From: allin
The Year of Autonomous Hardware and Robotics
Many experts predict that 2025 will be “the year of autonomous Hardware or robotics” [00:17:45]. Following 2024, which focused on compute build-out and AI system rollout in software, the shift is expected towards physical AI systems [00:17:55]. This means a significant increase in the presence and capabilities of robots and autonomous systems in daily life [00:19:12].
Key Players and Examples
- Unitree: This Chinese company’s Go2 robot, priced at 16,000 [00:20:10] and can be commanded for various tasks in homes or factories [00:20:22].
- Full Self-Driving (FSD): Expected to gain mainstream adoption, with the technology continuing to compound at an accelerating rate [00:21:28]. Some users already prefer Tesla’s FSD for safety during late-night Uber rides [00:21:39].
- Autonomous Drones: Projects like Zipline, which delivers almost anything to Suburban America, indicate that drones will become a significant wild card in delivery services [00:53:40]. Amazon is already deploying drones in Texas for delivering 60,000 SKUs, dropping items in backyards within 45 minutes [00:54:08]. Matuan in China also uses food delivery drones [00:54:22].
- Waymo: Having launched in San Francisco in August 2023, Waymo achieved a 22% market share of rides in SF by November 2024, matching Lyft’s share and reducing Uber’s to 55% [00:54:35]. With launches in Los Angeles, Austin, and existing operations in Phoenix, Waymo’s new hardware platform is expected to significantly reduce capital expenditure for new launches, improving return on invested capital metrics [00:55:01].
AI Advancements and Impact
In 2025, AI is expected to make more progress per quarter than it did per year in 2023 and 2024 [01:34:08]. This acceleration is attributed to new axes of scaling performance, including pre-training, inference-time compute, and reasoning [01:34:36]. AI models can generate synthetic data containing “reasoning traces” by solving verifiable problems and showing their step-by-step thought processes [01:35:28]. This process involves reinforcement learning, leading to multiplicative scaling [01:35:50].
Specifics of AI Impact
- Big Business Advantage: Large businesses that thoughtfully use AI will be winners [00:22:05]. The ability to invest millions for AI to analyze critical business questions for weeks will provide a profound advantage over smaller businesses [00:22:24].
- Compute Shortages: There will be a shortage of GPUs, accelerators, and compute in 2025, similar to 2023 [00:22:46]. High Bandwidth Memory (HBM) will be crucial, as it constitutes a larger part of GPU costs than Taiwan Semiconductor Manufacturing Company (TSMC) [01:14:10]. Only SK Hynix (HX) and Micron (MU) currently produce HBM, making it a key asset for NVIDIA, AMD, and Amazon’s Traniums [01:14:28].
- Agentic Software: The year 2025, particularly the second half, is expected to be the “year of agents” [01:24:41]. These AI models can take action and perform any online task a human can [01:24:45]. Given that labs and big cloud providers will likely dominate agents due to their low-cost production capabilities, Enterprise Application Software companies that do not own their models or compute infrastructure may face significant challenges [01:24:55].
- Impact on Jobs and Inequality: AI is expected to amplify inequality in the short term, as the amount of money spent on AI and test-time compute will give a massive advantage to companies and individuals [01:07:31]. This rapid change will cause significant employment and income disruption in 2025, affecting not just blue-collar jobs but also roles in venture capital, development, design, and writing [01:08:00].
Dollar-Denominated Stablecoins
Dollar-denominated stablecoins are predicted to be a major business winner in 2025 [00:23:13]. In 2024, stablecoins decoupled from crypto volatility and began to be used for “wholesale useful functions” in running businesses [00:23:33]. By the end of Q2 2024, stablecoin usage reached 1.1 billion transactions, totaling $8.5 trillion in volume, more than double Visa’s transaction volume for the same period [00:24:01]. This indicates a “point of no return” has been crossed [00:24:38]. Stablecoins are expected to challenge the Visa and Mastercard duopoly, leading to an “innumerable number of use cases” [00:24:51]. They could quadruple or quintuple by the end of 2025, due to their potential to remove 300 basis points of “drag” from the global economy, saving trillions of dollars [00:25:25].
Consolidation and Competition in Tech
Auto Industry Mergers
The collapse of traditional Auto Original Equipment Manufacturers (OEMs) is anticipated in 2025, leading to a massive wave of consolidation [00:47:10]. The merger agreement between Honda and Nissan (with Mitsubishi’s involvement) at the end of 2024 is seen as a signal for this trend [00:47:20]. Tesla’s strong position with vehicle quality, software, and FSD autonomy will trigger public capital markets to deem traditional OEMs “uninvestable” [00:47:32]. European OEMs like Volkswagen and Stellantis are in particular trouble [00:48:22].
Onshore Manufacturing Buildout
Massive funding deals for hardware-based manufacturing buildout in the United States are expected [00:50:00]. These deals may involve traditional private equity or government support to accelerate onshore manufacturing, particularly for autonomous and robotic systems [00:50:07]. This push is driven by China’s significant ramp-up in drone, robot, and autonomous vehicle production [00:50:40].
M&A Surge
Following four years of limited activity, a “tidal wave of M&A” is expected due to enormous pent-up demand [00:51:03]. This includes potential deals involving Intel [00:51:14] and the acquisition of independent Frontier AI Labs by larger entities [00:51:23]. The ultimate AI winner will be the one with the lowest infrastructure and compute costs, giving an advantage to full-stack companies that own their compute resources [00:51:41].
Autonomy Ecosystem Partnerships
Major partnerships are anticipated between companies in the autonomy, delivery, food delivery, and e-commerce spaces [00:52:34]. Potential scenarios include Tesla acquiring Uber (which would cost about 10% of Tesla’s market cap) [00:52:47], Waymo spinning out and partnering with Uber, or Amazon acquiring DoorDash and Uber [00:52:52]. These “mega deals” could lead to the creation of powerful “super apps” [00:53:06].
”Software Industrial Complex” Decline
Traditional, large, and often “bloated” enterprise software companies are expected to face significant challenges in 2025 [01:25:03]. These companies, characterized by high sales costs (golf trips, steak dinners) rather than product value [01:25:29], will be disrupted by agentic software and AI. Next-generation AI businesses can rebuild workflows more efficiently and operate at an “order of magnitude cheaper” cost structure [01:25:50]. This will lead to CEOs and CFOs exerting pressure on CIOs to manage IT spend, and the traditional “Cost Plus” model (time and materials, human labor arbitrage) will be heavily scrutinized [01:25:56]. This shift will fundamentally replace white-collar human employees with AI [01:27:49].
Nuclear Power Buildout
A significant trend anticipated for 2025 is the announcement of nuclear power plant buildout in the United States [01:33:00]. This will be driven by deregulation and new technologies, with the new government being more accommodating to meet the competitive demand for electricity production against countries like China [01:33:13]. Nuclear power is seen as an “inevitability” because no other renewable source can scale up fast enough to meet future power needs [01:33:36].