From: allin
Business Winners for 2025
Autonomous Hardware and Robotics
According to Freeberg, 2025 is poised to be the year of autonomous hardware and robotics, following 2024’s focus on compute and AI software systems [01:18:01]. He highlights companies like Unitree, whose Go2 robot, costing $1,600, offers capabilities such as an API, lidar, and intelligent guidance systems, making it suitable for scientific applications or even perimeter security [01:18:20], [01:18:50]. This low cost enables significant applications [01:18:50]. Gavin Baker profoundly agrees, adding that Full Self-Driving (FSD) will reach mainstream adoption [01:21:23], [01:21:28].
Big Businesses Leveraging AI
Gavin Baker predicts that large corporations that thoughtfully integrate AI will be major winners [01:22:05]. This is because advanced AI models, such as OpenAI’s 03, offer a combination of reasoning and test-time compute, providing a significant advantage to large businesses that can afford the substantial investment (e.g., $1 million for an AI to “think” for six weeks) [01:22:13], [01:22:27]. He also anticipates a shortage of GPUs, accelerators, and compute power in 2025, similar to 2023 [01:22:46].
Dollar-Denominated Stablecoins
Chamath Palihapitiya identifies dollar-denominated stablecoins as a significant business winner for 2025 [01:13:16]. He notes a critical development in 2024 where stablecoins decoupled from crypto volatility, becoming useful for various business functions [01:23:33]. By the end of Q2 2024, stablecoin transaction volume (8.5 trillion USD across 1.1 billion transactions) was more than double that of Visa over the same period [01:24:08], [01:24:18]. He believes this indicates a point of no return, suggesting stablecoin usage could quadruple or quintuple by the end of 2025, attacking the Visa and Mastercard duopoly and potentially saving the global economy trillions of dollars in transaction costs [01:24:51], [01:25:28], [01:26:13].
Tesla and Google
Jason Calacanis predicts Tesla and Google, as key players heavily investing in AI, will be the biggest business winners [01:29:05], [01:29:07]. He is particularly impressed by Google’s AI comeback, especially with Gemini Deep Research, which he describes as a “step function higher” than competitors [01:29:56], [01:30:18]. Gemini Deep Research can conduct real-time web searches, summarize information with citations, and produce reports that would typically cost hundreds of thousands of dollars and weeks of work from consulting firms [01:30:54], [01:31:01], [01:31:14]. Tesla’s efforts with xAI, specifically building out 100,000 GPUs in under 45 days, also contributed to this prediction [01:29:29].
Companies Manufacturing High Bandwidth Memory
Gavin Baker foresees that companies producing high bandwidth memory (HBM) will be the best performing asset in 2025 [01:14:10]. He notes that HBM accounts for a larger portion of the cost of goods sold (COGS) for GPUs than Taiwan Semiconductor Manufacturing Company (TSMC) [01:14:24]. With only two main manufacturers, Hynix (HX) and Micron, and HBM being sold out for the past two years, its importance is growing with the rise of test-time compute and inference [01:14:28], [01:14:55], [01:14:57].
Chinese Tech Stocks
Freeberg predicts Chinese tech stocks or ETFs will be the best performing assets in 2025 [01:17:42], [01:17:46]. This is a contrarian view, as most have divested from Chinese tech [01:17:49]. He attributes this to three main drivers:
- US-China Deal: The Trump administration may seek a “great deal” to open Chinese markets to American companies and vice versa, as indicated by recent actions like halting the TikTok ban [01:18:02], [01:18:10].
- Electricity Production: China’s unfathomable cost of electricity production and massive buildout of hydroelectric (e.g., $137 billion dam adding hundreds of gigawatts) and nuclear power will reduce costs and increase available electricity [01:18:37], [01:18:41].
- Free Market Enablement: The Chinese Communist Party has the ability to throttle free markets, and he believes they will enable more innovation and free market activity to motivate a deal [01:19:01].
Mag 7 (Magnificent Seven) Stocks
Jason Calacanis believes the “Mag 7” (Magnificent Seven) tech stocks will be the best performing asset [01:22:11]. He argues that these companies have learned to expand earnings by not hiring more people and instead outsourcing jobs and automating processes [01:22:21]. The internal application of AI within these companies will lead to unprecedented earnings growth that people will struggle to comprehend [01:22:37], [01:22:39].
