What did Dropbox announce on 5 November 2026?
Dropbox has agreed to buy Tigris Data, a storage startup built for AI workloads, for about $350 million in cash. The company announced the deal on 5 November 2026, two days after signing a definitive agreement on 3 November.
The deal still has to satisfy customary closing conditions. Dropbox expects it to close in the fourth quarter of 2026, its Form 8-K says. Tigris is described in the filing as a globally distributed object storage platform built for AI workloads.
Dropbox also unveiled DBX Agent Cloud, a storage and data service for companies building AI applications and agents. It will start with standalone object storage that developers can buy separately. The service is compatible with Amazon S3 tools.
The third piece is a new AI layer inside Dropbox itself. The company says the experience has reached more than 70,000 teams. It lets users ask for an outcome rather than a set of search results.
Crunchbase News reported that Dropbox described the purchase as its largest acquisition since it was founded in 2007. Earlier deals included HelloSign at $230 million, DocSend at $165 million and FormSwift at $95 million.
Webster framed the two sides as separate bets. He told Crunchbase News that DBX Agent Cloud opens a new developer market. AI inside Dropbox, he said, is meant to deepen engagement and retention among existing customers.
- •Deal: about $350 million in cash for Tigris Data
- •Signed: 3 November 2026, announced 5 November 2026
- •Closing: expected in the fourth quarter of 2026
- •Product: DBX Agent Cloud, starting with S3-compatible object storage
Why is Dropbox buying Tigris instead of building?
Dropbox is not new to running storage at scale. The New Stack notes that from 2015 the company moved over 90% of user data off Amazon S3. The data went onto an in-house system called Magic Pocket.
That history explains why Dropbox believes it can sell infrastructure. It does not explain why the company is paying for a startup instead of extending Magic Pocket to outside developers.
Eric Webster, Dropbox's chief business officer, drew a line between the two jobs. He told Crunchbase News that Tigris brings a product already running in production. He also cited its experienced team and established relationships with AI developers.
Tigris sells S3-compatible object storage spread across 11 regions on its own hardware. It does not charge egress fees, the charges cloud providers apply when data leaves their networks.
Those fees matter for AI teams that train or run models across several compute providers. TechCrunch reported in October 2025 that egress costs once accounted for the majority of cloud spending at Fal.ai, a Tigris customer.
Tigris had more than 4,000 customers at that point, mostly generative AI startups with large, latency-sensitive datasets. It raised a $25 million Series A led by Spark Capital, with Andreessen Horowitz participating. Crunchbase data puts its total funding at about $31.9 million.
The purchase price is far above those earlier Dropbox deals. It reflects what Dropbox will pay for a developer business with a product, customers and a team already in place.
Dropbox has made 35 known acquisitions, according to Crunchbase, including 23 publicly reported deals between 2012 and 2015. It made four across 2024 and 2025, among them Reclaim.ai, Nira, Promoted.ai and Mobius Labs.
The Tigris deal is larger than all of them and points at a different customer: developers rather than productivity users. That shift is the clearest sign that Dropbox sees its next phase as infrastructure as well as software.
Tigris was founded by former members of Uber's storage team, including CEO Ovais Tariq, CTO Himank Chaudhary and chief architect Yevgeniy Firsov. Its pitch is that AI workloads need storage that follows compute across clouds, not storage tied to one provider's data centres.
Tariq told TechCrunch that "without storage, compute is nothing." The line captures the gap Tigris is trying to fill between rented GPUs and the data those GPUs need to read.
What is DBX Agent Cloud, and who is it for?
DBX Agent Cloud is aimed at a different customer from Dropbox's file-sharing base. It targets companies that need somewhere to put training data, model artifacts, images, video and other content their AI systems produce.
Dropbox describes it as a storage and data layer. It is designed to work alongside the clouds and compute providers a customer already uses. The first product is object storage sold on its own.
Because it speaks the S3 API, developers can keep using familiar tools such as boto3, the AWS SDK, rclone and Terraform. The New Stack reports that Tigris already serves customers including Railway, Runpod and Replicate.
Dropbox plans to add filesystem and content-processing capabilities next. A filesystem would give agents a persistent place to manage files between tasks. Content processing would turn documents, images, audio and video into inputs that models can use.
Several important details are missing. Dropbox has not announced a launch date or pricing for DBX Agent Cloud. It has also not explained how the two companies' technology will be integrated. Dropbox told The New Stack it will share more as plans develop.
Existing Tigris customers will not need to migrate or change how they use the service when the deal closes, Dropbox says. They will continue to receive support for the Tigris product.
That pledge covers the immediate transition. It does not settle what happens once Dropbox layers its own filesystem and content-processing products on top.
AI workloads put different demands on storage than document sync does. Training jobs read large datasets repeatedly, inference services need low latency, and agents expect files to persist between sessions.
Tariq says the platform supports billions of small files and keeps data close to the GPUs that need it. Those claims come from the company and its chief executive, not from an independent benchmark.
The multi-cloud point is central. A team may store data with one provider while renting GPUs from another, which can trigger transfer fees and delays. Tigris avoids both by keeping data near compute and charging no egress.
