Nscale growth playbook

Publish /pricing with three blocks: serverless inference prices per million tokens for the models already live (the data the List endpoints API returns), "from" prices per GPU-hour for on-demand instances, and a reserved-capacity block with a "Talk to us for committed pricing" form, as Lambda does. Link it from the main navigation and every GPU node page, add Product structured data, and make it the landing page for price-intent search ads ("H100 cloud price", "GB300 rental").

Price is the first filter for developers and AI start-ups. A public "from" price keeps Nscale in the comparison, lowers cost per lead on price-intent keywords and sends sales better-qualified reserve requests, while committed deals stay negotiated.

[ ] Serverless price per 1M input and output tokens, per model
[ ] On-demand "from" price per GPU-hour, per GPU type
[ ] Reserved capacity block with a form (GPU type, count, term)
[ ] Token-cost calculator
[ ] Link from main navigation and all GPU node pages
[ ] Product / Offer structured data
[ ] "Last updated" date
[ ] Separate conversion events in GA4 and HubSpot

• Prices change often

  Feed serverless prices from the API and date the GPU block.

• Competitors undercut publicly

  Show "from" prices and sell the full stack, sovereignty and reserved terms.

• HiringPerformance Marketing Manager (London, posted October 7, 2026): owns "LinkedIn, Google Ads, YouTube, and other paid channels—including developer-centric platforms" for "account-based and developer-focused demand"; manages "CTR, CPL, CVR, SQL rate"; maintains tracking in "GA4, server-side GTM, and HubSpot" and builds "an experimentation roadmap for creative, targeting, and channel testing".Source
• Your websitenscale.com has no pricing page: the main navigation offers "Reserve GPUs", "Start Building" and "Contact". The GPU node pages (H100, H200, GB200, GB300 NVL72, Vera Rubin NVL72) list NVIDIA specifications with "Reserve GPUs" and "Get in touch" buttons and no price. The docs say AI services "use a credit-based billing system" ("To access AI services, you'll need to purchase credits"), and per-model prices appear only through the List endpoints API ("pricing as decimal USD").Source
• Competitor researchRivals publish prices on public pages: Lambda's pricing page ("Clear, straightforward pricing for Instances, 1-Click Clusters, and Superclusters. Contact us for reserved capacity at our lowest prices.") lists HGX B200 clusters from $8.87 to $9.86 per GPU-hour; Nebius lists HGX H100 on-demand at $4.50 per GPU-hour; CoreWeave lists an 8-GPU HGX H100 node at $49.24 per hour on demand.See the research Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source

Replace the three quotes with named, approved proof: the Microsoft and OpenAI programs already announced, plus two or three customer quotes from AI natives or telcos. Add a /customers page with short stories (workload, GPUs, time to production) and move the investor quotes to /about.

Buyers trust peers more than investors. Customer proof on the homepage also gives sales and paid ads something specific to say.

• Customers decline to be named

  Use anonymized stories with workload and scale.

• Your websiteHomepage (checked October 21, 2026): H1 "The engine of superintelligence"; the section "Trusted by leading AI labs and enterprises to run critical workloads" shows three quotes, from the CEO of investor Aker ASA, the founder of investor G Squared and the UK AI Minister, and no customer; the "Industry solutions" section and footer list only Telco and AI Native. The page promotes "abundant and renewable power resources", "Low PUE" and data centers in Norway, the UK, Iceland, Portugal and the US (Texas, West Virginia, North Carolina).Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source
• HiringUS Product Marketing Manager (New York or San Francisco): creates "pitch decks, one-pagers, competitor comparisons, FAQs, and talk tracks", supports "website and product page updates by helping develop copy, proof points, and supporting assets" and the "development of customer stories, proof points, analyst inputs".Source
• Press releaseNscale raised $2 billion in a Series C in March 2026 at a $14.6 billion valuation, contracted about 200,000 NVIDIA GB300 GPUs with Microsoft (October 2025), and on September 18, 2026 filed a Form S-1 for a proposed New York Stock Exchange listing under "NSCL". Series C backers include Dell, Lenovo, Nokia and NVIDIA, and the newsroom announces a Singtel partnership to unlock GPU capacity.Source

Re-encode the hero at a lower bitrate (aim under 3 MB), replace the five stock loops with posters that play only on hover or when scrolled into view on desktop, and serve no video on mobile data. Re-run PageSpeed and watch page weight in the monitoring.

A lighter page loads faster on mobile networks abroad and keeps the brand feel. It also cuts CDN bills.

• The brand team likes the motion

  Keep the hero, change only the file size and when it loads.

• PageSpeed testGoogle PageSpeed for the nscale.com homepage (October 2026): real Chrome visitors pass all three Core Web Vitals (page LCP 1.6 s, INP 137 ms, CLS 0.04; across the whole site LCP is 1.26 s). The mobile lab test scores 87, but the page downloads 42.4 MB. Most of that is video: a 3D hero clip (15.3 MB in the Safari version, 5.4 MB in the Chrome version) and five looping stock-footage backgrounds of 1.6 MB to 7.5 MB each.Source
• HiringPerformance Marketing Manager (London, posted October 7, 2026): owns "LinkedIn, Google Ads, YouTube, and other paid channels—including developer-centric platforms" for "account-based and developer-focused demand"; manages "CTR, CPL, CVR, SQL rate"; maintains tracking in "GA4, server-side GTM, and HubSpot" and builds "an experimentation roadmap for creative, targeting, and channel testing".Source

Fill the organization card with a short pitch, a link to the pricing page and docs with UTM tags, publish a Space that compares the 19 models on Nscale, and offer free starter credits for developers coming from Hugging Face. Report which models bring sign-ups.

Developers already reach Nscale through Hugging Face routing; a direct path lets them become Nscale customers instead of anonymous traffic.

• Credit abuse

  Limit credits per verified account.

• Hugging FaceNscale has a verified Hugging Face organization (64 followers) and serves 19 models as a Hugging Face inference provider, including Llama 3.1 8B Instruct, FLUX.1-schnell, gpt-oss-120b and the Qwen3 family.Source
• HiringEcosystem Marketing Manager (New York or San Francisco, posted September 8, 2026): builds Nscale's presence "with an initial focus on PyTorch, Hugging Face, Ray, vLLM", turning relationships into "joint technical content, demos, integrations, workshops"; one of four pillars is developer adoption: "Create stronger journeys from the tools and communities developers already use into Nscale."Source
• Your websitenscale.com has no pricing page: the main navigation offers "Reserve GPUs", "Start Building" and "Contact". The GPU node pages (H100, H200, GB200, GB300 NVL72, Vera Rubin NVL72) list NVIDIA specifications with "Reserve GPUs" and "Get in touch" buttons and no price. The docs say AI services "use a credit-based billing system" ("To access AI services, you'll need to purchase credits"), and per-model prices appear only through the List endpoints API ("pricing as decimal USD").Source

When an event ends, swap the countdown for a recap: talk recordings or slides, a short summary, the offer shown at the booth and a "book a follow-up" form. Use the same page in follow-up emails.

Visitors from events arrive at a useful page instead of a dead one, and sales follow-ups have a link to send.

• No recordings

  Use slides and a written summary.

• Your channelsYouTube @NscaleCloud: 654 subscribers, 34 videos, about 3.9 million total views. The PyTorch Conference 2026 event page (October 20-21, San Jose) still shows a countdown at zero after the event.Source
• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• HiringDemand Generation & Lifecycle Marketing Manager (San Francisco, still open): owns "account-based demand generation programs" with "buying-committee mapping", builds "nurture journeys by persona, funnel stage, product interest, intent and account fit", runs pre-event and post-event follow-up with Global Events, and will "Stand up entirely new motions from a blank page where no program, template or playbook exists yet".Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source

Publish a public summary page with the five headline charts, citation text and a press kit, and keep the full report gated. Use the charts in social posts and paid ads that link to the summary.

Public numbers earn links and citations, while the gate still captures readers who want the detail.

• Fewer gated leads

  Gate the full data and the methodology.

• Your websiteState of AI Natives 2026: a survey of "500 AI-native businesses in the United States"; "76% saying infrastructure limited their ability to ship features in the past year" and "77% of AI Natives are highly concerned about data residency and compliance". The full report is behind a form.Source
• HiringSenior Social Media Lead (posted August 27, 2026): "Define and execute audience growth strategies that scale communities into the millions", "Design and implement AI-powered content pipelines", and "Own reporting frameworks that connect social to pipeline and brand outcomes" across LinkedIn, X and YouTube.Source
• HiringDemand Generation & Lifecycle Marketing Manager (San Francisco, still open): owns "account-based demand generation programs" with "buying-committee mapping", builds "nurture journeys by persona, funnel stage, product interest, intent and account fit", runs pre-event and post-event follow-up with Global Events, and will "Stand up entirely new motions from a blank page where no program, template or playbook exists yet".Source

Publish /industries/financial-services and /industries/healthcare-life-sciences: workloads (risk models, trading research; drug discovery, genomics, imaging), compliance and sovereignty, an ROI model, and a briefing form. Use them as ABM landing pages.

