Paperspace

Name: Paperspace
Brand: Paperspace (now DigitalOcean)
Rating: 2.1 (134 reviews)

Paperspace is a cloud compute platform that provides GPU-accelerated virtual machines, notebook-based ML development (Gradient), and a simplified UI/API for provisioning and managing GPU workloads.

Est 2014|Valuation $100.4M|Raised $27.8M|23 Ref Customers

LinkedIn 4K|Web traffic 312K|FTEs 33|HQ New York, New York, US

MarketsData Science & Machine Learning Platforms Development Environments Cloud Infrastructure Providers (IaaS/PaaS)

Published Oct 15, 2026

01Common complaints

1. **When enterprise-grade compliance transparency is required out of the box.** - Public materials emphasize datacenter certifications but do not supply clear, service-level attestations in a buyer-friendly format. 2. **When billing surprises cannot be tolerated.** - Multiple review sources cite unexpected charges associated with instance lifecycle semantics and limited alerting. 3. **When guaranteed GPU capacity (especially premium GPUs) is required on demand.** - H100 availability is region-limited and typically requires a support request and capacity approval. 4. **When robust support continuity and ticket history are required without contractual safeguards.** - Documentation indicates changes to support email domains and potential unavailability of historic ticket data for ongoing issues. 5. **When a stable, standalone product identity is a procurement requirement.** - DigitalOcean recommends new users explore DigitalOcean GPU offerings; organizations seeking a product with a fixed standalone roadmap should consider transition risk.

Deal disqualifiers (poor fit)

Requirement for publicly available, independently verifiable compliance artifacts without NDAs.
Workloads that require guaranteed immediate capacity for specific GPUs without manual approval processes.
Low tolerance for billing ambiguity or a need for default guardrails and automated cost enforcement.

Product and operational limitations evidenced in public sources

Incidents occur; the status page documents service disruptions (e.g., Jan 16, 2026 incident affecting Gradient/Core Machines).
Premium GPU access gating: H100 access requires a support request and is region-limited to NY2 as documented.
Support process changes: documentation indicates that support email domains changed effective Jan 14, 2026, and that past ticket history may be unavailable, requiring customers to reference prior ticket IDs.

Vendor-acknowledged roadmap / future direction

DigitalOcean's messaging indicates a migration of Paperspace capabilities into DigitalOcean-branded AI/ML offerings; legacy Paperspace products remain accessible for existing users while new users are directed to DigitalOcean’s GPU products.

02Why they buy

Documented strengths (based on user feedback and documentation)

Rapid provisioning of notebooks and virtual machines, which supports iterative development and experimentation.
Historically low entry pricing for individuals via a Pro tier, with actual costs dependent on utilization patterns.
An integrated ML workflow concept (Gradient) that combines notebooks, workflows, and deployment features.

Claims that require validation

Marketing statements about large GPU catalogs and performance multipliers are not always accompanied by a public methodology or independent benchmarks; they should be validated for a specific workload and region.

Areas where competitors may have clearer advantages

Public compliance artifacts and enterprise readiness documentation.
Pricing transparency and built-in FinOps guardrails.

03Field reports

Below are representative excerpts and paraphrases from public review platforms. These examples are not exhaustive and reflect publicly accessible feedback.

Review platform: Capterra (balanced)

Positive examples (paraphrased): - Reviewers praised a smoother experience relative to Colab for uploading datasets and creating Jupyter notebooks. - Some reviewers indicated the platform met their needs for student and personal ML work, citing access to GPU time at the Pro tier price point. - Other positive feedback noted responsive customer support in individual cases while observing that lower-tier support can be slower. Negative examples (paraphrased): - Multiple reviewers report service unavailability, login issues, and an unreliable console. - Other low-rated reviews allege billing problems and slow support response times. - Some reviewers reported continued charges despite attempting to stop services. Interpretation: Capterra feedback shows a split pattern—some users report successful setup and effective cost management, while others report availability, billing, and support issues that impacted their continued use.

Trustpilot (skewed negative)

The product is poorly rated on Trustpilot, with repeated complaints regarding unexpected charges, confusion around free-tier limits, the requirement to manually shut down machines, and difficulties obtaining timely support.

Community/Social: Reddit (anecdotal)

Discussion threads include strong negative anecdotes about billing and quality, while other participants offer mitigation steps such as contacting support to remove saved payment information. These are anecdotal and reflect individual user experiences.

Recurring reasons cited for choosing the platform

Ease of provisioning notebooks and virtual machines compared with some alternatives.
Affordability for students and individual projects based on the Pro tier plus paid usage.
Relatively straightforward access to GPU resources compared with hyperscaler onboarding.