Whether that advantage survives inside Dropbox is untested. Dropbox has run storage for its own product. Selling it to outside developers, however, requires pricing, support and reliability commitments the company has not had to offer before.
What changes for Dropbox users?
The AI features inside Dropbox follow a separate track from the infrastructure bet, but they share the same logic. Dropbox wants AI to work on files where they already live, rather than in a chat window that requires uploading them again.
Users can now describe a task and Dropbox will find relevant files, propose a plan and carry out supported actions after approval. The actions include summarising folders, comparing versions, editing PDFs and images, transcribing recordings and trimming video.
Dropbox says the system works within existing permissions, keeps a record of actions and lets users undo supported changes. Those controls matter because the software is changing business files, not just answering questions about them.
The rollout is limited. More than 70,000 teams have the new experience. The initial release covers eligible Standard and Advanced plans in the United States, the United Kingdom and Australia, in English. Broader availability is planned for 2027.
The limits are practical as well as technical. Dropbox has not said when other markets and plans will get the features.
Customers can switch between supported models during a session, according to Dropbox. They do not need a separate account or paid subscription from the AI provider. Files can also be used inside ChatGPT, Claude, Gemini, Muse, Cursor and Copilot, with results returning to Dropbox.
Dropbox published two customer examples. Intelligent Communities Group said Dropbox sorted a folder of 1,200 images in about 10 minutes. Creative agency YYES said it reorganised about 100 files in two or three minutes, against 15 to 20 minutes by hand.
These are company-supplied anecdotes, not independent benchmarks. They show the intended workflow, but they do not establish accuracy across messy real-world folders.
How does the deal fit Dropbox's finances?
The acquisition lands against a mixed financial picture. Dropbox reported second-quarter 2026 revenue of $631.5 million, up 0.9% from a year earlier. Annual recurring revenue was $2.566 billion, up 1.0%.
Paying users stood at 18.19 million, an increase of 96,000 from the prior quarter. GAAP net income was $95.8 million, and the company held $1.114 billion in cash, cash equivalents and short-term investments.
At about $350 million, the Tigris purchase would use roughly a third of that cash pile. Dropbox can afford it, but the deal is large enough to matter on its balance sheet.
The price is small relative to $2.566 billion in annual recurring revenue. It is large relative to Dropbox's recent deal history. It also equals more than half the revenue the company reported in the second quarter.
The company is also cutting costs. On 28 October it said it would eliminate about 200 roles. That is roughly 9% of its global workforce, across engineering, product, design and administrative teams. It estimated $27 million to $33 million in cash costs.
Chief technology officer Ali Dasdan is leaving on 18 November. The job cuts and the acquisition were announced eight days apart, and Dropbox's filings treat them as separate events.
The strategic logic is easier to see against slow core growth. A file-sync business growing at about 1% needs a new market, and AI infrastructure is one where Dropbox's storage experience is relevant.
What are the risks and limitations?
The first limitation is timing. The deal has been signed but not completed. It has not yet cleared the customary closing conditions. Dropbox has warned that the transaction may not close on the expected schedule, or at all.
The second is product uncertainty. DBX Agent Cloud has no public launch date, no published pricing and no technical integration plan. Developers cannot evaluate a service that is still being assembled.
The third is competition. Tigris sells against Amazon S3, Google Cloud Storage and Microsoft Azure, which offer deep integration with their own compute. Dropbox will need to convince AI teams that portability and zero egress are worth switching for.
Tigris already has a developer customer base, which reduces that risk. Dropbox's challenge is to keep those customers while turning a startup product into a larger platform.
The no-migration pledge gives Dropbox time, but it does not guarantee retention once the product changes. The startup's customers include GPU and model-hosting companies that can move workloads if pricing or performance shifts.
The AI features carry their own risks. Giving software permission to move, rename and edit business files raises questions about errors, audit trails and model behaviour. Dropbox's approval step and undo function address some of this, but enterprises will test them.
The financial risk is modest relative to Dropbox's cash, but not relative to its growth. AI infrastructure demand may cool, and Tigris customers may leave. In that case, Dropbox will have spent a large sum on a business whose revenue has not been disclosed.
Dropbox has not said whether Tigris is profitable or how much revenue it generates. Without those figures, the price is hard to judge against the business being bought.
What should you watch next?
Dropbox had scheduled its third-quarter 2026 results for 5 November, the same day as the announcement. The report was expected to show whether the 0.9% growth rate reported in the second quarter had changed.
The next milestone is the closing of the Tigris deal, expected in the fourth quarter. Watch whether Dropbox discloses the acquisition's effect on revenue and costs.
Developers should watch for DBX Agent Cloud pricing and availability. The first test will be whether existing Tigris customers stay and whether new AI teams adopt it for training and inference data.
Pricing will matter for adoption. Without published rates, it is impossible to compare DBX Agent Cloud with Amazon S3 or with specialist storage providers on cost.
Dropbox also needs to deliver the promised filesystem and content-processing features. Those are the pieces that would make DBX Agent Cloud more than a resold object store.
The 2027 expansion will show whether Dropbox can move the AI experience beyond 70,000 teams and three English-speaking markets.
Finally, watch the technology leadership transition. Ali Dasdan leaves on 18 November, while Dropbox works to close and integrate a storage startup and build a new infrastructure business.