Regulated buyers want to see their own use cases and controls. A vertical page also makes paid targeting cheaper because the message matches the audience.

• Claims about regulated use

  Only state certifications Nscale holds.

• HiringVertical go-to-market leads: a Head of Financial Services GTM ("author our compliance narrative, build sophisticated ROI models tailored specifically to the unique needs of trading desks", with "SOC2, sovereign data requirements") and a Head of Healthcare & Life Sciences GTM ("drug discovery, genomics, molecular simulation (e.g., AlphaFold), and advanced medical imaging"; "a critical zero-to-one challenge").Source
• Your websiteHomepage (checked October 21, 2026): H1 "The engine of superintelligence"; the section "Trusted by leading AI labs and enterprises to run critical workloads" shows three quotes, from the CEO of investor Aker ASA, the founder of investor G Squared and the UK AI Minister, and no customer; the "Industry solutions" section and footer list only Telco and AI Native. The page promotes "abundant and renewable power resources", "Low PUE" and data centers in Norway, the UK, Iceland, Portugal and the US (Texas, West Virginia, North Carolina).Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source
• HiringPerformance Marketing Manager (London, posted October 7, 2026): owns "LinkedIn, Google Ads, YouTube, and other paid channels—including developer-centric platforms" for "account-based and developer-focused demand"; manages "CTR, CPL, CVR, SQL rate"; maintains tracking in "GA4, server-side GTM, and HubSpot" and builds "an experimentation roadmap for creative, targeting, and channel testing".Source

Turn each competitor comparison into a factual public page: GPU types, pricing model, regions and sovereignty, managed services (Kubernetes, Slurm, inference), energy and data center ownership, and switching help.

Comparison searches are high intent, and fair tables also feed AI assistants' answers.

• Out-of-date claims

  Date each page and cite sources.

• HiringUS Product Marketing Manager (New York or San Francisco): creates "pitch decks, one-pagers, competitor comparisons, FAQs, and talk tracks", supports "website and product page updates by helping develop copy, proof points, and supporting assets" and the "development of customer stories, proof points, analyst inputs".Source
• Competitor researchRivals publish prices on public pages: Lambda's pricing page ("Clear, straightforward pricing for Instances, 1-Click Clusters, and Superclusters. Contact us for reserved capacity at our lowest prices.") lists HGX B200 clusters from $8.87 to $9.86 per GPU-hour; Nebius lists HGX H100 on-demand at $4.50 per GPU-hour; CoreWeave lists an 8-GPU HGX H100 node at $49.24 per hour on demand.See the research Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source

Publish a "Run Ray on Nscale" quickstart, a cookbook for distributed training and batch inference, and a credit offer for Ray users. Run workshops at Ray meetups and track sign-ups.

Ray users already need many GPUs; a clean path turns that community into customers.

• Deal timing uncertain

  Build on open-source Ray, not the acquisition.

• Press releaseOctober 20, 2026: Nscale joined the PyTorch Foundation as a Platinum member. The release says Nscale came to the Foundation through Ray and refers to "Nscale's planned acquisition of Anyscale, the company behind Ray", with a commitment to keep investing in Ray and its contributor community.Source
• HiringEcosystem Marketing Manager (New York or San Francisco, posted September 8, 2026): builds Nscale's presence "with an initial focus on PyTorch, Hugging Face, Ray, vLLM", turning relationships into "joint technical content, demos, integrations, workshops"; one of four pillars is developer adoption: "Create stronger journeys from the tools and communities developers already use into Nscale."Source
• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source

Instrument the sign-up flow (sign-up, credit purchase, first key, first call), publish a 5-minute quickstart, offer small free credits so the first call needs no card, and review drop-off weekly.

Activation is the biggest lever for developer growth; every paid click is worth more if more people make a first call.