Recurring reasons cited for leaving / churn

Billing transparency and unexpected charges (instances left running, unclear defaults, limited alerting).
Slow or inconsistent support responsiveness.
Reliability and availability complaints for critical workflows.
Capacity constraints and approval requirements for certain machine types.

Vendor-published materials and press

- Public materials include press releases and vendor pages that describe GPU availability and performance, particularly around announcements for high-end GPUs and partner program activities. Critical evaluation: - These materials emphasize capability and performance potential but typically lack independently audited ROI methodologies. - Marketing claims (for example, large GPU catalogs or performance multipliers) are not consistently defined in methodology or scope and should be validated in specific customer environments.

Reliability & SLA claims

Enterprise product materials reference an uptime SLA. The enforceability and relevance of that SLA depend on contract scope, exclusions, and whether the customer has negotiated the SLA within a contract. Public status records show incidents; assessment should compare incident frequency and recovery characteristics across options.

04Best fit

Company sizes

Individuals and students for prototyping and learning.
Startups and SMBs (approximately 5–500 employees) conducting periodic fine-tuning, inference, or internal ML projects.
Enterprises (500+ employees) primarily via negotiated contracts with explicit SLA and security add-ons.

Industries

AI SaaS and developer tools
Media and generative AI
E-commerce and recommendation systems
Computer vision and analytics

Roles and personas

Decision makers: CTO, Head of Engineering, Head of Data/ML
End users: ML engineers, data scientists, research engineers, DevOps/Platform engineers

Qualification criteria (good fit)

Fast notebook-to-GPU provisioning and an integrated notebook/VM workflow are priorities.
The customer can implement internal cost controls (shutdown policies, budgets, automation) to manage unexpected usage costs.
The buyer understands the platform is in a brand transition to DigitalOcean and accepts associated product transition risk.

05The pitch

What the product does (core value proposition)

Paperspace is a cloud compute platform historically known for: - GPU-accelerated virtual machines ("Machines") for AI/ML training, fine-tuning, and inference. - Notebook-based development (via the Gradient platform) oriented around progressing from experiment to production. - A UI/API interface intended to be less complex than hyperscalers for certain GPU workloads. Important market reality: Paperspace was acquired by DigitalOcean and has been increasingly integrated into DigitalOcean’s AI/GPU offerings; the Paperspace name is positioned as a legacy/transition brand for many users.

Target market and positioning

- **Primary users:** ML engineers, data scientists, AI researchers, startups, and SMBs that require GPU compute without constructing bespoke infrastructure. - **Secondary users:** VDI / remote desktop workstation use cases (legacy positioning), including design and visualization. - **Positioning:** Market materials position the product as an interface with lower operational complexity for certain GPU workflows compared with major cloud providers, though user feedback is mixed on reliability, support, and billing transparency.

Get a comprehensive analysis of Paperspace including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →

Published Oct 15, 2026

01Common complaints

When enterprise-grade compliance transparency is required out of the box.
- Public materials emphasize datacenter certifications but do not supply clear, service-level attestations in a buyer-friendly format.
When billing surprises cannot be tolerated.
- Multiple review sources cite unexpected charges associated with instance lifecycle semantics and limited alerting.
When guaranteed GPU capacity (especially premium GPUs) is required on demand.
- H100 availability is region-limited and typically requires a support request and capacity approval.
When robust support continuity and ticket history are required without contractual safeguards.
- Documentation indicates changes to support email domains and potential unavailability of historic ticket data for ongoing issues.
When a stable, standalone product identity is a procurement requirement.
- DigitalOcean recommends new users explore DigitalOcean GPU offerings; organizations seeking a product with a fixed standalone roadmap should consider transition risk.

Deal disqualifiers (poor fit)

Requirement for publicly available, independently verifiable compliance artifacts without NDAs.
Workloads that require guaranteed immediate capacity for specific GPUs without manual approval processes.
Low tolerance for billing ambiguity or a need for default guardrails and automated cost enforcement.

Product and operational limitations evidenced in public sources

Incidents occur; the status page documents service disruptions (e.g., Jan 16, 2026 incident affecting Gradient/Core Machines).
Premium GPU access gating: H100 access requires a support request and is region-limited to NY2 as documented.
Support process changes: documentation indicates that support email domains changed effective Jan 14, 2026, and that past ticket history may be unavailable, requiring customers to reference prior ticket IDs.

Vendor-acknowledged roadmap / future direction

DigitalOcean's messaging indicates a migration of Paperspace capabilities into DigitalOcean-branded AI/ML offerings; legacy Paperspace products remain accessible for existing users while new users are directed to DigitalOcean’s GPU products.

02Why they buy

Documented strengths (based on user feedback and documentation)

Rapid provisioning of notebooks and virtual machines, which supports iterative development and experimentation.
Historically low entry pricing for individuals via a Pro tier, with actual costs dependent on utilization patterns.
An integrated ML workflow concept (Gradient) that combines notebooks, workflows, and deployment features.

