{"slug": "ai-is-making-chips-more-expensive", "title": "AI Is Making Chips More Expensive", "summary": "SMIC, China's largest chip foundry, reported record quarterly revenue exceeding $3 billion, driven by strong AI-related demand, and raised prices for its most sought-after manufacturing capacity. The company's average selling price rose 5.7% while wafer shipments increased 14%, reflecting tight supply in the semiconductor supply chain as AI growth drives demand for physical computing infrastructure.", "body_md": "When we talk about the AI boom, we usually talk about:\n\nChatGPT\n\nClaude\n\nGemini\n\nAI agents\n\nGPUs\n\nLLMs\n\nBut underneath all of those technologies is something much more fundamental:\n\nSemiconductors.\n\nAnd a recent report about China's largest chip foundry, SMIC, shows just how strongly AI demand is reaching into the semiconductor supply chain.\n\nSMIC says AI-related demand remains strong enough that it has raised prices for some of its most sought-after manufacturing capacity.\n\nThat's a big signal.\n\nBecause AI growth isn't happening only at the software layer.\n\nIt's creating enormous demand for physical computing infrastructure.\n\n🏭 First: What Does SMIC Actually Do?\n\nSMIC stands for Semiconductor Manufacturing International Corporation.\n\nThink of a chip company as having several layers.\n\nA simplified version looks like this:\n\nChip Design\n\n↓\n\nChip Architecture\n\n↓\n\nManufacturing\n\n↓\n\nPackaging\n\n↓\n\nTesting\n\n↓\n\nComputer / Server\n\nCompanies can design chips without actually manufacturing them.\n\nA foundry manufactures chips designed by other companies.\n\nSMIC is primarily a semiconductor foundry.\n\nIts customers provide chip designs, and SMIC manufactures those designs on semiconductor wafers.\n\n🧠 Why Does AI Need So Many Chips?\n\nModern AI models require enormous amounts of computation.\n\nTraining a large model might involve:\n\nMassive Dataset\n\n↓\n\nNeural Network\n\n↓\n\nMillions/Billions of Parameters\n\n↓\n\nRepeated Computation\n\n↓\n\nThousands of Accelerators\n\n↓\n\nHuge Energy Consumption\n\nAnd training isn't the only cost.\n\nOnce a model is deployed, millions of users may interact with it.\n\nEvery request requires inference.\n\nFor example:\n\nUser\n\n↓\n\nAI Application\n\n↓\n\nAPI\n\n↓\n\nModel Server\n\n↓\n\nGPU / AI Accelerator\n\n↓\n\nInference\n\n↓\n\nResponse\n\nMultiply that by millions of users and billions of requests.\n\nThe hardware requirements become enormous.\n\n📈 SMIC's Numbers Tell an Interesting Story\n\nAccording to the Reuters report, SMIC generated more than:\n\n$3 billion in quarterly revenue\n\nfor the first time.\n\nIts second-quarter performance was helped by strong AI-related demand.\n\nSome other important numbers:\n\nMetric Result\n\nQ2 Revenue >$3B\n\nWafer shipments +14% QoQ\n\nAverage wafer selling price +5.7%\n\nMonthly capacity 1.1M 8-inch-equivalent wafers\n\nCapacity utilization 93.7%\n\nQ2 shareholder profit $479.2M\n\nH1 capital spending $3.4B\n\nExpected 2026 amortization ~$5B\n\nThese numbers show something important:\n\nDemand is strong, but semiconductor manufacturing is extremely capital-intensive.\n\n💰 Why Are Chip Prices Going Up?\n\nThe simplest explanation is:\n\nHigh AI Demand\n\n↓\n\nMore Chip Orders\n\n↓\n\nLimited Manufacturing Capacity\n\n↓\n\nHigher Utilization\n\n↓\n\nTighter Supply\n\n↓\n\nHigher Prices\n\nSMIC's average selling price increased by 5.7% while wafer shipments increased by 14%.\n\nAnd SMIC said it raised prices for its most sought-after capacity after negotiating with customers.\n\nThat's basic supply and demand.\n\nBut in semiconductors, increasing supply isn't as simple as:\n\n“Let's build another factory.”\n\nA semiconductor fabrication plant can require billions of dollars and years of planning, construction, equipment installation and qualification.\n\n🔬 What Does “7nm” Mean?\n\nThis is another important concept for anyone interested in AI hardware.\n\nSMIC is reported to be the only Chinese foundry currently capable of mass-producing logic chips such as CPUs and GPUs using a 7-nanometre process.\n\nBut what does 7nm mean?\n\nVery roughly, it's a label associated with a particular semiconductor manufacturing technology generation.