🚀 Milestone: ~50% of @Bitdeer_AI’s upcoming 9.5MW A102 facility in #Malaysia# is already contracted (5-yr offtake) ahead of energization! 🇲🇾 Part of our 350MW of AI Cloud capacity target by Q1 2028. 🔥 Purpose-built for @NVIDIA GB300 NVL72, this liquid-cooled site delivers both GPU cloud & data hosting. We execute to power your AI! ⚡️ $BTDR @NVIDIAAP #GB300# #MalaysiaTech#

🥇 '@Bitdeer_AI is one of the lowest-cost providers for DeepSeek V4 Flash, sitting in the most attractive quadrant while remaining competitive on speed according to the latest @ArtificialAnlys benchmark. 🦾Our goal with this pricing is to accelerate the adoption of cutting-edge intelligence. We hold the advantage through full vertical integration: from owned GPUs down to token serving, backed by deep optimization capabilities. ،🚀Sign up, get your $5 voucher and use Model Studio now: @deepseek_ai ⚡ #OpenModels#

📣 @Bitdeer_AI is heading to @WorldAIShow Malaysia! 🇲🇾 With our data center infrastructure built locally in #Malaysia#, we’re empowering startups and enterprises with the cloud resources needed to build and scale. Scaling AI? Let’s talk! From raw GPU compute to open-source models on our Model Studio.🤝 📅 Sept 9–10 📍 @DoubleTree by Hilton KL (Level 10) 🎪 Booth C10 Will you be in KL? Drop by and meet our experts for a quick chat over coffee☕️ Ready to scale? Let’s ship. 🚀 #MalaysiaTech#

DeepSeek V4 Pro now on @Bitdeer_AI Model Studio. ⚡ 💫💫💫 𝗜𝗻𝗽𝘂𝘁: $0.413 / 1M tokens 𝗖𝗮𝗰𝗵𝗲𝗱: 0.0036 / 1M tokens 𝗢𝘂𝘁𝗽𝘂𝘁: $0.8265 / 1M tokens ✈️ Faster decode, zero extra infra: DSpark speculative decoding and no secondary draft model. 🧠 The Agentic Leap is Real: Massive gains vs. Pro Preview: up to 49.9 points increase on DeepSWE, @terminalbench and CyberGym 📉 1M Context & Fraction of the Cost: Hybrid CSA+HCA uses. Production ready for full-codebase agent loops. 🤑 Compute on Your Terms: 3 reasoning_effort dials (Non-Thinking/High/Max). Pay for what you need. The lowest rates are available today on @Bitdeer_AI Model Studio: @deepseek_ai 🔥

🚀 Qwen3.8-2.4T-A95B is Day-0 live on @Bitdeer_AI Model Studio! 🎉🥳@Alibaba_Qwen! Releasing this 2.4T Max-class open-weights monster (95B active/token, 512 experts MoE). ☄️ 📝 256K context 👌 @OpenAI-compatible + Parallel Tool Calls 🎓 reasoning_effort: tune (low/medium/xhigh) to balance depth vs. cost 🧠 preserve_thinking: keeps reasoning state across multi-turn chats 🚀 Run It Now:

At last week’s Building the Agentic AI Infrastructure session at @SgSMUYPHSL, we proved that the most useful question about AI agents is: 𝙒𝙝𝙞𝙘𝙝 𝙢𝙤𝙙𝙚𝙡, on 𝙬𝙝𝙞𝙘𝙝 𝙘𝙤𝙢𝙥𝙪𝙩𝙚, at 𝙬𝙝𝙖𝙩 𝙘𝙤𝙨𝙩 𝙥𝙚𝙧 𝙧𝙪𝙣? 💡 And we showed exactly how to solve this using @Bitdeer_AI Model Studio. 📌Highlights📌 🤖Real-world Video Gen, Finance & PM agent workflows 💰 Complex tasks via Kimi K3 for ~$1 🔥 Multi-step, multi-model agent calls for <$0.10 Open models give us choice, but robust infrastructure turns that choice into business value.  @Bitdeer_AI is committed to providing your agents the environment they need to actually thrive and scale. Huge thanks to those who spent their evening with us🍷 📊 Catch up on the presentation slides here:

🚀 Day-0 on @Bitdeer_AI Model Studio: @nvidia Nemotron 3.5 Lightning is now live. The fastest open model in its class for always-on agents. A customizable open 30B MoE (3B active) built for your high-volume, specialized AI workloads. Deploy now 👉 #Nemotron# #AgenticAI# @NVIDIAAP 🔥

"In a digital AI economy, economic capacity equals energy capacity." — @PGelsinger at @Ai4Conferences last week. @OpenAI's Sachin Katti @sk7037 also added: the hard part now it's building enough physical infrastructure to support it. As the application layer scales, the reality check hits the infrastructure layer✅: power, memory, network, and the economics of turning compute into useful output. Every inefficiency between layers surfaces as cost, latency or reliability.⚠️ At @BitdeerAI we treat power, data centers, GPU compute, and model serving and tokens on Model Studio⭐as one system. Infrastructure efficiency is becoming a product strategy.🔥

Happy 61st, Singapore! 🇸🇬 🦁 ˗ˏˋ ✸ ˎˊ˗ A nation that turned vision into infrastructure, and is now building one of the world's leading AI ecosystems. 🌱 Proud to be headquartered here, powering AI from SG to the world. 🚀 Majulah Singapura, Go Beyond! 🇸🇬🎉 #SG61# #GoBeyond# #BitdeerAI#

Welcome to the era of the AI Factory. The new gold standards? 𝗖𝗼𝘀𝘁 𝗽𝗲𝗿 𝗧𝗼𝗸𝗲𝗻 and 𝗧𝗼𝗸𝗲𝗻𝘀 𝗽𝗲𝗿 𝗪𝗮𝘁𝘁. 💫 Our core thesis at the @SGTechSocial 𝗔𝗜𝗗𝗖 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 2026: 💡 𝗘𝘃𝗲𝗿𝘆 𝗪𝗮𝘁𝘁 𝗠𝘂𝘀𝘁 𝗘𝗮𝗿𝗻 𝗜𝘁𝘀 𝗧𝗼𝗸𝗲𝗻: A watt lost to an idle queue or mid-job restart is stranded capital. Smart 𝙤𝙧𝙘𝙝𝙚𝙨𝙩𝙧𝙖𝙩𝙞𝙤𝙣 and 𝙥𝙧𝙤𝙖𝙘𝙩𝙞𝙫𝙚 𝙥𝙤𝙤𝙡𝙞𝙣𝙜 turn raw power into relentless AI throughput. ⚙️ 𝗦𝗵𝗮𝘁𝘁𝗲𝗿 𝗧𝗵𝗲 𝗦𝗶𝗹𝗼𝘀: Tuning facility PUE and software in isolation kills efficiency. 𝙄𝙣𝙛𝙧𝙖𝙨𝙩𝙧𝙪𝙘𝙩𝙪𝙧𝙚, 𝙥𝙤𝙬𝙚𝙧, 𝙘𝙤𝙤𝙡𝙞𝙣𝙜, and 𝙨𝙚𝙧𝙫𝙞𝙣𝙜 must be co-optimized as a single, unified engine. At @Bitdeer_AI, we own the full stack, engineering true AI workload economics from the grid all the way to the generated token. 🌐💰 Massive thanks to @SGTechSocial for bringing the ecosystem together and sparking such an insightful conversation! 👏 @Dell @nvidia @NVIDIAAP Bridge Data Centers

⚡Efficient 1M Token context. 🛡️100% Request Success Rate on @Bitdeer_AI Model Studio 🔌@OpenAI Compatible. 👩🏻‍💻Built specifically for coding & agentic workflows. Elite performance at a fraction of the cost. Try the @deepseek_ai DeepSeek V4 Flash now😎

Open models = access. Local adaptation = relevance. AI Infrastructure = impact. 💡 As the Open Models AI Codefest wraps up on 31 July, we’re seeing what becomes possible when innovators get powerful open models, software & GPU compute. 🧵👇 Co-hosted by @nvidia & @Bitdeer_AI, teams tackled challenges in: 🏥 Healthcare 📈 Finance 🗣️ Regional-Language 🔬 Science AcceleratedPowered by NVIDIA technologies software & Bitdeer AI infrastructureGPUs, builders are scaling concepts into solutions & driving sovereign AI. Kudos to every team! ✨ More builder stories and behind-the-scenes highlights coming next. 👀 👏👏OneByZero, @AiKatonic, Amity, LynxKite, MISA, Little Fruit, Panic, Sharpmind, A👏👏 #OpenModels# #AIHackathon#

'@JensenHuang just used his first X post to support open models. We agree, open models are how we diffuse AI for all. But to make open models to work, right infrastructure is a must. As more open models enter the market, access to intelligence is becoming broader, but access alone is not the competitive advantage. The real advantage is to know which model to call for each task, what infrastructure is required to run the workflow reliably, and how much each interaction and workflow will cost. That is where agentic infrastructure comes in. On 5 Aug at @SgSMUYPHSL, we’re showing you exactly how to build agents using open models across finance, marketing, and product, and reveal the model selection and token costs behind real workflows. Build with @Bitdeer_AI Model Studio: #OpenModels# #AgenticAI# #AIInfrastructure#