Claims that require validation

Marketing statements about large GPU catalogs and performance multipliers are not always accompanied by a public methodology or independent benchmarks; they should be validated for a specific workload and region.

Areas where competitors may have clearer advantages

Public compliance artifacts and enterprise readiness documentation.
Pricing transparency and built-in FinOps guardrails.

03Field reports

Below are representative excerpts and paraphrases from public review platforms. These examples are not exhaustive and reflect publicly accessible feedback.

Review platform: Capterra (balanced)

Positive examples (paraphrased):

Reviewers praised a smoother experience relative to Colab for uploading datasets and creating Jupyter notebooks.
Some reviewers indicated the platform met their needs for student and personal ML work, citing access to GPU time at the Pro tier price point.
Other positive feedback noted responsive customer support in individual cases while observing that lower-tier support can be slower.

Negative examples (paraphrased):

Multiple reviewers report service unavailability, login issues, and an unreliable console.
Other low-rated reviews allege billing problems and slow support response times.
Some reviewers reported continued charges despite attempting to stop services.

Interpretation: Capterra feedback shows a split pattern—some users report successful setup and effective cost management, while others report availability, billing, and support issues that impacted their continued use.

Trustpilot (skewed negative)

The product is poorly rated on Trustpilot, with repeated complaints regarding unexpected charges, confusion around free-tier limits, the requirement to manually shut down machines, and difficulties obtaining timely support.

Community/Social: Reddit (anecdotal)

Discussion threads include strong negative anecdotes about billing and quality, while other participants offer mitigation steps such as contacting support to remove saved payment information. These are anecdotal and reflect individual user experiences.

Recurring reasons cited for choosing the platform

Ease of provisioning notebooks and virtual machines compared with some alternatives.
Affordability for students and individual projects based on the Pro tier plus paid usage.
Relatively straightforward access to GPU resources compared with hyperscaler onboarding.

Recurring reasons cited for leaving / churn

Billing transparency and unexpected charges (instances left running, unclear defaults, limited alerting).
Slow or inconsistent support responsiveness.
Reliability and availability complaints for critical workflows.
Capacity constraints and approval requirements for certain machine types.

Vendor-published materials and press

Public materials include press releases and vendor pages that describe GPU availability and performance, particularly around announcements for high-end GPUs and partner program activities.

Critical evaluation:

These materials emphasize capability and performance potential but typically lack independently audited ROI methodologies.
Marketing claims (for example, large GPU catalogs or performance multipliers) are not consistently defined in methodology or scope and should be validated in specific customer environments.

Reliability & SLA claims

Enterprise product materials reference an uptime SLA. The enforceability and relevance of that SLA depend on contract scope, exclusions, and whether the customer has negotiated the SLA within a contract. Public status records show incidents; assessment should compare incident frequency and recovery characteristics across options.

04Best fit

Company sizes

Individuals and students for prototyping and learning.
Startups and SMBs (approximately 5–500 employees) conducting periodic fine-tuning, inference, or internal ML projects.
Enterprises (500+ employees) primarily via negotiated contracts with explicit SLA and security add-ons.

Industries

AI SaaS and developer tools
Media and generative AI
E-commerce and recommendation systems
Computer vision and analytics

Roles and personas

Decision makers: CTO, Head of Engineering, Head of Data/ML
End users: ML engineers, data scientists, research engineers, DevOps/Platform engineers

Qualification criteria (good fit)

Fast notebook-to-GPU provisioning and an integrated notebook/VM workflow are priorities.
The customer can implement internal cost controls (shutdown policies, budgets, automation) to manage unexpected usage costs.
The buyer understands the platform is in a brand transition to DigitalOcean and accepts associated product transition risk.

05The pitch

What the product does (core value proposition)

Paperspace is a cloud compute platform historically known for:

GPU-accelerated virtual machines ("Machines") for AI/ML training, fine-tuning, and inference.
Notebook-based development (via the Gradient platform) oriented around progressing from experiment to production.
A UI/API interface intended to be less complex than hyperscalers for certain GPU workloads.

Important market reality: Paperspace was acquired by DigitalOcean and has been increasingly integrated into DigitalOcean’s AI/GPU offerings; the Paperspace name is positioned as a legacy/transition brand for many users.

Target market and positioning

Primary users: ML engineers, data scientists, AI researchers, startups, and SMBs that require GPU compute without constructing bespoke infrastructure.
Secondary users: VDI / remote desktop workstation use cases (legacy positioning), including design and visualization.
Positioning: Market materials position the product as an interface with lower operational complexity for certain GPU workflows compared with major cloud providers, though user feedback is mixed on reliability, support, and billing transparency.

Get a comprehensive analysis of Paperspace including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →