Investment bank Morgan Stanley has warned that a major breakthrough in artificial intelligence (AI) could arrive in the first half of 2026. This could be driven by a rapid expansion of computing power at the leading AI labs in the United States. In its latest report, the bank says this could strain power grids and disrupt jobs. And according to it, the pace of AI progress is faster than what many policymakers, industries, and infrastructure providers are prepared for.
Survey
`; // trigger animation const inner = qaContainer.querySelector(“.qa-inner”); setTimeout( => inner.classList.add(“show”), 50); document.querySelectorAll(“input[name=’answer’]”).forEach(radio => { radio.addEventListener(“change”, => submitAnswer(radio.value)); });}function showThankYou { thankYouBox.style.display=”block”; setTimeout( => thankYouBox.classList.add(“show”), 50);}function submitAnswer(answerValue) { const finalPayload = { campaign_name: “Digit Questionaire”, form_name: “Digit Questionaire_Form 1”, form_data: [ { key: “uuid”, value: deviceId, type: “text” }, { key: “question”, value: questions[currentQuestionIndex].question, type: “text” }, { key: “response”, value: answerValue, type: “text” } ], verification_data: { captcha_verification: “”, captchaValue: null, captchaId: null, phone_verification: false, email_verification: false }, meta_data: { referer: document.referrer || window.location.href, user_agent: getDeviceType } }; const myHeaders = new Headers ; myHeaders.append(“accept”, “*/*”); myHeaders.append(“origin”, “https://www.timesdrive.in”); myHeaders.append(“user-agent”, navigator.userAgent); const formdata = new FormData ; formdata.append(“finalPayload”, JSON.stringify(finalPayload)); fetch(“https://apivelocitynext.tnn.in/submit-form-data/68b6d8aac3aa7094b919ac4f/68b6d8f5c3aa7094b919ac89”, { method: “POST”, headers: myHeaders, body: formdata }) .then(res => res.text ) .then(result => console.log(“Submitted:”)) .catch(err => console.error(“Error:”)); currentQuestionIndex++; progress++; setCookie(“progress”, progress); if (currentQuestionIndex { qaBox.style.display=”none”; showThankYou ; }, 600); }}if (currentIndex qaBox.classList.add(“show”), 1200); showQuestion(currentIndex);}
Rapid compute growth could speed up AI progress
Morgan Stanley estimates that as companies dramatically increase the amount of computing power in terms of datasets and associated resources, AI capability would continue to improve. This concept is called ‘scaling laws’, and recently, Tesla and xAI chief Elon Musk also argued that applying ten times more compute to training large language models could effectively double their intelligence.
The bank agrees that the scaling law is still holding up, and future models could improve quickly if computing resources keep expanding.
Executives from major AI companies are reportedly signalling to investors that upcoming model releases may deliver performance improvements beyond current expectations. One example is OpenAI’s GPT-5.4 ‘Thinking’ model, which reportedly achieved an 83% score on the GDPVal benchmark, a test designed to evaluate AI performance on economically valuable tasks. Morgan Stanley says such results indicate that some AI systems are beginning to approach or match expert-level performance in certain domains.
Self-improving AI systems may emerge
The Morgan Stanley report also mentions predictions from AI researchers about the possibility of recursive self-improvement. It is a concept where AI systems help design and improve future versions of themselves.
Jimmy Ba, co-founder of Elon Musk’s AI company xAI and a professor at the University of Toronto, has suggested that such systems could begin appearing as early as the first half of 2027 if current development trends continue.
If it pans out, recursive AI improvement could accelerate technological progress significantly because systems would no longer rely entirely on human researchers to upgrade their capabilities.
Adverse effects of AI progress
Energy demand may become the biggest constraint. The report estimates the United States could face a power shortage of between 9 and 18 gigawatts by 2028, equivalent to roughly 12 to 25% of the electricity required to support projected AI data centre demand.
Training and running large AI models requires enormous computing clusters, often powered by thousands of specialised graphics processing units (GPUs). These clusters consume large amounts of electricity, and data centre developers are increasingly seeking alternative energy sources to avoid delays.
According to Morgan Stanley, some companies are repurposing former Bitcoin mining facilities into high-performance computing sites for AI workloads. Others are installing natural gas turbines or fuel cell systems directly at data centre locations to generate electricity locally.
The bank describes an emerging investment model around AI infrastructure that it calls the ’15-15-15′ dynamic. This refers to data centre leases lasting around 15 years, delivering roughly 15% yields, and generating about $15 per watt in estimated net value creation.
Morgan Stanley also warns that the economic impact of advanced AI could extend into the labour market.
The report suggests that increasingly capable AI systems may act as a ‘deflationary force’ in the economy by allowing companies to automate tasks that were previously performed by humans. Businesses adopting AI tools could complete certain types of work faster and at lower cost.
As a result, the bank says some companies are already restructuring teams or laying off people as AI systems improve productivity in areas such as coding, data analysis, and content generation.
OpenAI CEO Sam Altman has previously argued that AI could enable extremely small teams to build highly competitive companies. In some cases, he has suggested that startups with only one to five people may eventually be able to compete with much larger organisations by relying heavily on AI tools.
Governments around the world are increasingly paying attention to the energy demands of AI infrastructure and the potential economic disruption that could happen. Let’s hope policymakers and industry leaders prepare infrastructure and workforce strategies to address this.
Keep reading Digit.in for similar stories around AI development.
