Signals Inbox·August 27, 2026·AI Infrastructure

What startups Amazon bought recently?

Amazon’s recent startup purchases are surprisingly focused: DuckLabs, Fauna Robotics, RIVR, Rightbot, Feenix.ai, Axio and Bee, with robotics emerging as the clearest theme and a broader pattern of Amazon buying companies it already knows well.

Q1What startups has Amazon bought most recently?

As of now, Amazon’s clearest recent startup purchases are DuckLabs, Fauna Robotics, RIVR, Rightbot, Feenix.ai, Axio and Bee.

We count deals from 2025 onward where Amazon either completed an acquisition or signed a definitive agreement to acquire the company. That distinction matters because Amazon has also been hiring founders, licensing technology and buying selected assets from startups without actually buying the companies themselves.

The latest name is DuckLabs, the Amsterdam company behind DuckDB. AWS has signed a definitive agreement to acquire DuckLabs, although the transaction still needs to close. Before that came a striking run of robotics deals: Fauna Robotics, RIVR and Rightbot. Bee, Axio and Feenix.ai make the recent list more varied, covering wearable AI, lending and cloud procurement.

Amazon’s clearest recent startup purchases

Q2What actually counts as Amazon buying a startup?

For this article, Amazon has to acquire the company or substantially take control of the business before we call it a startup acquisition.

That sounds obvious until we look at Amazon’s dealmaking. In several recent transactions, founders and engineers joined Amazon while the original startup remained alive. Amazon sometimes licensed the technology as part of the same deal. Other transactions involved assets and hiring offers rather than ownership of the company.

Those deals can be strategically almost as important as acquisitions, but calling all of them “Amazon acquisitions” would inflate the list and hide an interesting part of Amazon’s strategy. We therefore separate full acquisitions, pending acquisitions and these narrower talent or technology deals.

Q3Is DuckLabs really Amazon’s latest startup acquisition?

Yes. AWS’s agreement to acquire DuckLabs is currently Amazon’s newest publicly announced startup acquisition deal we could identify.

DuckLabs matters because Amazon is buying the company behind DuckDB without buying the open-source DuckDB project itself. Amazon says DuckDB will remain under the independent DuckDB Foundation and keep its MIT license, while DuckLabs founders Hannes Mühleisen and Mark Raasveldt will join AWS and continue leading the team.

AWS also knew the company well before making the offer. Amazon says AWS and DuckLabs had been working together since 2024. DuckDB, meanwhile, has become a widely used analytical database because it can run complex analytics directly on local files and inside applications without requiring a traditional database server.

The structure makes AWS’s interest pretty clear. Owning DuckDB itself could upset the open-source ecosystem that helped make it popular. Owning DuckLabs gives AWS the engineering team and much closer access to the expertise around the technology while leaving the project open.

Q4Why has Amazon bought so many robotics startups lately?

Amazon’s recent startup buying has become unusually concentrated in robotics, with Rightbot, RIVR and Fauna Robotics all acquired in 2026.

These three companies attack very different problems. Rightbot automates the messy job of unloading boxes from trailers and containers. RIVR builds robots that can travel on wheels but use legs when they encounter stairs or other obstacles on the way to a customer’s door. Fauna Robotics builds smaller humanoids designed to operate safely around people.

Amazon is already far beyond the stage of experimenting with warehouse robots. The company passed one million deployed robots in 2025, spread across more than 300 facilities, and says robotics now assists with roughly three quarters of customer orders. Amazon also developed DeepFleet, an AI model designed to improve robot travel efficiency by about 10%.

So the interesting part today is where Amazon is buying rather than whether it believes in robotics. These acquisitions push automation into environments that are still difficult to control: packed trailers, sidewalks and staircases, and spaces where robots have to work directly around humans.

Q5Is Amazon investing in startups before it buys them?

Yes, and the pattern is strong enough to take seriously: Amazon had already invested in both Rightbot and RIVR before acquiring them, while several other recent targets had long commercial relationships with Amazon.

Amazon launched its $1 billion Industrial Innovation Fund to invest in technologies around fulfillment, logistics and supply chains. Rightbot raised $6.25 million in 2023 in a round led by the fund. RIVR later raised a $22.2 million seed round that included the same Amazon fund and Bezos Expeditions.

As seen above, both companies eventually ended up inside Amazon. The earlier investments gave Amazon a much closer view of their technology and teams than a normal buyer would get during a few weeks of acquisition due diligence.

The pattern goes beyond venture investments. Axio had worked with Amazon for about six years before the acquisition. AWS had collaborated with DuckLabs since 2024. Feenix.ai built specifically around AWS Marketplace before joining AWS.

Across our seven clearest recent targets, five had an identifiable investment, partnership or ecosystem relationship with Amazon before the deal. “Try before you buy” is one of the clearest patterns in Amazon’s recent startup M&A.

Amazon’s relationship with recent targets before the deal

Q6Is Amazon buying RIVR to replace delivery drivers?