• Free credits attract abuse

  Verify accounts and cap usage.

• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source
• Your websitenscale.com has no pricing page: the main navigation offers "Reserve GPUs", "Start Building" and "Contact". The GPU node pages (H100, H200, GB200, GB300 NVL72, Vera Rubin NVL72) list NVIDIA specifications with "Reserve GPUs" and "Get in touch" buttons and no price. The docs say AI services "use a credit-based billing system" ("To access AI services, you'll need to purchase credits"), and per-model prices appear only through the List endpoints API ("pricing as decimal USD").Source
• HiringPerformance Marketing Manager (London, posted October 7, 2026): owns "LinkedIn, Google Ads, YouTube, and other paid channels—including developer-centric platforms" for "account-based and developer-focused demand"; manages "CTR, CPL, CVR, SQL rate"; maintains tracking in "GA4, server-side GTM, and HubSpot" and builds "an experimentation roadmap for creative, targeting, and channel testing".Source

Connect console events to HubSpot and trigger emails: after the first call (next steps), at low credit (auto-recharge), after heavy serverless use (Dedicated Inference) and when usage stops.

Messages based on behavior beat time-based sequences, and the step to dedicated capacity is where revenue grows.

• Too many emails

  Cap at one a week.

• HiringDemand Generation & Lifecycle Marketing Manager (San Francisco, still open): owns "account-based demand generation programs" with "buying-committee mapping", builds "nurture journeys by persona, funnel stage, product interest, intent and account fit", runs pre-event and post-event follow-up with Global Events, and will "Stand up entirely new motions from a blank page where no program, template or playbook exists yet".Source
• Your websiteDedicated Inference page: "Deploy supported models on reserved Nscale GPU capacity and access managed endpoints for production inference", presented as "a third option" between running the serving stack yourself and adding a separate inference platform; the call to action is "Talk to an expert".Source
• Your websitenscale.com has no pricing page: the main navigation offers "Reserve GPUs", "Start Building" and "Contact". The GPU node pages (H100, H200, GB200, GB300 NVL72, Vera Rubin NVL72) list NVIDIA specifications with "Reserve GPUs" and "Get in touch" buttons and no price. The docs say AI services "use a credit-based billing system" ("To access AI services, you'll need to purchase credits"), and per-model prices appear only through the List endpoints API ("pricing as decimal USD").Source

Publish two engineering videos a month (benchmarks, how-tos, data center tours) with links to docs and pricing, and judge the channel on subscribers and sign-ups rather than views.

Developers subscribe to useful content, and subscribers come back without ad spend.

• Slow growth

  Commit to six months.

• Your channelsYouTube @NscaleCloud: 654 subscribers, 34 videos, about 3.9 million total views. The PyTorch Conference 2026 event page (October 20-21, San Jose) still shows a countdown at zero after the event.Source
• HiringSenior Social Media Lead (posted August 27, 2026): "Define and execute audience growth strategies that scale communities into the millions", "Design and implement AI-powered content pipelines", and "Own reporting frameworks that connect social to pipeline and brand outcomes" across LinkedIn, X and YouTube.Source
• HiringPerformance Marketing Manager (London, posted October 7, 2026): owns "LinkedIn, Google Ads, YouTube, and other paid channels—including developer-centric platforms" for "account-based and developer-focused demand"; manages "CTR, CPL, CVR, SQL rate"; maintains tracking in "GA4, server-side GTM, and HubSpot" and builds "an experimentation roadmap for creative, targeting, and channel testing".Source

Add a partner page that explains referral rewards and MDF, track referred leads in HubSpot by partner, and report results to partners each quarter.

Partners refer more when they see what happens to their leads.

• Low-quality referrals

  Reward only qualified leads.

• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source
• Press releaseNscale raised $2 billion in a Series C in March 2026 at a $14.6 billion valuation, contracted about 200,000 NVIDIA GB300 GPUs with Microsoft (October 2025), and on September 18, 2026 filed a Form S-1 for a proposed New York Stock Exchange listing under "NSCL". Series C backers include Dell, Lenovo, Nokia and NVIDIA, and the newsroom announces a Singtel partnership to unlock GPU capacity.Source

Publish a clear page: location of each data center, which services run there, legal entity, certifications and how customers choose a region. Use it in vertical and EU campaigns.

Sovereignty is a buying criterion that hyperscalers struggle to answer simply; Nscale can.