\n\nSmaller process technologies generally aim to provide:\n\nMore transistors\n\nBetter performance\n\nImproved power efficiency\n\nGreater computational density\n\nA simplified evolution might look like:\n\nOlder Process\n\n↓\n\nLarger Transistors\n\n↓\n\nFewer Transistors / Area\n\n↓\n\nNewer Process\n\n↓\n\nSmaller Features\n\n↓\n\nMore Computational Density\n\nModern AI accelerators depend heavily on advanced semiconductor processes.\n\n🤖 AI Is More Than GPUs\n\nOne detail from the SMIC report is particularly interesting.\n\nSMIC said much of the increase in shipments was driven by AI-related demand for chips other than CPUs and GPUs, particularly from Chinese customers.\n\nThat's important.\n\nWhen people hear “AI hardware,” they usually think:\n\nGPU\n\nBut an AI infrastructure stack can contain many different kinds of chips.\n\nFor example:\n\n```\n         AI Data Center\n              │\n   ┌──────────┼──────────┐\n   ↓          ↓          ↓\n CPU        GPU       Networking\n   │          │          │\n   └──────────┼──────────┘\n              ↓\n          Storage\n              ↓\n         Controllers\n              ↓\n        Power Systems\n```\n\nAI demand can therefore affect an entire semiconductor ecosystem.\n\n🌐 The AI Supply Chain Is Huge\n\nA modern AI system isn't simply:\n\nGPU + Model\n\nIt's more like:\n\nSemiconductor Manufacturing\n\n↓\n\nAI Chips\n\n↓\n\nServers\n\n↓\n\nNetworking\n\n↓\n\nData Centers\n\n↓\n\nCloud Platforms\n\n↓\n\nAI Models\n\n↓\n\nDeveloper APIs\n\n↓\n\nAI Applications\n\n↓\n\nUsers\n\nA bottleneck at almost any layer can affect the layers above it.\n\nThis is why semiconductor manufacturing is becoming strategically important to the AI industry.\n\n⚡ Why 93.7% Utilization Matters\n\nSMIC reported capacity utilization of approximately:\n\n93.7%\n\nCapacity utilization tells us how much of the available manufacturing capacity is actually being used.\n\nImagine a factory can theoretically produce:\n\n100 units\n\nbut is currently producing:\n\n50 units\n\nIts utilization is:\n\n50 / 100 = 50%\n\nIf it produces:\n\n94 / 100\n\nutilization becomes:\n\n94%\n\nA utilization rate around 94% indicates the factory is operating at a very high level.\n\nThat's good for revenue.\n\nBut it also means there isn't an enormous amount of unused capacity available if demand suddenly increases.\n\n🏗️ The Hidden Cost of AI: CapEx\n\nThere's another number developers should pay attention to:\n\nCapital expenditure\n\nSMIC spent approximately $3.4 billion during the first half of the year.\n\nWhy so much?\n\nBecause semiconductor manufacturing requires expensive infrastructure.\n\nThink:\n\nFab\n\n↓\n\nLithography\n\n↓\n\nDeposition\n\n↓\n\nEtching\n\n↓\n\nIon implantation\n\n↓\n\nMetrology\n\n↓\n\nPackaging\n\n↓\n\nTesting\n\nThe equipment involved can be extremely expensive.\n\nThis is why semiconductor companies aren't like typical software startups.\n\nA software company might scale by adding:\n\nServers + Engineers\n\nA semiconductor manufacturer needs:\n\nFactories\n\n+\n\nSpecialized equipment\n\n+\n\nClean rooms\n\n+\n\nEngineers\n\n+\n\nMaterials\n\n+\n\nEnergy\n\n+\n\nYears of investment\n\n💡 The Interesting Business Lesson\n\nThis creates a fascinating economic relationship.\n\nAI companies want:\n\nMore compute.\n\nChip designers want:\n\nMore manufacturing capacity.\n\nFoundries want:\n\nMore equipment and factories.\n\nCloud providers want:\n\nMore data centers.\n\nAnd everyone is trying to scale simultaneously.\n\nSo the AI boom creates a chain reaction:\n\nAI adoption\n\n↓\n\nMore model usage\n\n↓\n\nMore compute\n\n↓\n\nMore chips\n\n↓\n\nMore semiconductor capacity\n\n↓\n\nMore factories\n\n↓\n\nMore capital investment\n\n🇨🇳 Why SMIC Matters Beyond One Company\n\nThere is also a broader strategic dimension.\n\nSMIC is China's largest chip foundry.\n\nThe Reuters report says approximately 90% of SMIC's second-quarter revenue came from China, while the United States contributed around 8%.\n\nThat tells us how strongly its business is tied to the domestic Chinese technology ecosystem.\n\nAnd as AI becomes increasingly important to:\n\nCloud computing\n\nRobotics\n\nAutonomous systems\n\nDefense\n\nSoftware\n\nData centers\n\nConsumer electronics\n\nsemiconductor manufacturing becomes a strategic capability.\n\n🔥 AI Infrastructure Is Becoming a Competitive Advantage\n\nFor years, software developers could think mostly about:\n\nCode\n\n+\n\nServers\n\nAI changes the equation.