Not anytime soon. Amazon buying RIVR gives it a serious experiment in doorstep robotics, but RIVR has not yet proved that robots can replace delivery drivers at Amazon scale.

RIVR’s robot looks a little like a four-legged machine on wheels. Wheels handle the efficient part of the journey, while legs let the robot deal with stairs and uneven terrain. That combination targets one of the hardest parts of delivery automation: the final stretch from the vehicle to the actual front door.

Before Amazon bought the company, RIVR was already testing deliveries in Austin with Veho. Founder Marko Bjelonic had talked about scaling to 100 robots, but public evidence never showed deployment at anything close to the millions of deliveries Amazon handles.

Amazon’s warehouses can be redesigned around machines. A neighborhood cannot. Stairs, dogs, curbs, gates, rain, parked cars and unusual entrances make doorstep delivery much harder. RIVR gives Amazon a way to work on that specific problem now, without pretending the driverless future has already arrived.

Q7Is Fauna Robotics Amazon’s bet on home humanoid robots?

Fauna Robotics gives Amazon a route into humanoid robots for everyday human spaces, but a mass-market Amazon home robot still looks premature.

Fauna was founded in 2024 and introduced its Sprout humanoid in early 2026. Sprout is roughly child-sized, lightweight and deliberately designed to be safer and less intimidating than the heavy humanoids being developed for factories.

As seen above, Amazon bought Fauna only weeks after the company formally introduced Sprout. Fauna says it will continue selling the Creator Edition and supporting existing customers while operating as an Amazon company. That suggests the technology is still in the learning and developer stage rather than ready for a huge consumer rollout.

Amazon has an obvious reason to care. Alexa gives it a software assistant inside millions of homes, while Ring, Echo and other devices give it a large consumer hardware footprint. A useful mobile robot could eventually connect those pieces. For now, Fauna is better understood as an early bet on robots that can safely exist around people.

Q8Why did Amazon buy Bee when it already has Alexa?

Amazon bought Bee because Bee follows the user through the day, while Alexa has historically understood much more of what happens around Amazon devices.

Bee sells a small wearable that can listen to conversations, turn them into transcripts and remember commitments, ideas and tasks. The device launched at a dramatically lower price than many earlier AI wearables, and Bee had raised only around $7 million before Amazon acquired it.

Since joining Amazon, Bee has kept moving toward what its founders call personal or ambient intelligence. New features can turn conversations into actions, connect with email and calendars, and surface recurring patterns from a user’s day.

That gives Amazon something Alexa never naturally had: continuous context outside the home. Amazon already knows how to build voice assistants and hardware at enormous scale. Bee brings a different interaction model where the AI remembers what happened before the user explicitly asks for help.

The privacy question will be difficult because a device that understands more of someone’s life also needs access to far more personal information. But the product logic behind the acquisition is easy to see.

Q9Why did Amazon buy Axio in India?

Amazon bought Axio to own a real lending operation in India rather than relying entirely on outside financial partners.

Axio, formerly known as Capital Float, was much bigger operationally than most of the startups on this list. Before the acquisition, the company said it had served more than 10 million customers and managed a loan book of roughly ₹2,200 crore. Amazon had already spent around six years working with and investing in Axio.

After approval from the Reserve Bank of India, Amazon completed the purchase and turned Axio into a fully owned subsidiary. Indian financial press put the all-cash deal at roughly $200 million, although Amazon did not publicly disclose the exact price.

The regulatory piece is especially useful. Buying Axio gave Amazon control of an established non-bank financial company rather than forcing it to build a lending operation from scratch. Amazon can now connect commerce, payments and credit more directly in one of its largest international markets.

Among Amazon’s recent startup purchases, Axio looks the most like traditional M&A: Amazon bought customers, a loan book, a regulated entity and years of operating history, not just a small technical team.

Q10Why would AWS buy Feenix.ai?

AWS bought Feenix.ai because the startup had already built software around one annoying problem inside AWS Marketplace: making enterprise software transactions easier to manage.

Feenix.ai helped vendors automate marketplace listings, private offers, reporting, procurement integrations and other administrative work. The company was founded in 2023 and built its product directly around AWS Marketplace.

When Feenix.ai joined AWS, one of the AWS executives involved publicly described the transaction as M&A. The founders said customers would continue receiving support while the team worked inside AWS Marketplace on custom storefronts, procurement-system integrations and end-to-end automation.

This is a small deal compared with something like Axio, but the fit is unusually tight. AWS Marketplace makes money and becomes more useful when companies can buy and sell software through it without weeks of procurement friction. Feenix.ai had already spent its entire life solving that problem inside Amazon’s own ecosystem.

Q11Which supposed Amazon startup acquisitions do not actually count?

Adept, Covariant and Pixwise should not be presented as normal Amazon acquisitions, while Viziotix remains harder to classify because public disclosure of the proposed deal is clearer than public evidence of its completion.

Adept is the clearest example. Amazon hired co-founder David Luan and several other Adept employees while licensing the startup’s agent technology, models and datasets. Adept stayed alive under new leadership.

