The godfather of AI warns: a new wave of layoffs could arrive in 2026 The evolution of artificial intelligence (AI) is following a pattern reminiscent of the phases of a financial bubble, according to Hyman Minsky's framework: displacement, boom, euphoria, profit-taking, and panic. Although Minsky did not consider AI, his analysis helps to understand the current moment. The question is no longer whether AI will transform the economy, but whether massive investment in the sector will generate the expected returns. Thus, 2026 is shaping up to be a key year, not necessarily of collapse, but of adjustment to reality. Experts cited by Business Insider Spain point to three indicators that will define the financial future of AI in the next 12 to 18 months: 1. Real profitability: If in 2026 the major players in AI fail to increase their margins beyond revenue growth, it will signal that competition has turned AI into a commodity too quickly. 2. Sustainable financing: As long as investment in infrastructure is covered by cash flows, the system will be stable. If growth depends on debt, the sector will enter a vulnerable phase. 3. Tangible productivity: Companies must demonstrate measurable improvements in efficiency and costs. Without concrete results, the narrative around AI could weaken. The comparison with the tech bubble of 2000 is inevitable. However, the current situation resembles 1998 more, when the market still had room to grow before the burst. Examples like Global Crossing and Worldcom show how enthusiasm can precede abrupt declines, although this time the fundamentals seem more solid. The AI sector shows speculative signs: investments from major providers already account for nearly 30% of their projected revenues, and semiconductor stocks have risen significantly. Commitments from OpenAI and other players, totaling hundreds of billions of dollars, reinforce the perception of an intense investment environment. However, most companies are already discounting the risk that these commitments will not reach the announced $1.4 trillion. Despite speculation, the financial fundamentals of major providers are robust, with exceptions like Oracle. The investment cycle in AI, representing about 1.4% of U.S. GDP, is modest compared to other major historical cycles. Therefore, investors are advised to focus on companies with solid balance sheets and diversify into related sectors, such as energy or infrastructure. Euphoria for AI coexists with setbacks: competition from cheaper Chinese models and studies questioning the impact on productivity have raised doubts about the sector's sustainability. The main risk is that competition erodes margins and pricing power, both in chip manufacturing and model development. Greater dispersion in the sector is expected, moving away from the "winner takes all" model. However, the situation is different from the dot-com bubble: current profits are backed by solid earnings, capital spending is more contained, and leverage is lower. While there may be a mania, it is not yet time to panic. Looking ahead to 2026, greater clarity is expected regarding the direction of AI. The current model, based on intensive computing, could consolidate or give way to alternatives if the anticipated leap in capabilities does not occur. The potential for a new productivity cycle exists, but it is not guaranteed. The investment cycle in AI is still in its early stages and has spread throughout the economy, not just the tech sector. The next phase will depend on "AI scalers," companies with the resources to meet investment commitments of up to $2.1 trillion by 2027. However, these investments will require new forms of financing, such as leasing, credit, and capital issuance. While an imminent bubble is not perceived, investors must monitor whether investments in AI generate profitability and whether financial conditions remain favorable. Unlike the telecom bubble of the 90s, most investments in AI are financed with free cash flow, although this is starting to change. Additionally, the valuations of major tech companies today are more moderate than those during the dot-com bubble, and monetary policy is more lenient. In summary, while there are risks and signs of speculation, the AI sector has not yet reached the critical point of a bubble, but 2026 will be a decisive year to see if the promise of AI translates into tangible results. (Source: businessinsider.es)