Deepgram is acquiring Daily. The announcement is framed around conversational AI, speech models, and the Pipecat ecosystem. If your team uses Daily for human-to-human video calls, that framing leaves a practical question unanswered: what should you check today, before any product, pricing, or service change has been announced?

The answer is to audit your dependency while the service is still operating normally. Do not migrate because a transaction was announced. Do not assume that a public assurance replaces your contract. Record what you use, what you pay, what your SLA covers, where recordings go, and how difficult it would be to move. That gives you a decision based on your own exposure rather than on speculation about either company.

What the announcement says—and what it does not

Daily’s announcement says that Deepgram is acquiring Daily. The same announcement says that Daily’s customers and the Pipecat ecosystem should expect no immediate change, and that Deepgram will sell, support, and develop Daily’s current products and services.

Those are useful statements about the announced direction. They are not a substitute for your Order Form, support package, SLA, DPA, or account-specific pricing. They also do not answer questions such as whether your renewal terms change, which entity handles a future support escalation, whether your data-processing paperwork needs an update, or how a negotiated enterprise rate will be treated.

The transaction should prompt a review of your operating and commercial dependencies. Keep the production architecture in place while that review produces evidence about whether a change is necessary.

Map the Daily surface your video product actually uses

Daily describes its current video product as SDKs and APIs for real-time video and audio applications across web, React Native, iOS, Android, Python, and Flutter, with recording, streaming, analytics, REST, and webhook integrations. The current product page is the right starting point for an inventory, because “we use Daily” usually hides several separate dependencies.

  • Client entry: identify every web and mobile SDK, version, authentication flow, and supported platform.
  • Room control: record how your backend creates rooms, issues meeting tokens, applies permissions, and reacts to webhooks.
  • Media features: list recording, live streaming, screen sharing, transcription, layouts, analytics, and any custom tracks your product depends on.
  • Operations: document regions, domains, firewall rules, TURN behaviour, monitoring, incident escalation, and the people who can change the integration.
  • Data handling: trace live media, recordings, transcripts, logs, and metadata from capture to deletion.

This inventory separates a replaceable API call from an operational dependency. A team may be able to replace room creation in a sprint while still facing a much larger migration for recording, mobile behaviour, stored media, or customer-facing support commitments.

Check the contract before you check the roadmap

Pull the documents that govern your account: the signed Order Form, pricing schedule, SLA, support package, DPA, security addendum, renewal notice, and any amendments. The public pages describe a baseline. They cannot establish what an enterprise customer has negotiated.

Ask your account team or legal owner to answer these questions in writing:

  1. Which legal entity is your contracting party, and what change-of-control language applies?
  2. Which Daily products, environments, regions, and support channels are covered?
  3. What availability target, exclusions, response targets, and service credits apply to your account?
  4. Can pricing, usage allowances, or included features change at renewal, and what notice is required?
  5. How can either party terminate, and what happens to prepaid balances, recordings, and account data?
  6. Which export and deletion obligations apply to media, recordings, logs, and metadata?
  7. Who owns the escalation if a production incident crosses the transaction or account teams?

Daily’s public pricing page says that free developer support includes email and Discord support with best-effort response times, while enterprise support includes packages with SLAs. It does not publish the terms of your enterprise SLA. Treat the signed support package as the authoritative document.

The practical recommendation is simple: ask for clarification where your documents are silent, and preserve the written answer with the rest of your vendor record.

Establish your current cost baseline

Do not compare a future quote with a remembered estimate. Export invoices, usage reports, rate cards, and credits now. Then separate the bill into the meters your product actually creates.

Daily’s public video pricing page lists 10,000 free participant minutes each month on the applicable plan. It also displays a volume-rate range of $0.0015–$0.004 per participant-minute, with automatic discounts as usage grows and possible enterprise discounts. Your account may use a different rate or set of included allowances.

For a comparable internal baseline, calculate:

participant minutes × contracted participant-minute rate
+ recording minutes × recording rate
+ recording storage
+ streaming or other enabled meters
+ support or platform charges

Daily lists recording at $0.01349 per recorded minute, plus $0.003 per minute for additional storage. The page describes these as public list prices and notes that discounts may apply at scale. Check whether your account uses those meters, a negotiated rate, or a different recording configuration.

Keep three baselines separately: the last twelve months of actual spend, the current public list-price equivalent, and the cost of moving. Mixing them makes a vendor review look precise while hiding the assumptions that matter.

Test the parts of Daily you cannot replace with a hostname

An API abstraction helps when your application calls one interface for rooms, tokens, participants, and events. It does not automatically make media portable. Test the functions that affect customers and operations.

DependencyWhat to verifyEvidence to keep
Room and participant controlCan another provider reproduce your permissions, identity model, webhooks, and failure handling?Token code, webhook handlers, integration tests
Client SDKsWhich browser and mobile behaviours are product-critical?Supported-device matrix and release versions
RecordingCan you retrieve, move, replay, and delete recordings in the required format?Sample exports, storage paths, retention jobs
Network pathWhich regions, TURN paths, domains, and firewall rules are required?Network diagrams and runbooks
SupportWho responds during a degraded call experience, and how is severity assigned?SLA, escalation contacts, incident records

Daily says its Global Mesh Network has 75+ points of presence. It also reports a median first-hop latency of less than 13 milliseconds. Those are vendor-reported product figures; the product page does not specify the measurement period, geography, percentile methodology beyond “median,” or workload. Use them as questions for your own measurement. They do not establish an end-to-end experience for every participant.

Daily also says its service can support up to 100,000 simultaneous calls or sessions with a card on file, subject to the number of programmatic rooms on a domain, with higher requirements handled through contact with Daily. If your workload approaches that boundary, document the account-specific limit rather than relying on the public headline.