• Overpromising

  State facts only.

• Your websiteState of AI Natives 2026: a survey of "500 AI-native businesses in the United States"; "76% saying infrastructure limited their ability to ship features in the past year" and "77% of AI Natives are highly concerned about data residency and compliance". The full report is behind a form.Source
• HiringVertical go-to-market leads: a Head of Financial Services GTM ("author our compliance narrative, build sophisticated ROI models tailored specifically to the unique needs of trading desks", with "SOC2, sovereign data requirements") and a Head of Healthcare & Life Sciences GTM ("drug discovery, genomics, molecular simulation (e.g., AlphaFold), and advanced medical imaging"; "a critical zero-to-one challenge").Source
• Your websiteHomepage (checked October 21, 2026): H1 "The engine of superintelligence"; the section "Trusted by leading AI labs and enterprises to run critical workloads" shows three quotes, from the CEO of investor Aker ASA, the founder of investor G Squared and the UK AI Minister, and no customer; the "Industry solutions" section and footer list only Telco and AI Native. The page promotes "abundant and renewable power resources", "Low PUE" and data centers in Norway, the UK, Iceland, Portugal and the US (Texas, West Virginia, North Carolina).Source

Launch a program with credits, technical support and a path to reserved capacity, and partner with accelerators. Track graduation to paid.

Start-ups that grow on a cloud tend to stay, and the program feeds the AI Natives team.

• Credits burned without conversion

  Tie later credits to milestones.

• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source
• Your websiteState of AI Natives 2026: a survey of "500 AI-native businesses in the United States"; "76% saying infrastructure limited their ability to ship features in the past year" and "77% of AI Natives are highly concerned about data residency and compliance". The full report is behind a form.Source

Publish a monthly note: benchmarks, new GPU availability, energy figures. Send it to a list of reporters and analysts, and post a summary on the blog.

Regular, useful data builds relationships with reporters that announcements alone do not.

• IPO quiet-period rules

  Clear every note with legal while the offering is pending.

• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• Press releaseNscale raised $2 billion in a Series C in March 2026 at a $14.6 billion valuation, contracted about 200,000 NVIDIA GB300 GPUs with Microsoft (October 2025), and on September 18, 2026 filed a Form S-1 for a proposed New York Stock Exchange listing under "NSCL". Series C backers include Dell, Lenovo, Nokia and NVIDIA, and the newsroom announces a Singtel partnership to unlock GPU capacity.Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source

Build an online booking flow for small reservations (GPU type, count, term, region) with a price and a checkout, and route larger requests to sales.

Self-serve captures smaller deals at low cost and frees sales for large contracts.

• Capacity conflicts

  Show only real available capacity.

• Your websitenscale.com has no pricing page: the main navigation offers "Reserve GPUs", "Start Building" and "Contact". The GPU node pages (H100, H200, GB200, GB300 NVL72, Vera Rubin NVL72) list NVIDIA specifications with "Reserve GPUs" and "Get in touch" buttons and no price. The docs say AI services "use a credit-based billing system" ("To access AI services, you'll need to purchase credits"), and per-model prices appear only through the List endpoints API ("pricing as decimal USD").Source
• Competitor researchRivals publish prices on public pages: Lambda's pricing page ("Clear, straightforward pricing for Instances, 1-Click Clusters, and Superclusters. Contact us for reserved capacity at our lowest prices.") lists HGX B200 clusters from $8.87 to $9.86 per GPU-hour; Nebius lists HGX H100 on-demand at $4.50 per GPU-hour; CoreWeave lists an 8-GPU HGX H100 node at $49.24 per hour on demand.See the research Source
• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source

Publish an index of public GPU and token prices across providers, with Nscale's energy and efficiency data, updated quarterly.

Original data earns press and AI citations and positions Nscale on cost.

• Seen as self-serving

  Publish the method and sources.

