From: allin
Discussions around technological advancements in 2024 highlighted significant shifts expected in the AI landscape, business models, and various industries.
Business Winners in Technology
Predictions for the biggest business winners in 2024 included a focus on efficiency, cost reduction, and new AI-driven capabilities.
- Bootstrapped Startups Chamath identified bootstrapped and profitable startups as major winners, anticipating that it will become significantly cheaper to copy existing businesses in 2024 [00:21:26]. This is due to models becoming “10 and 100 times better,” the cost of compute becoming “10 and 100 times cheaper,” and the cost of energy becoming “10 times cheaper” [00:21:56]. He suggests that companies can now be created very cheaply to attack existing businesses that have high overheads in people and processes, which AI can replicate for free [00:22:20].
- Anduril’s Roadrunner Product David Sachs predicted Anduril’s Roadrunner product would be a significant business winner [00:22:55]. This drone interceptor is designed for ground-based air defense and offers a more sustainable solution than using expensive missiles against cheap drones [00:23:05]. The Roadrunner system, costing hundreds of thousands of dollars, is reusable [00:23:46] and can either intercept or return to base [00:24:17].
- Training Data Owners Jason picked owners of training data, such as The New York Times, Reddit, X (Twitter), and YouTube, as major winners [00:24:59]. He noted that language models are quickly reaching parity, making the real value lie in the training data itself, suggesting that open-source models may dominate [00:25:10]. He proposed a market-based solution where language models respect copyright owners and pay licensing fees for data usage [00:25:32].
Business Losers in Technology
Predictions for business losers in 2024 highlighted the disruptive power of AI and shifting market dynamics.
- Vertical SaaS Companies Chamath believes vertical SaaS (Software as a Service) companies will face significant challenges [00:32:41]. The availability of tools for code generation, no-code development, and co-piloting will enable engineers to become “20, 50, 100x more productive” in building custom applications for their enterprises [00:32:51]. This allows companies to create homegrown solutions at a very low cost, disrupting the market for expensive vertical SaaS [00:33:18].
- LLM Startups Chamath also predicted that LLM (Large Language Model) startups would be among the worst performing assets, believing they are “massively overvalued” [00:58:10]. He argued that open source is making an “incredible run” at proprietary models, with too many players leading to parity and unsustainable valuations [00:58:14].
Most Contrarian Beliefs (Tech/AI Focused)
- Decline in OpenAI’s Enterprise Value Chamath’s most contrarian belief for 2024 was that the enterprise value of OpenAI would decrease [00:42:09]. He attributed this not to OpenAI specifically, but to the broader AI industry [00:42:15]. Key reasons included:
- Latency: The current latency of AI tools (30-50 seconds per request) makes building production-quality code impossible [00:42:38].
- Cost: The economic cost of a million tokens on current platforms is “untenable” [00:43:01].
- He believes that capitalism will arbitrage these issues, leading to new cloud services with millisecond latency and significantly lower token pricing (10-20 cents per million tokens) [00:43:28]. This will cause open-source models to “proliferate” and put pressure on proprietary models, leading to a reallocation of profits [00:43:58].
- Apple’s Gains in Generative AI Jason’s contrarian belief was that Apple would make “huge gains” in generative AI or AI in general, perhaps by rebooting Siri [00:50:57]. He feels Apple will not remain on the sidelines in the AI race [00:51:08].
Most Anticipated Trends
- AI-Driven Discovery in BioPharma and Chemical Engineering Friedberg expressed excitement for the advancements in “predictive models, AI-driven discovery of novel molecules, materials, and methods of production” [01:08:54]. This includes new drugs and materials emerging from software rather than traditional lab discovery, promising reduced costs and environmental footprints [01:09:13].
- Exponential Pace of AI Advancement Sachs anticipates that the “exponential pace of advancement in AI will continue” in 2024, leading to innovations percolating down to mainstream consumers [01:10:59].
- Efficiency via AI and Outsourcing Jason’s most anticipated trend was efficiency, driven by AI advances and outsourcing [01:12:24]. He noted that high salaries and a desire for remote work among American workers are forcing companies to build AI and software to automate processes and outsource work globally at lower costs [01:12:31]. This “work from home philosophy” makes hiring internationally as easy as hiring locally, but at a third of the price [01:12:00].
Most Anticipated Media (AI Focused)
- AI-Generated News and Personalized Newscasters Friedberg is interested in AI-generated news, where real-time generative video takes news feeds as input and presents them in a user-desired visual format [01:18:49]. This could lead to a “personal newscaster” that allows users to interact, curate their news, and even filter for specific topics [01:19:01]. Chamath echoed this, mentioning AI models that produce “stunningly symmetrically beautiful people” who could deliver news, allowing users to interact with and tailor the content [01:19:42]. He noted that such products are expected to emerge in 2024 [01:20:15].