Kimi K3 is live on @Bitdeer_AI Model Studio. The world's first open 3T-class model: 2.8T params, 1M context, native vision. Top 3 on the @ValsAI, #1# on LMArena Frontend Code Arena. Compared to Fable 5, it delivers solid results at a significantly lower token cost. Kudos to @Kimi_Moonshot and the open-source AI community! Experience open frontier intelligence today! API key →

Sovereign AI Codefest is officially coming to SEA - 🌏 fully virtually. Co-hosted by Bitdeer AI and @NVIDIAAP, we’re bringing Sovereign #AI# to the region with a bang! More than a demo, a high-impact hackathon focused on developing and fine-tuning Sovereign AI models and applications tailored to local needs. 🔥Your agent, you fine-tune! If you’ve been sitting on an AI idea but held back by compute costs, this is your moment. Worried about resources? Here’s what you get for 4 weeks: 🔹 Get hands-on mentorship directly from Bitdeer AI and NVIDIA experts. 🔹 Harness @Bitdeer AI GPU cloud powered by NVIDIA HGX H200 🔹 Explore the full-stack ecosystem to apply AI to real-world use cases. 📅3-31 July 2026, unlimited GPU compute, NVIDIA & Bitdeer AI mentors, impactful outcomes. Bring your problem, we’ll help you build the solutions. Apply before 19 June! 👉 #SovereignAI# #AIHackathon# #GenerativeAI# #BitdeerAI# #OpenHackathon#

After 2 inspiring days at #Echelon2026#, we left with more than conversations but stories worth remembering✨: founders turning bold ideas into impactful products, builders transforming complex workflows into intuitive systems, and competitors from #AI# Workflow improving bottlenecks into smarter flows. One thing was also clear: the demand for GPU resources is real, and the need for reliable AI infrastructure is growing fast⚡️. That's where Bitdeer AI comes in. As a full-stack AI cloud provider, we support builders and businesses with GPU compute, access to various AI models, and scalable cloud solutions, helping teams build faster and experiment easier. Shoutout to @e27co, the inspiring AI Workflow Competition teams, and everyone who dropped by our booth with your questions, making this show truly meaningful! Got an idea, an infrastructure question, or just a coffee chat☕️? We're only one message away! #neocloud# #GPUCloud# #AIInfrastructure# #SGTech#

🔔 New Model Available: NVIDIA Nemotron 3 Ultra is now live on Bitdeer AI Model Studio from day 0! The open 550B Mixture-of-Experts model serves frontier reasoning at 5x higher throughput, and a single NVFP4 checkpoint runs across @NVIDIAAI GPUs. Built for long-running autonomous agents across coding, deep research, and enterprise automation, Nemotron 3 Ultra delivers: 🚀 Up to 5x faster inference 📉 Up to 30% lower cost ⚡️ Frontier-level reasoning 🧠Up to 1M-token context Start building agentic AI today on Bitdeer AI Model Studio: Discover more insights in our latest blog: #NVIDIA# #Nemotron3Ultra# #AgenticAI# #AIInference#

Day 2 at Echelon, and the energy is real🔥 The road from an #AI# “fantasy” to a powerful application is a grind. It takes immense brainpower, sleepless coding nights, and, of course, real resources. At Bitdeer AI, we believe AI can be empowered to anyone with big dreams. And that passion is the beating heart behind the Bitdeer AI for AI Workflow Competition. We’ve been incredibly proud to sponsor our cloud services, host workshops, and mentor builders through those tough times. Today, we proudly presented the grand prize! 🏆 A massive congratulations to our Top 5 builders, selected from 147 amazing participants. Huge shoutout to these brilliant minds! We can't wait to see the incredible applications you bring to the world next! #e27# #Echelon2026# #AIWorkflow# #TechStartups# #AICommunity#

A closer look at the NVIDIA GB200 NVL72 deployment in our Malaysia data centers. Bitdeer AI is scaling high-performance GPU infrastructure to power enterprise AI workloads, from large-scale training to low-latency inference and production-ready deployment.

Day 1 of Echelon Singapore 2026 is officially here — and the Bitdeer AI booth is ready 🇸🇬! Come find us at Booth M21 to meet the team and explore how Bitdeer AI Cloud brings scalable GPU compute and model deployment together in one integrated cloud platform, powering the next wave of AI builders. And don’t miss our fireside chat tomorrow: AI for Everyone: How Bitdeer AI is Building Responsible Infrastructure for Communities, Startups, and Universities Plus SMEs – and Why That’s the Winning Strategy 📅 June 4 ⏰ 11:20 AM – 11:50 AM 📍 Disrupt Stage See you at Booth M21 — let’s build what’s next. #EchelonSingapore2026# #AIIfra# #Datacenters# #neocloud#