From: allin
Federal Reserve Decisions and Economic Outlook
The Federal Reserve recently decided to pause interest rate hikes, marking a break after what would have been the eleventh consecutive rate increase [00:08:27]. Despite the pause, the Fed anticipates two more 25 basis point hikes before the end of the year [00:08:40].
Chairman Powell’s statements have led to some perplexity, as he indicated the Fed is data-dependent but also stated that more rate hikes are likely and rates will remain high for a couple of years [00:16:20]. This creates a contradictory outlook: if there’s a hard landing, rates would be cut, but the Fed forecasts 1% growth and year-end inflation at 3.2% [00:17:25].
One perspective suggests that a hard landing for the US economy is unlikely, especially given China’s recent move to stimulate its economy by “ripping in trillions of dollars” [00:20:04]. This stimulation by a critical global artery could lead to sticky rates and persistent inflation for the rest of the decade [00:20:26]. This aligns with the broader macroeconomic discussion on inflation and Federal Reserve policy.
Market Performance and Tech Rally
The market has seen a significant rally this year, particularly in tech stocks [00:13:28]. This rally is seen as a “reversion to the mean” after a devastating 2022 for the tech sector [00:14:07]. In late 2021, multiples were 50-70% above their 10-year average, while at the trough, they were 30-40% below [00:14:22]. Currently, internet and software companies are still trading below their 10-year average multiples [00:14:51].
The shift in market framework from concerns about rising inflation and rates in 2022 to the type of economic landing in 2023 has influenced this rebound [00:15:01]. With confidence that inflation has peaked and rates are nearing their “final destination,” the market has moved [00:15:19]. The Nasdaq has moved 30% to start the year [00:16:12].
However, despite overall market gains, particularly in tech, some individual stocks may have gotten ahead of themselves [00:17:45]. Many AI-related stocks are at or within 10% of year-end price targets [00:17:58]. The seven or eight most valuable tech stocks are now “priced to perfection,” with enterprise value over net income yields less than half of the two-year note, making little sense compared to government bonds [00:20:51]. If these top companies are excluded, the S&P 500 has not been a great asset [00:21:17].
IPO Market and SoftBank’s ARM
There’s significant news in the IPO market with ARM confidentially filing for an IPO, seeking to raise 32 billion in 2016 [00:09:28]. While original valuation targets were 40 billion fell through due to regulatory scrutiny [00:12:22]. One analyst believes ARM’s cash cow business is likely worth only $25 billion [00:11:53].
AI’s Influence on Market Trends
The “AI rally” is seen as a significant factor in the current market run-up, with many companies pushing an AI angle [00:22:00]. While some AI applications are valid (e.g., Microsoft, Google), others seem speculative [00:22:07]. The market is witnessing a “radical realization” that businesses dependent on services replaceable by AI may lack longevity [00:22:57].
AI is immediately driving demand for chips and infrastructure, leading to a “disproportionate bubbling” in valuations for related businesses [00:23:32]. However, the impact on other sectors like legal or investment banking is further out, with a higher discount rate applied to their potential AI benefits [00:23:51].
The cost curve for AI model development and training is rapidly decreasing, possibly faster than Moore’s Law [01:35:11]. This implies that huge capital investments in training models today (e.g., 5-10 million) in 18-24 months [01:35:31]. Therefore, investing in model development could be risky unless there’s a sustainable competitive advantage through new data generation or business model innovation [01:36:17].
Investment Philosophy and Venture Capital
The current market conditions, influenced by the current state of the US economy and Federal Reserve policies, highlight shifts in investment strategies. Investors are becoming very “data dependent” [01:19:22].
There’s concern about large “heat check” seed rounds in AI startups, with some arguing that these are financially illiterate [01:16:25]. If most of the capital is going to buying hardware and compute resources, it’s essentially subsidizing CapEx, a low-yield activity better suited for banks offering lease lines [01:25:09]. This leads to high dilution for founders and limited returns for investors [01:28:31].
The history of over-funded startups, particularly in the dot-com and crypto bubbles, shows that constraint is vital for founders and huge early funding can be a distraction [01:22:04]. The inability to acquire smaller AI companies due to regulatory scrutiny further complicates the exit strategy for these heavily funded startups [01:33:00].
CalPERS’ Venture Capital Strategy
CalPERS, the largest public pension in the US, managing 800 million to $5 billion) [00:57:27]. This move is controversial given their historically “atrocious” returns in venture capital and past decision to keep exposure low [00:56:11]. Some view this as a potentially good time to invest, as venture capital may be at the “bottom third” of valuations [00:56:40], and aligns with the broader state of the market and economic outlook. However, there’s a strong critique regarding the “sloppy” and “imbecilic risk management infrastructure” at CalPERS for not capitalizing on the tech boom in its own backyard for decades [00:59:12].
The Future of Search and Data Monetization
The fundamental question of how the web’s open architecture will be re-architected in the age of AI remains [00:34:48]. The current “10 Blue Links” search model, which has been Google’s prolific business model, is suspect [00:37:30]. The shift is towards “knowledge extraction” and “intelligent agents” [00:45:29].
Two main models are emerging:
- Hyper-fragmented Relationships: Users will maintain independent relationships with specialized AI agents (e.g., travel, medical, financial) for privacy and control [00:39:19]. The concept of users becoming “servers” of their own data, with permission-based access to services, is proposed [00:40:05].
- General Intelligence Agents: Some, like those behind ChatGPT and Inflection AI’s Pi, believe in general intelligence that solves “vertical problems horizontally,” akin to a single personal assistant knowing all preferences [00:44:03].
The future likely lies in between, where a personal assistant AI handles general tasks but can “subcontract” work to specialized agents [00:45:10]. Google is making rapid progress with Bard, integrating images and linking to external sites, potentially monetizing through sponsored links [00:45:37]. While Google has massive distribution power and may eventually lead, it will face competition in this new AI-driven landscape [00:48:47].
Reddit’s API Strike and User-Generated Value
Reddit experienced a widespread strike where 95% of subreddits went dark, protesting Reddit’s decision to charge for API access [01:04:25]. This impacts third-party apps, with some facing annual costs of $20 million [01:04:57]. Reddit’s stated reason was to get paid for its data being used to train AI models by companies like Google Bard and ChatGPT [01:07:05].
This situation echoes similar historical issues with Facebook and Twitter, which initially had open APIs but later restricted them to control the user experience and monetize content directly [01:05:58]. The core issue is that much of Reddit’s value lies in its community and user-generated content, not solely in the company’s management or software [01:08:50]. This highlights a shift where content creators on platforms need to be compensated or given control over their data, akin to YouTube creators earning millions [01:12:06]. A potential solution could be allowing mods to monetize subreddits through subscriptions and revenue sharing with Reddit [01:13:59].