of those jobs will move towards what Garman calls a flatter workforce: “[One] with fewer layers, where you may have more people doing some types of roles and fewer people doing other types of roles, is going to be what we move to …It doesn't mean that they'll have fewer employees necessarily. It just means you may actually shuffle around.”
Rahul Pathak, vice president of data and AI at AWS, says agentic AI is presenting exciting opportunities and already delivering benefits for businesses using it.
“We've seen remarkable results with customers in terms of what they've been able to do with agentic and generative AI, but I think we're just scratching the surface,” says Pathak.
“We're going to innovate in the models, in agents and inference and interaction with data and infrastructure sites. Twelve months ago, I don't think we saw all the things we're working on today, and in 12 months from now, there’s going to be exponential change.”
Pathak says AI will catapult the world to super efficiency, and it will come in the form of AI and other technologies combined: “AI is a very important technology, but it's one of many. A lot of the classic machine learning, the classic optimisation techniques, analytics, all of those have a role to play, and ultimately, it's about working backwards from the problem and thinking about all of these various capabilities as tools in our tool chest to solve a problem.”
Customer demand is driving decision-making around the significant investment in AI at the moment, that and the company wanting to stay ahead of others in the development race.
“We're seeing a ton of interest, whether that's in software development, in knowledge work and agentic tools like Quick, access to core inference through Bedrock. We also see some of the world's largest model providers wanting to train on us, and so things like Project Rainier and Anthropic are driving a bunch of our investments in core infrastructure. There’s very broad based interest.”
Future of work
There is, of course, the broader question of job losses with AI adoption, and even Amazon itself is not immune to job losses. The company is planning to reduce its workforce by 14,000 as it continues to make the business case for further investment in artificial intelligence.
The cuts are expected to target areas such as human resources, advertising, and management in a group that has 350,000 office positions, out of a total of more than 1.5 million employees - roughly 8% of its workforce.
The company denies the cost-cutting measures come as AI replaces roles, but say it is rather about “culture”.
Garman could not say whether AWS would cut more jobs. But he did say Amazon would continue to try to remove “layers” of management to incorporate a “flat structure”, and acknowledged that AI was permanently changing the nature and roles of those who worked at the company.
“Our goal is to have a flatter organisation with fewer layers, where all employees are closer to the end customer.
”One of the things we can do is remove layers of management. We've seen that when you remove layers of bureaucracy, it's a better place to be. You move faster. I think you're going to see more companies make changes like that,” Garman said of what people could expect from the future of work.
He says AI was changing how people worked at a speed that was not yet fully acknowledged. “All of our jobs are going to change over the next 10 years and we're going to have to, as a society, adapt and learn new skills.
“Does this mean robots are going to have all the jobs and there's no jobs for humans any more? That seems to be quite far away. I see that nowhere in the near term, particularly because I actually think that in many ways the technology supercharges people that are great at their jobs today.”
NZ adoption
For a New Zealander a lot of the talk may sound like hype, even if it is hard not to catch the enthusiasm pulsating through not just the AWS headquarters but the city itself. It suggests New Zealand’s low adoption rate has something to do with the country being so far away from mass development of the technology.
Datacom director of artificial intelligence Lou Compagnone thinks so. She says: “If we look around different places in the world, we're seeing different horizons of AI that haven't quite hit us. It's easy when you're not surrounded by that to go, ‘Oh, it's progressing, but it's not very fast’, but that’s just because we're not surrounded by that level of technological advancement.”
Just 33% of New Zealand firms are using AI day-to-day, but Spark chief technology and AI officer Matt Bain says executive attitudes towards AI are changing in New Zealand.
There was a stark difference in levels of optimism noticeable among leaders at its Spark Accelerate tech summit this year.
“Last year there was a lot of doom and gloom and fear about the risks of AI and, a fear of getting started because of the risks. This year there's been a lot more optimism around how do we start to get on the front foot and make the most of this technology,” Bain says.
Consumers, however, were still ahead of businesses when it came to regularly using and adopting the technology, he says.
Accenture Cloud First global lead Andy Tay says the technology consultancy has seen enterprises trial AI with small pilots and quickly move to company-wide deployment after seeing how transformative it can be.
“Some of our clients have changed the way they select use cases, and also change the way they're measuring ROI [Return On Investment]. For example, a contact centre agent - a physical human agent - is now spending less time on a repetitive task, and more time on empathetic human-to-human touch, that's a different form of ROI, and I think we'll see more of that come to the fore.”
But there is the question of that pesky - and contested - AI bubble, which, if true, could throw the brakes on further AI development. Already the bubble forming has caused commentators to compare it to that of the internet bubble of the early 2000s - although they also note that the impact hardly stopped the rise of the internet as it became a daily staple for vast swathes of the world’s population.
Still, there is a lot at stake. The market capitalisation of hot-in-demand computer chip maker Nvidia, for example, is more than US$4 trillion - more than the GDP of India, the world’s second-largest economy.
New Zealand-based tech expert David Downs, chair of The Icehouse and a member of the Government’s AI Expert Advisory panel, says there is no doubt technology companies and investors are benefiting from a steep rise in technology company share prices right now.
Even if that is short-lived, AI was not a fad and would continue to develop, he says. “Even if the pace of change slows, and in some ways I hope it does; to allow the policy and ethical considerations to be properly considered, even the stuff we have today is having - and will continue - to have an impact. The genie is out of the bottle,” says Downs.
“The bubble is most often talked about in terms of tech stocks and investment into the big tech providers. There is no doubt they are benefiting from increased investment and individually getting higher valuations than the fundamentals would suggest, but that is quite different at an industry impact level.”
AI will continue to advance, he says: “AI is a structural shift in how technology works that will impact many areas - business, education, government. When we look at the capabilities of AI today, let alone what the upcoming developments are likely to achieve - working at an exponential rate - there can be no doubt.”
Down says New Zealand firms needed to “move from simply understanding and playing with AI” to serious adoption. “We need more focus and a more developed and resourced approach nationally. There are moves afoot from the government and others, but sorry to say I feel it's too little and too slow.”
As Downs puts it, if you use a keyboard in your job, you are likely going to have to change what you do in the next five years.
And after that will come a rapid development in robotics. AI and robotics combined will result in “a very different world” in the future, he says.
“My advice is not to be scared, but certainly be curious and put in the effort to understand AI so that you are well set up as these changes continue.
“Done well, AI and robotics will make lives easier. But like any advancement, it won’t be evenly spread and we have to ensure that the benefits of AI are accessible to all, and don’t further expand inequity and the digital divide.”
Aimee Shaw travelled to Seattle courtesy of Amazon.