Verify recording ownership and exit paths

Recording is often the least portable part of a video integration because it combines media processing, storage, permissions, retention, playback, and downstream workflows.

Daily says recordings can be saved in Daily’s Cloud or in a customer’s own Amazon S3 storage. The recording documentation establishes that storage option; it does not prove that every customer uses customer-owned storage or that every recording mode has identical export behaviour.

Run an exit test with a representative recording. Confirm the original and processed formats, timestamps, participant identity, layout, captions or transcripts if enabled, access controls, deletion behaviour, and the time required to retrieve the asset. Then test whether your downstream systems can consume the export without a Daily-specific transformation.

Also decide where new recordings should land during a transition. Customer-owned storage can reduce dependence on a provider’s retention layer, but it does not remove the need to recreate recording triggers, permissions, processing, monitoring, and playback.

Build, buy, or deploy?

The acquisition is a useful point to choose the right level of control. Your options are different engineering commitments, so compare them against the dependency map rather than treating “open” or “commercial” as a guarantee of portability.

Build on open-source components

Building on open-source WebRTC and SFU components can give your team control over media routing, TURN, recording, deployment locations, and release timing. It also transfers responsibility for capacity planning, upgrades, security response, observability, mobile and browser compatibility, incident response, and regional operations.

This path fits a team that already operates real-time infrastructure and needs control over the media path and surrounding services. It is a poor fit when the main requirement is to ship a reliable video feature without becoming a media-infrastructure operator.

Buy a cloud video API

A cloud video API keeps the media service outside your infrastructure while giving your product SDKs, APIs, recording, and provider-operated network services. This is usually the fastest path when your team values implementation speed and has a manageable portability risk.

The trade-off is the dependency you are reviewing now: pricing, product direction, support, regions, and contract terms sit partly outside your control. A provider-neutral application boundary, regular export tests, and a current cost baseline make that dependency easier to manage.

Deploy a commercial platform

A commercial platform that can be deployed in your environment can give you deeper control over the media path, TURN, recording processing, and recording storage while retaining vendor software and support. It still requires infrastructure ownership, upgrades, capacity planning, and an operating model that matches your availability requirements.

Samvyo is one option in this category. It is based on SFU architecture and provides embeddable SDKs with OEM and white-label depth, alongside a managed-service option. It fits teams that need control over the media path, TURN, and recording processing and storage, while keeping the core platform resilient by design. It is less suitable when a fully provider-operated cloud API is the only acceptable operating model or when a team does not want responsibility for deployment choices.

The recommendation for a current Daily customer is to keep the existing service while it meets your product and contractual needs, complete the audit now, and prepare a tested alternative only in proportion to your exposure. The limits are clear: a public announcement cannot tell you what your contract says, a public rate card cannot tell you your negotiated price, and an abstraction layer cannot prove that media, recording, or operations can move without work.

Bottom Line

Deepgram’s planned acquisition of Daily is a reason to document your dependency while your video product is stable. Review the announcement against your contract, SLA, pricing, recording path, support model, and migration evidence.

That work gives you two useful outcomes: confidence that Daily still fits your requirements, or a measured plan for changing providers. Either outcome is stronger than predicting a roadmap from the transaction alone.

What’s Next

Daily is one example of a broader shift in which real-time video capabilities are being folded into larger AI and conversation platforms. The companion analysis maps the dated product and platform moves behind that shift.

Read it when you need to compare Daily’s situation with changes at other major video API providers.

FAQs

What does Deepgram acquiring Daily mean for Daily video customers?

It means Daily’s ownership and product context are changing through the announced acquisition, while Daily’s announcement says its current customers should expect no immediate change and that current products and services will continue to be sold, supported, and developed. Customers should still verify their own contract, SLA, pricing, support, data-processing terms, and roadmap dependencies.

Should I migrate from Daily because Deepgram is acquiring it?

No immediate migration follows from the announcement alone. Keep Daily if it meets your requirements, but complete a dependency and exit audit now so that a future change can be evaluated against evidence from your account rather than assumption.

How much does Daily video cost?

Daily’s public pricing page displays a volume range of $0.0015–$0.004 per participant-minute and lists 10,000 free participant minutes each month on the applicable plan. It also lists recording at $0.01349 per recorded minute plus $0.003 per minute for additional storage; enterprise discounts, account terms, and enabled meters can change the amount a customer actually pays.

What should I ask Daily about my SLA after the acquisition?

Ask which legal entity, products, regions, availability target, exclusions, response targets, service credits, and escalation paths your signed SLA covers. Public support descriptions do not establish the terms of an individual enterprise account, so retain the written answer with your Order Form and support package.

Can Daily recordings be stored outside Daily’s cloud?

Daily documents an option to save recordings in a customer’s own Amazon S3 storage as well as in Daily’s Cloud. Verify which recording modes, export formats, processing steps, retention rules, and deletion workflows apply to your implementation before treating that option as an exit plan.

Is building on an open-source SFU safer than using a cloud video API?

Open-source infrastructure can give you more control over the media path, TURN, recording, and deployment, but it also gives your team responsibility for operating those components. A cloud API reduces infrastructure work while creating dependence on the provider’s pricing, support, product direction, and contract terms; the safer choice depends on which responsibility your team can reliably carry.

Where does Samvyo fit for a Daily customer?

Samvyo is one commercial-platform option for a team that wants embeddable SDKs, OEM and white-label depth, and control over the media path, TURN, and recording processing and storage. It is a poor fit if the team requires only a fully provider-operated cloud API and does not want deployment responsibility.