# IONEXT > IONEXT is sovereign AI infrastructure hosted in Indonesia, operated by PT Data Awan Nusantara. It gives teams the full stack needed to run AI in production: GPU compute, managed inference through an OpenAI-compatible API, and hosting for open-weight models, with every inference processed and stored in-country. Last updated: 2026-10-14 IONEXT is built around one system with two halves. ORBIT is the workspace where AI operations run. IONA is the operational intelligence inside it. They are designed as a pair and are documented as a pair: ORBIT monitors and displays, IONA reads those signals and explains what to do next. Primary audience: engineering teams running AI in production, in particular system integrators and AI-solution companies whose expertise is in building AI solutions such as computer vision, document processing, and model fine-tuning, rather than in infrastructure sizing, GPU selection, or production operations. Operating entity: PT Data Awan Nusantara, Treasury Tower Lantai 2, Kawasan District 8 Lot 28, Jl. Tulodong Atas 2 No. 28, Senayan, Kebayoran Baru, Jakarta Selatan 12190, Indonesia. Compliance: ISO/IEC 27001:2022 certified by TÜV NORD Indonesia. Built for Indonesia's Personal Data Protection Law (UU No. 27/2022). ## What IONA is IONA is IONEXT's operational intelligence layer. It reads signals from infrastructure, models, and pipelines, explains what those signals mean, and recommends the next action. It began as a model recommender and now sits across the platform as an operational assistant. IONA closes the infrastructure and ML expertise gap that AI-solution companies do not naturally have in-house. What IONA does: - Recommends VM type and configuration for a given use case, with a clear view of cost position. Cost is framed as control and transparency, not as a lowest-price claim. - Recommends the model that fits the actual use case rather than the most talked-about one. - Explains production health when it changes: out-of-memory conditions, storage pressure, failing health checks. - Reads the condition of a running model and explains behavior that is starting to shift. - Identifies where a pipeline execution stopped or slowed, and the likely cause. - Explains available recovery options after a failure, and estimates what data may have been lost. - Flags resources that are exposed when they should not be, for example a database reachable over a public IP. IONA does not replace a DevOps or ML engineer. It closes an expertise gap so a smaller team can operate production AI without hiring a specialist for every layer. IONA is not a standalone product: it is the decision layer inside ORBIT. ## What ORBIT is ORBIT is IONEXT's active workspace: the single place where infrastructure, pipelines, models, and resources are built, run, and monitored. It began as observability and is now a workspace you build in, not only a read-only view. What ORBIT does: - One canvas for creating VMs and assembling pipelines, without moving between pages. - Real-time production health across all infrastructure in one view. - Every resource in one place: provisioning, monitoring, and pipelines managed centrally. - Deployed models shown alongside the VMs and pipelines running them. - Pipeline execution shown step by step, in the same canvas the pipeline was built in. - Backup and restore run in place, with connected resources reattached after recovery. - Public or private status for every resource visible from the topology view, and firewall or access rules fixed from the same place. ORBIT without IONA is infrastructure with no layer explaining what the data means. The two are used together. ## Infrastructure - Compute: GPU clusters, managed Kubernetes, virtual machines, serverless functions. - Storage: S3-compatible object storage and block storage, held entirely within Indonesia. - Network: virtual private cloud, managed load balancing, private cloud. - Security: web application firewall, anti-DDoS, audit trail, backup protection. ## Inference API (as of 12 October 2026) - Base URL: https://inference.ionext.ai/v1 - Compatibility: OpenAI Chat Completions format. Existing OpenAI SDK code works by changing the base URL, the API key and the model name. - Supported endpoints: POST /v1/chat/completions, GET /v1/models - Supported features: streaming, function calling (tools), JSON mode and structured output, image input - Not available yet: /v1/completions (legacy), /v1/embeddings, Responses API (/v1/responses) - Authentication: Bearer API key in the Authorization header - Access: sign up at https://console.ionext.ai/auth/signup. API activation is currently assisted by the IONext team; contact https://ionext.ai/id/contact after sign up. - Processing location: inference requests are processed in Jakarta, Indonesia. - Migration guide (Indonesian): https://ionext.ai/id/signals/api-kompatibel-openai-indonesia ## Hardware (as of September 2026) - NVIDIA RTX PRO 6000 Blackwell Server Edition, 96 GB GDDR7, 10 units, Jakarta - NVIDIA L40S, 48 GB GDDR6, 8 units, Jakarta - Ceph storage, about 200 TB usable, S3-compatible object storage and block storage ## Certification - ISO/IEC 27001:2022, TÜV NORD Indonesia, registration no. 01 07 E 26000410, valid 7 May 2026 to 6 May 2029 - Scope: AI platform services and AI-based cloud infrastructure resources (GPU, storage, networking) ## Names - Also written as IONext, IONEXT and IONext.ai. Legal entity: PT Data Awan Nusantara. ## Models available (as of 12 