• Your websiteState of AI Natives 2026: a survey of "500 AI-native businesses in the United States"; "76% saying infrastructure limited their ability to ship features in the past year" and "77% of AI Natives are highly concerned about data residency and compliance". The full report is behind a form.Source
• Competitor researchRivals publish prices on public pages: Lambda's pricing page ("Clear, straightforward pricing for Instances, 1-Click Clusters, and Superclusters. Contact us for reserved capacity at our lowest prices.") lists HGX B200 clusters from $8.87 to $9.86 per GPU-hour; Nebius lists HGX H100 on-demand at $4.50 per GPU-hour; CoreWeave lists an 8-GPU HGX H100 node at $49.24 per hour on demand.See the research Source
• Your websiteSitemap (190 URLs, checked October 21, 2026): 81 blog posts, 46 press releases, 10 event pages, four gated reports (including State of AI Natives 2026 and AI Grid), five GPU node pages, two industry pages (telco, AI native) and a partner lead referral form; the newest blog post is "Why Are AI Costs So Difficult to Predict" (October 21, 2026); no case studies, no pricing page and no comparison pages.Source

Host hands-on workshop days near the US sites (and online), with tours for enterprise buyers and labs for developers.

Physical infrastructure is Nscale's difference; showing it builds trust that a cloud console cannot.

• Site safety and security

  Tour only finished areas.

• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source
• Your websiteHomepage (checked October 21, 2026): H1 "The engine of superintelligence"; the section "Trusted by leading AI labs and enterprises to run critical workloads" shows three quotes, from the CEO of investor Aker ASA, the founder of investor G Squared and the UK AI Minister, and no customer; the "Industry solutions" section and footer list only Telco and AI Native. The page promotes "abundant and renewable power resources", "Low PUE" and data centers in Norway, the UK, Iceland, Portugal and the US (Texas, West Virginia, North Carolina).Source

Publish a live page of power, renewable share and PUE per site, and post monthly highlights on social.

Live data backs the sustainability claims and gives social a steady stream of content.

• Data looks bad some months

  Show yearly averages too.

• Your websiteHomepage (checked October 21, 2026): H1 "The engine of superintelligence"; the section "Trusted by leading AI labs and enterprises to run critical workloads" shows three quotes, from the CEO of investor Aker ASA, the founder of investor G Squared and the UK AI Minister, and no customer; the "Industry solutions" section and footer list only Telco and AI Native. The page promotes "abundant and renewable power resources", "Low PUE" and data centers in Norway, the UK, Iceland, Portugal and the US (Texas, West Virginia, North Carolina).Source
• HiringSenior Social Media Lead (posted August 27, 2026): "Define and execute audience growth strategies that scale communities into the millions", "Design and implement AI-powered content pipelines", and "Own reporting frameworks that connect social to pipeline and brand outcomes" across LinkedIn, X and YouTube.Source

Publish open benchmark scripts for training and inference on Nscale GPUs, with results, and invite the community to submit runs.

Reproducible benchmarks earn trust and links, and bring developers to try the platform.

• Unfavorable results

  Publish anyway; it builds trust.

• Hacker NewsNscale's most-upvoted Hacker News story (66 points) is its benchmark "AMD MI300x GPUs with GEMM tuning improves throughput and latency by up to 7.2x"; Nscale has six Hacker News mentions in total.Source
• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source
• HiringEcosystem Marketing Manager (New York or San Francisco, posted September 8, 2026): builds Nscale's presence "with an initial focus on PyTorch, Hugging Face, Ray, vLLM", turning relationships into "joint technical content, demos, integrations, workshops"; one of four pillars is developer adoption: "Create stronger journeys from the tools and communities developers already use into Nscale."Source

Invite 12 customers to a quarterly board: roadmap preview, capacity plans, feedback. Share results with the product team.

Customers who shape the product stay and refer others.

• Competitive information

  Use NDAs.

• HiringMore go-to-market hiring open now: two Business Development Directors, AI Natives (London; New York or San Francisco), "the founding member of a new team" for "frontier model labs, fast-scaling AI start-ups"; a Director of Media Relations (posted October 13, 2026) and a Director Communications, AI Infrastructure; a North America Events Lead for "flagship trade shows, executive dinners, roundtables, and partner activations"; a Regional Marketing Lead, USA; and a Marketing Operations Specialist who will "create accurate MDF request and reporting processes" with the partnerships team.Source
• HiringDeveloper hiring: a Director, Developer Relations to "lead and build how developers discover, learn, build, and scale" on Nscale and act as "the voice of the developer internally, identifying friction in onboarding, documentation, services, and tooling"; a Developer Relations Engineer; and a Staff Product Engineer, Developer Experience to "define the paved roads developers use to adopt, operate, and scale on Nscale, from first API call to production workloads", including "onboarding & activation".Source