Technology Sector

Gas turbine manufacturing has become the binding constraint on AI data center expansion, with GE Vernova's backlog reaching 116 GW in Q2 2026—nearly six years of production at current throughput of roughly 20 GW annually, with newest orders slated for 2031 delivery. Siemens Energy holds a 69 GW backlog with lead times exceeding three years; Mitsubishi Heavy Industries reported 35 GW of large-frame orders. Combined, the three manufacturers face 220 GW of commitments against global annual production capacity of 60–70 GW, creating a structural shortage that will persist through 2028–2030 even as GE plans to ramp to 30 GW yearly by 2030. Demand is relentless: the International Energy Agency projects data-center electricity consumption will more than double to 945 TWh by 2030 from 415 TWh in 2024, with US capacity alone climbing from 31 GW in 2025 to 66 GW by 2027. Hyperscalers have committed $660–690 billion of capex for 2026 alone, roughly double 2025 levels, yet money cannot compress manufacturing lead times. The constraint cascades: generator step-up transformer lead times exceeded 160 weeks in early 2026; grid interconnection queues stretch to seven years; an estimated 30–50% of large data-center projects scheduled for 2026 have been delayed or canceled, including at least 75 projects worth more than $130 billion halted in Q1 2026. Unable to wait for grid access, hyperscalers have shifted to behind-the-meter generation—private power plants feeding directly into facilities—but those still require the same gas turbines everyone else is fighting for. Turbine prices have surged more than 195% since 2019, and manufacturers now charge reservation fees just to hold a place in the queue. The bottleneck is not temporary scarcity but structural: the speed mismatch between data-center construction (2–3 years) and power infrastructure build-out (5–15 years) means the turbine queue will remain the binding constraint on AI expansion through the end of the decade.

Unitree, a Chinese humanoid robotics maker, surged 629% at its peak on its Shanghai stock market debut Wednesday, raising 6.1 billion yuan ($904 million) at an IPO price of 150.80 yuan per share. The company issued 40.45 million shares, and retail investor demand was extraordinary—the offering was oversubscribed more than 8,000 times. Shares closed 460% higher at 845 yuan, briefly valuing the company at over 400 billion yuan ($59 billion), among China's most valuable listed firms. Unitree shipped 5,500 humanoid robots in 2025 and reported 1.7 billion yuan in revenue that year, with more than 40% from overseas sales. The company will use IPO proceeds for advanced robotics R&D and manufacturing capacity expansion. The debut reflects investor appetite for embodied AI—artificial intelligence embedded in machines that interact with the physical world—and China's dominance in humanoid robot production and supply chains. However, most current demand comes from demonstrations and research rather than commercial deployment. A major headwind looms: the U.S. Federal Communications Commission banned imports of new foreign-made humanoid robots in July on national security grounds, affecting Unitree's future U.S. sales, which represented a portion of 2025 revenue. Unitree's trailing twelve-month P/E ratio stands at nearly 1,200, and analysts caution that the valuation prices in a robot revolution before large-scale industrial adoption has proven viable.

Apple's board authorized $100 billion in share buybacks on April 30, 2026, reloading what has become the largest repurchase machine in corporate history—but the number itself signals a shift, not a crescendo. The authorization is a ceiling, not a commitment: Apple may spend up to $100 billion, and the actual outlay depends on cash flow. What matters is that this $100 billion is a cut from the $110 billion authorized in May 2024, which had been described as the largest buyback ever announced by a US company. The timing is the tell. Tim Cook announced his departure as CEO on April 20, and the $100 billion reload was his last major act before handing the job to John Ternus on September 1. Since Cook took over in 2011, Apple has returned more than $1 trillion to shareholders, of which more than $850 billion went to buybacks. In the March quarter alone, Apple repurchased $61.8 billion in the first nine months of its fiscal year. The buyback works mechanically: fewer shares outstanding means each remaining share claims a larger slice of earnings. In the March quarter, revenue rose 17 percent to $111.2 billion while EPS rose 22 percent to $2.01; roughly five percentage points of that gap came from the share count shrinking by about 4 percent a year. Apple funds this from operating cash flow—$29.6 billion in quarterly profit—and a net cash pile of about $62 billion, with no new debt. The share count has collapsed under Cook: from roughly 26 billion shares when he arrived to 14.6 billion as of July 2026, a 44 percent decline. But the new CEO's priorities may differ. Market skepticism toward buyback announcements is rising, as Apple's valuation now hinges on AI spending and R&D rather than the mechanical lift from share retirement.