From: allin
The landscape of venture capital and startup strategies has been significantly shaped by recent economic shifts, market performance, and technological advancements like AI [08:27:00]. Discussions highlight market conditions for exits, funding approaches for new ventures, and the evolving role of investors.
Current Market Conditions
The Federal Reserve recently paused rate hikes, which would have been the 11th consecutive increase [08:27:00]. However, two more 25-basis-point hikes are anticipated before the end of the year [08:40:00]. The market, particularly tech stocks, experienced a significant rally, driven by a “reversion to the mean” after a devastating 2022 [14:08:00]. While the NASDAQ moved 30% to start the year, valuations for internet and software companies are still below their 10-year average [16:12:00], [14:51:00].
Despite the rally, some stocks, particularly AI-related ones, are perceived as having “gotten ahead of themselves” [17:48:00].
State of Initial Public Offerings (IPOs)
Several companies, including Reddit and Stripe, are being floated as 2023 IPO candidates [08:52:00]. ARM confidentially filed for an IPO in April, seeking to raise 32 billion write-down [09:10:00], [12:28:00].
For current IPOs, investors are demanding a “significant margin of safety” compared to deals done in 2020 and 2021 [10:11:00]. Companies like ARM and Databricks, despite being “fantastic companies,” must offer a “bigger discount” to get into the public markets [10:58:00], [11:14:00].
ARM’s valuation is a point of contention; while SoftBank acquired it for 40 billion due to regulatory scrutiny, current estimates for its IPO valuation range up to $70 billion [09:28:00], [12:22:00], [11:37:00]. Some argue ARM is a “mid-20s billion dollar company” now, based on its “Cash Cow business” and lack of significant product decisions beyond mobile processors [11:43:00], [12:02:00].
AI’s Influence on Valuations and Business Models
The recent market rally is partially attributed to enthusiasm for AI [22:00:00]. However, there is also a “radical realization” that businesses dependent on services replaceable by AI might not have long-term longevity [22:57:00]. The impact on venture capital and investment is visible, with significant demand for chips and infrastructure to build AI models [23:32:00].
A high “discount rate” is applied to industries where AI’s positive impact is further out (e.g., lawyers, investment banks), whereas industries like chip manufacturing see immediate valuation bubbles [24:08:00], [24:32:00], [24:38:00].
Venture Capital Funding and Strategies
”Heat Check” Rounds and Capital Allocation
Recent funding rounds, such as Mistral AI’s 240 million valuation, are described as “heat checks” [01:15:19], [01:16:25]. These large early-stage investments are driven by the need to acquire expensive H100 and A100 GPUs for AI model training [01:16:00].
Critics argue that VCs are “subsidizing capex” by funding hardware purchases rather than groundbreaking IP [01:25:12], [01:29:55]. This is likened to the dot-com bubble where money was “flushed down the drain” on data center capacity [01:28:54].
“When you put in a hundred million dollars into a startup to buy compute, you are not buying whiz-bang Next Generation IP, you are subsidizing capex” [01:25:09]
The cost curve for AI model training is rapidly decreasing, similar to DNA sequencing, with potential 100x cost reductions in 18-36 months [01:35:13], [01:38:33]. This makes large investments in foundational models today a risky “lottery ticket” [01:25:02].
Optimal Fund Size and LP/GP Dynamics
The discussion touches on the “optimal Venture fund size” [01:40:01]. Large funds (over $1 billion) can participate in companies requiring more capital, but questions arise about the quality of investment decisions [01:41:33]. Critics express concern that “too many things are getting funded,” with a lower “margin of return” being required [01:41:57].
There is significant pressure on GPs from LPs for distributions, especially after market downturns [01:54:26]. This can lead to firms distributing shares too early, missing out on potential upside [01:54:41]. The lack of Public Market investing experience among many venture capital challenges and market resets is noted as a problem [01:55:27].
Calpers’ Strategy
California Public Employees’ Retirement System (Calpers), the largest public pension fund in the US, is increasing its venture capital allocation six-fold, from 5 billion [01:57:27], [01:57:47]. This comes after years of low exposure (1%) and “atrocious” returns [01:57:42], [01:56:13]. While some view this as a smart move given that venture is “at the bottom or bottom third” of valuations [01:56:40], others question the “imbecilic risk management infrastructure” that led to such low historical allocation in Silicon Valley’s backyard [01:59:43]. The concern is that if the process isn’t fixed, increased allocation could just lead to more losses [02:00:15].
The Importance of Writing Deal Memos
Writing clear, long-form deal memos (2-5 pages) rather than relying solely on “stupid performative decks” is emphasized for clarity of thought and attracting thoughtful critique [01:02:41], [01:04:04]. Decks are seen as dangerous for decision-making as they can “dupe somebody else” through graphics and lead to “group think” [01:03:10], [01:03:51].
Creator Economy and Data Ownership
The Reddit API changes, which led to a widespread mod strike, highlight the tension between platforms seeking to monetize user-generated content and the communities that create the value [01:04:23], [01:07:11]. Reddit’s move to charge for API access, in part for AI model training, mirrored similar actions by Facebook and Twitter to control user experience and prevent third-party apps from “sucking users off the platform” [01:05:07], [01:06:00], [01:10:13].
The “value of Reddit is inherent in the community” and its content creators [01:08:58]. This event suggests a shift where “the value is going to go…more towards the individual people…and not to the centralized organization” [01:14:40]. A proposed solution for Reddit is to offer a “healthy rev share” or allow paid subscriptions to subreddits, splitting revenue with mods [01:11:59], [01:13:59].
The broader shift towards users becoming “servers” of their own data, with individual IP addresses controlling access and being able to “rent” their data to services, could lead to hyper-fragmented, best-in-class interactions rather than a single aggregated service [01:40:05], [01:39:19].
Regulatory Impact on Investment and Innovation
The current regulatory environment in Washington D.C. makes it difficult for hyperscalers to acquire AI companies for over $1 billion [01:33:00]. This “unintended consequence” means entrepreneurs and venture capitalists building traction might not find an exit home like Instagram or WhatsApp did, potentially hindering innovation as many startups cannot become big, profitable businesses on their own [01:33:34], [01:33:55].
This situation creates a “two-sided problem”: more capital is being spent to fund and start companies, while the traditional “downside protection” of acquisition has been undermined [01:34:32].
The Role of Constraint in Startups
Over-funding can be a “huge distraction to Founders” [01:22:17].
“Constraint makes for great art, constraint makes for great startups. You need to have pressure on a startup for them to deliver. You cannot give startups five years of runway and expect it’s going to work.” [01:30:21]
The milestone-based funding system, where companies raise in stages (e.g., 25-50 million after a year or two), is seen as ideal [01:22:12]. Excessive early funding can lead to high salaries and disincentivize efficient product development [01:22:31], [01:23:32].