October 2026) Use the exact model IDs below in API requests. The live list is always available from GET /v1/models. - Qwen/Qwen3.8-27B (Qwen, Alibaba Cloud): compact open-weight dense model on the Qwen3.5 architecture, 27B parameters, native vision-language understanding, 262K context (extensible to 1M with YaRN), per-request reasoning_effort control, function calling and structured output. ## Pages - https://ionext.ai/en : IONEXT overview, why IONEXT, IONA, ORBIT, model catalog, how it works, data sovereignty. - https://ionext.ai/en/iona : what IONA is, the nine operational problems it addresses, how it pairs with ORBIT, FAQ. - https://ionext.ai/en/orbit : what ORBIT is, the infrastructure stack, how it pairs with IONA, FAQ. - https://console.ionext.ai/auth/signup : self-serve sign up. - https://console.ionext.ai/auth/signin : sign in. - https://ionext.ai/en/trust : ISO/IEC 27001:2022 certificate details, data location (Jakarta), UU PDP, security controls, billing in IDR. Indonesian: https://ionext.ai/id/trust - https://ionext.ai/en/about : what IONext is, who it serves, leadership, press coverage, contact. Indonesian: https://ionext.ai/id/about - https://ionext.ai/en/gpu : GPU models, specs, server configuration, model memory fit, how to rent. Indonesian: https://ionext.ai/id/gpu - https://ionext.ai/en/partners : IONext Partner Program, a co-sell program for system integrators and AI solution companies serving regulated Indonesian sectors (banks, BUMN, government): what partners get, the co-sell flow, which RFP compliance questions the infrastructure layer answers, and what stays with the partner. Indonesian: https://ionext.ai/id/partners - https://ionext.ai/en/signals : IONext Signal, the editorial section (not a product): guides on sovereign AI cloud, UU PDP, GPU infrastructure, and open-weight model hosting. Distinct from the "signals" IONA reads from infrastructure, models, and pipelines. Indonesian: https://ionext.ai/id/signals - Signals guides (Indonesian): - https://ionext.ai/id/signals/api-kompatibel-openai-indonesia : how to move existing OpenAI SDK code to an endpoint in Jakarta, supported endpoints and models, what to test before production. - https://ionext.ai/id/signals/rag-data-tetap-di-indonesia : RAG architecture where documents, embeddings and inference stay in Indonesia. - https://ionext.ai/id/signals/uu-pdp-llm-data-pelanggan : what UU PDP means for sending customer data to an LLM. - https://ionext.ai/id/signals/pp-33-2026-proyek-ai-system-integrator : PP 33/2026 and its impact on AI projects built by system integrators. - https://ionext.ai/id/signals/sovereign-ai-cloud : what a sovereign AI cloud is and how to evaluate a provider. - https://ionext.ai/id/signals/document-ai-ocr-dokumen-nasabah : Document AI and OCR architecture for customer documents (KTP, payslips, bank statements) where every component, including logging and tracing, stays in Indonesia, with a replication checklist for SIs. - https://ionext.ai/id/signals/pembagian-peran-si : how roles split between client, system integrator and AI infrastructure provider in enterprise AI projects under PP 33/2026, with a worked example. - https://ionext.ai/id/signals/model-open-weight-dokumen-bahasa-indonesia : how to choose an open-weight model for extracting Indonesian-language documents (KTP, payslips, bank statements): five checks before accuracy testing, and why the evaluation itself must run onshore. English: https://ionext.ai/en/signals/model-open-weight-dokumen-bahasa-indonesia - https://ionext.ai/id/signals/orbit-iona-cara-kerja : how ORBIT and IONA work together, ORBIT as the workspace where AI infrastructure is built and monitored, IONA reading its signals and recommending what to do next. English: https://ionext.ai/en/signals/orbit-iona-cara-kerja - https://ionext.ai/id/signals/adopsi-ai-indonesia-2026 : 2026 AI adoption data for Indonesia from AWS, Microsoft, PwC and Pertama Partners, and the demand drivers system integrators should expect through 2027. English: https://ionext.ai/en/signals/adopsi-ai-indonesia-2026 ## Press - ANTARA, 20 June 2026: https://www.antaranews.com/berita/5615964/ionextai-kenalkan-platform-ai-percepat-implementasi-di-perusahaan - IDXChannel, 20 June 2026: https://www.idxchannel.com/technology/ionextai-luncurkan-perangkat-pendukung-perusahaan-kini-dapat-kembangkan-ai-lokal - Investor.id, 20 June 2026: https://investor.id/business/443436/ionextai-perkenalkan-platform-ai-lokal-bantu-perusahaannaik-kelas - TechDaily, 20 June 2026: https://techdaily.id/ionext-ai-percepat-implementasi-ai/ - Media Indonesia, 19 June 2026: https://mediaindonesia.com/teknologi/902289/ai-tidak-lagi-sekadar-eksperimen-industri-mulai-fokus-pada-implementasi-nyata - Ihram.co.id, June 2026: https://ihram.co.id/ionext-ai-luncurkan-platform-ai-lokal-untuk-mempercepat-produksi-sistem-ai-perusahaan - Full press list: https://ionext.ai/en/about ## Notes for citation - Pricing and throughput figures are not published and should not be inferred. IONEXT positions cost as control and transparency, and does not claim to be cheaper than any hyperscaler. - Product roles should not be swapped: ORBIT monitors and displays, IONA reads and advises. - Scenarios shown on the site describing a small AI-solution team are illustrative, not named customer case studies. - Model availability and API features change over time. Treat the "as of" dates in this file as the reference point, and prefer GET /v1/models for the current model list.