\n\nNow the competitive stack increasingly looks like:\n\nAlgorithms\n\n↓\n\nModels\n\n↓\n\nInference\n\n↓\n\nAccelerators\n\n↓\n\nSemiconductors\n\n↓\n\nManufacturing\n\n↓\n\nEnergy\n\nA company can have an incredible AI model.\n\nBut if it doesn't have enough compute to serve that model efficiently, scaling becomes difficult.\n\nThis is why hardware availability can directly influence software and AI companies.\n\n👨💻 What Does This Mean for Developers?\n\nYou might be wondering:\n\n“I'm learning Python and AI engineering. Why should I care about semiconductor manufacturing?”\n\nBecause understanding the infrastructure beneath AI makes you a better engineer.\n\nYou don't necessarily need to become a chip designer.\n\nBut you should understand concepts like:\n\nAI Compute\n\nCPU\n\nGPU\n\nTPU\n\nNPU\n\nAI accelerators\n\nSystems\n\nMemory\n\nNetworking\n\nStorage\n\nDistributed computing\n\nAI Infrastructure\n\nModel serving\n\nBatching\n\nQuantization\n\nParallelism\n\nInference optimization\n\nGPU utilization\n\nCloud\n\nContainers\n\nKubernetes\n\nData centers\n\nLoad balancing\n\nObservability\n\nThese concepts help you understand what actually happens when your Python code calls an AI model.\n\n🧠 The Developer Perspective\n\nImagine you build an AI application.\n\nYour code might look incredibly simple:\n\nresponse = client.generate(\n\nmodel=\"some-large-model\",\n\nprompt=user_prompt\n\n)\n\nBut underneath that one line could be:\n\nYour Python Application\n\n↓\n\nAPI Gateway\n\n↓\n\nLoad Balancer\n\n↓\n\nInference Server\n\n↓\n\nModel Runtime\n\n↓\n\nGPU Cluster\n\n↓\n\nNetworking\n\n↓\n\nData Center\n\n↓\n\nSemiconductor Hardware\n\nThat's the hidden infrastructure behind modern AI.\n\nAnd understanding that stack is becoming increasingly valuable.\n\n🚀 The Bigger Picture\n\nThe SMIC story shows that the AI revolution isn't happening only inside research labs.\n\nIt's happening inside:\n\nFactories.\n\nIt's happening inside:\n\nData centers.\n\nIt's happening inside:\n\nCloud platforms.\n\nIt's happening inside:\n\nChip-design teams.\n\nAnd of course, it's happening inside:\n\nSoftware applications.\n\nThe AI industry is effectively building a new computing infrastructure layer.\n\n🔮 What's Next?\n\nIf AI demand continues growing, we'll likely see increasing pressure across the entire compute ecosystem:\n\nMore AI applications\n\n↓\n\nMore inference\n\n↓\n\nMore accelerators\n\n↓\n\nMore semiconductor demand\n\n↓\n\nMore manufacturing investment\n\n↓\n\nMore capacity\n\n↓\n\nLower bottlenecks\n\nBut achieving that isn't easy.\n\nFactories take years.\n\nEquipment is expensive.\n\nAdvanced manufacturing is technically difficult.\n\nEnergy requirements are significant.\n\nAnd geopolitical constraints can complicate the supply chain even further.\n\n💭 Final Thought\n\nWhen someone says:\n\n“AI is growing rapidly.”\n\ndon't think only about ChatGPT, Claude or Gemini.\n\nThink about everything underneath them.\n\nAI Applications\n\n↓\n\nAI Models\n\n↓\n\nInference\n\n↓\n\nGPU / Accelerators\n\n↓\n\nServers\n\n↓\n\nData Centers\n\n↓\n\nSemiconductors\n\n↓\n\nManufacturing\n\nSMIC's latest results are a reminder that the AI revolution is also a hardware revolution.\n\nThe companies building the models may get most of the attention.\n\nBut underneath every AI model is an enormous physical infrastructure of chips, factories, electricity, cooling systems, networks and engineers.\n\nThe future of AI will be written in software—but it will run on silicon. 🧠⚡\n\nSource: Reuters, August 14, 2026. Financial and operational figures in this article are based on the Reuters report.", "url": "https://wpnews.pro/news/ai-is-making-chips-more-expensive", "canonical_source": "https://dev.to/techytcm/ai-is-making-chips-more-expensive-3ej9", "published_at": "2026-08-15 05:50:52+00:00", "updated_at": "2026-08-15 06:11:33.908632+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-infrastructure", "ai-chips"], "entities": ["SMIC", "Reuters"], "alternates": {"html": "https://wpnews.pro/news/ai-is-making-chips-more-expensive", "markdown": "https://wpnews.pro/news/ai-is-making-chips-more-expensive.md", "text": "https://wpnews.pro/news/ai-is-making-chips-more-expensive.txt", "jsonld": "https://wpnews.pro/news/ai-is-making-chips-more-expensive.jsonld"}}