Amazon did something very similar with Covariant shortly afterward. It hired founders Pieter Abbeel, Peter Chen and Rocky Duan, along with roughly a quarter of the company, and received a non-exclusive license to Covariant’s robotic foundation models. Covariant continued serving customers.

Pixwise later appeared in European regulatory disclosures as a transaction involving a hiring offer and assets. Again, commercially meaningful, but different from buying the company.

Viziotix sits closer to the acquisition side. Amazon disclosed an intention to acquire the French computer-vision startup under European rules for large digital platforms. We found much clearer public evidence of the intended transaction than of a conventional completed acquisition, so we would not put it in the same confirmed bucket as Axio or Bee today.

Deals that should not be treated as conventional Amazon acquisitions

Q12Is Amazon mainly buying AI startups now?

AI is everywhere in Amazon’s recent deals, but calling this an “AI startup buying spree” misses what Amazon is actually shopping for.

Three recent acquisitions are robotics companies. DuckLabs strengthens data analytics inside AWS. Feenix.ai improves AWS Marketplace. Bee gives Amazon a wearable interface for personal AI. Axio provides lending infrastructure in India.

Several of these companies obviously use AI, especially the robotics businesses and Bee. But Amazon already spends tens of billions of dollars building AI infrastructure, models and services internally. It does not need to buy a startup every time it wants another generic AI capability.

Its targets lately are much narrower. Amazon seems interested when a startup has solved one awkward piece of a larger Amazon system: unloading a trailer, reaching a doorstep, operating around humans, making cloud procurement easier, adding local analytical database expertise or building a persistent personal-AI interface.

Amazon’s own 2025 annual report says the main reasons for its acquisitions were acquiring technology and know-how or expanding the customer base. That broad corporate wording actually fits this group quite well.

Q13Is Amazon buying these startups for revenue or for their technology?

Amazon is mainly buying capabilities, teams and time; Axio is the big exception where the existing business itself was clearly part of the prize.

The scale difference makes that obvious. Amazon generated hundreds of billions of dollars in annual revenue. Bee had raised about $7 million before being acquired. Rightbot’s 2023 funding round was $6.25 million. RIVR had raised around $25 million and was last reported at a roughly $100 million valuation before Amazon bought it. Fauna Robotics had existed for only about two years.

None of those companies could materially change Amazon’s revenue immediately. What they can do is save Amazon years of development on difficult technical problems.

Amazon’s filings reinforce that point. The company reported $780 million of aggregate cash consideration for acquisition activity in 2024, while completed acquisition activity in 2025 was described as immaterial to Amazon’s consolidated financial results. Amazon does not disclose enough transaction prices for us to produce a credible total for the startups in this article.

Axio stands apart. The reported purchase price was around $200 million, and Amazon acquired millions of customers, a loan portfolio and regulated lending infrastructure. The economics of that transaction already existed before Amazon arrived.

For the smaller technology companies, revenue today is almost beside the point. Amazon is paying for engineers, intellectual property, deployment experience and a head start.

Q14Is Amazon buying startups more aggressively now?

Yes, Amazon has clearly become more active in direct startup acquisitions lately, although there is not enough disclosed spending data to call it a giant M&A boom.

The contrast with 2024 is useful. Two of Amazon’s most discussed AI deals that year, Adept and Covariant, used hiring and licensing structures while leaving the startups independent. Amazon also spent $780 million across its total acquisition activity that year, according to its annual filing.

The next phase looks different. Amazon completed Bee, Axio and Feenix.ai transactions in 2025, and 2026 has already produced a concentrated run of direct technology deals, particularly in robotics, followed by AWS’s agreement for DuckLabs.

The change is easier to see in the number and type of deals than in dollars because Amazon keeps most purchase prices private. Direct ownership is showing up more often in Amazon’s startup strategy these days, especially when the company finds a technology that fits an existing Amazon operation immediately.

Q15So, what startups has Amazon bought recently?

Amazon has recently bought a surprisingly focused group of startups, led today by robotics companies and small businesses that plug directly into AWS, devices or financial services.

The clearest recent names are DuckLabs, Fauna Robotics, RIVR, Rightbot, Feenix.ai, Axio and Bee. DuckLabs is still a signed deal awaiting completion, while the other names are completed or clearly integrated transactions. Adept, Covariant and Pixwise belong in a separate category because Amazon obtained people, assets or licenses without buying the whole company in the conventional sense.

Looking across the deals gives us a sharper answer than the company list alone. Amazon is buying things it can use. The recent targets solve specific problems inside businesses Amazon already operates at huge scale, and five of our seven core companies had already worked with Amazon, taken Amazon money or built directly around its ecosystem before the transaction.

Robotics is currently the standout theme. Amazon already operates more than one million robots, yet its newest purchases are pushing automation into the harder parts that remain: unpredictable freight, the journey to the doorstep and environments shared directly with humans.

That tells us more about Amazon’s acquisition strategy than simply saying it is buying AI startups. These days, Amazon appears most willing to buy when a small company has already solved one difficult piece of a very large Amazon problem.