Business Losers for 2025
Government Service Providers
Gavin Baker predicts that government service providers, particularly those with over 35% of their revenue from the U.S. government, will be the biggest business losers in 2025 [01:39:29]. This is due to a more restrained approach to spending under the new administration, which may scrutinize bills and eliminate waste, fraud, and abuse [01:39:38], [01:40:07].
Mag 7 Drawdown
Chamath Palihapitiya warns that the “Mag 7” tech stocks could experience a multi-trillion dollar absolute drawdown in 2025 [01:40:10], [01:40:14]. While acknowledging the exceptional quality of these businesses, he is concerned about the unprecedented concentration of these companies in market indices, approaching 40% [01:40:30], [01:40:39]. Historically, such high concentrations have foreshadowed significant drawdowns [01:40:43], [01:40:45]. Even a 10% pullback would equate to trillions of dollars [01:41:10], [01:41:13].
Old Defense and Aerospace Providers
Freeberg predicts that traditional defense and aerospace companies, such as Boeing, Lockheed Martin, and Raytheon, will be major losers [01:41:29], [01:41:31]. He believes U.S. defense budgets will shift towards more tech-oriented, AI-driven, and robot-driven solutions, benefiting companies like Palantir and Anduril [01:41:40], [01:41:46]. He also points to a “failure at scale” and increased bureaucracy within these older businesses, citing Boeing’s space program failures and airplane business challenges in 2024 [01:41:51], [01:41:55]. Cost-plus contracting, which incentivizes inefficiency, will be heavily scrutinized [01:43:00], [01:43:08].
OpenAI
Jason Calacanis predicts that OpenAI will experience its peak valuation in 2025 and ultimately be the biggest business loser, potentially seeing a “total collapse” [01:44:34], [01:44:38], [01:44:47]. He argues that the company’s headwinds are “absolutely underappreciated” [01:44:44]. Google’s strong AI advancements and xAI’s significant infrastructure build-out pose major competitive threats [01:44:47], [01:44:50]. Microsoft’s access to OpenAI’s source code and ownership of “big iron” (compute infrastructure) suggest they no longer need OpenAI [01:45:01]. He doubts OpenAI can maintain its current pricing and suggests a non-zero chance of legal challenges over its transition from non-profit to for-profit, potentially overturning a $157 billion valuation transfer [01:45:12], [01:45:31], [01:45:37]. He also points out that developers show no loyalty to any single AI service and often split queries across multiple platforms, gravitating towards open-source alternatives [01:46:40], [01:46:42].
Enterprise Application Software (“Software Industrial Complex”)
Gavin Baker predicts pain for Enterprise Application Software companies in 2025, especially with the rise of “agents” – AI models that can perform any online human action [01:23:41], [01:23:45]. He believes that dominant cloud providers and labs will control these agents, rendering existing enterprise software companies (who don’t own their own models or compute) less competitive [01:23:55], [01:24:13].
Chamath Palihapitiya doubles down on this, coining the term “software industrial complex” to describe large, bloated enterprise software companies that have perfected sales motions and heuristics around basic database functions rather than delivering true product value [01:25:03], [01:25:07], [01:25:22]. He expects CEOs and CFOs to pressure CIOs to manage spending, as next-generation AI businesses can offer similar features at an order of magnitude lower cost due to increased productivity from AI tools [01:25:56], [01:26:10]. He notes that while traditional software makes white-collar workers more efficient, AI aims to replace them entirely, representing a fundamental paradigm shift [01:27:49], [01:27:54].
Freeberg reiterates this by tripling down on his prediction for vertical SaaS (Software as a Service) [01:28:17]. He foresees pricing models being challenged and compressed as companies develop in-house AI tools to replace traditional business practices [01:28:19], [01:28:21].
Legacy Car Companies and Real Estate
Jason Calacanis predicts that legacy car companies and real estate will be the two worst-performing asset classes in 2025 [01:29:41], [01:29:44]. He notes an oversupply of cars and a consumer base burdened by record debt [01:29:49], [01:29:51]. In real estate, he points to declining housing values and rents in states that allow building (like Texas) and people leaving states with restrictive building policies [01:30:04], [01:30:06]. This aligns with Chamath’s earlier prediction of a wave of consolidation and “melting icebergs” in the traditional auto OEM industry [01:47:10].