Liberation Day, Tariffs, US v China Open Source, OpenAI, $CRWV, TikTok | BG2

This episode unpacks the seismic shifts from Trump's tariff announcements impacting global markets to the strategic battles unfolding in open-source AI and landmark funding deals shaping the compute landscape.

Trump's Tariff Announcement ("Liberation Day") and Market Reaction

• The discussion kicks off analyzing the immediate market fallout following President Trump's "Liberation Day" tariff presentation. Brad notes this announcement was anticipated, highlighting his long-held view that the administration is doctrinally serious about reshaping trade policy, not just using tariffs as a negotiation tactic. The initial market reaction was volatile; a premature report of 10% across-the-board tariffs caused a brief market jump, but as the presentation detailed higher reciprocal tariffs, futures (S&P, Nasdaq) plummeted, indicating significant market anxiety about the potential economic impact.
• Market Volatility: S&P and Nasdaq futures experienced a sharp 600 basis point swing from their initial jump to their eventual low during the presentation, reflecting immediate negative sentiment.
• Brad Gerstner observes: "the S&P futures the Q the NASDAQ futures start sinking they had a 600 basis point fall between where they initially jumped and where they ended up so the market is not liking this at all."

Decoding the Tariffs: Details and Exemptions

• The speakers delve into the specifics of the announced tariffs, noting the administration's concept of "reciprocal tariffs" which includes not just explicit tariffs but also non-tariff trade barriers (factors like currency manipulation or restrictive judicial actions). This calculation method allows for significantly higher effective tariff rates, such as the proposed 54% on China (building on an existing 20%). Key tariff buckets include 25% on autos (targeting Mexico, Canada, Germany), the broad reciprocal tariffs applied country-by-country effective April 9th, a minimum 10% tariff on all countries, and the exceptionally high rate for China.
• Significant Exemptions: Crucially, Brad highlights exemptions announced shortly after, including pharmaceuticals and, notably for this audience, semiconductors. This significantly alters the total impact, potentially bringing the estimated $750 billion headline figure closer to Peter Navarro's $600 billion estimate once exemptions are factored in.
• Taiwan Example: Despite a 32% proposed tariff on Taiwan, semiconductors are exempt, showcasing the strategic importance placed on this sector.

Market Uncertainty and Business Impact

• Bill Gurley emphasizes the detrimental effect of ambiguity on business planning and investment. He argues that even with announced tariffs or exemptions, the lack of predictability regarding future policy changes makes long-term capital expenditure (like relocating factories) extremely difficult and risky. This uncertainty freezes decision-making, potentially leading to hiring pauses and slower economic activity, aligning with recent Fed forecast adjustments (lower GDP growth, higher inflation/unemployment).
• CEO Uncertainty: Brad shares anecdotes from CEO conversations during the announcement, highlighting their confusion ("Do you see us exempted?") and reports of cancelled European/Asian contracts due to geopolitical friction.
• Bill Gurley states: ...you're left not knowing what what's going to be the policy 3 months from now 6 months from now 12 months from now which makes it very hard to allocate capex in any meaningful way whatsoever.

Geopolitical Fallout and Strategic Implications

• The tariffs are causing unusual geopolitical alignments. Brad notes reports of potential collaboration between China, Korea, and Japan in response to US tariffs – a rare occurrence. Furthermore, European leaders (like President Macron) are engaging more closely with China, seeking stronger trade relationships, a development Bill had previously warned about. This suggests the tariffs could inadvertently push traditional US allies towards competitors.
• Strategic Risk: The potential for strengthened economic ties between China and other major economies (including Europe) poses a long-term strategic risk for US influence and market access.

Looking Ahead: Tariff Negotiations and Key Watch Points

• Brad expresses a belief that the announced high tariff levels represent a starting point for negotiation, predicting the final impact will be closer to $300-400 billion rather than $600-750 billion. He suggests political realities, like the need for Republican unity to pass the President's desired tax reconciliation package, will moderate the administration's stance. Key factors to watch in the next 30-60 days include whether semiconductor/pharma exemptions hold, the progress of country-by-country negotiations, and, most critically, the trajectory of US-China trade talks.
• Investment Timing: Brad suggests that while clarity is lacking, the current fear might present buying opportunities for investors willing to act amidst uncertainty.

The Strategic Landscape of Open Source AI: US vs. China

• The conversation shifts to the rapidly evolving open-source AI landscape, focusing on developments in China (e.g., DeepSeek) and the US (OpenAI, Meta). This follows recent announcements, including OpenAI signaling a move towards releasing powerful open-weight models.

China's Embrace of Open Source: History and Motivation

• Bill Gurley provides crucial historical context, countering the narrative that China's recent open-source AI moves (like DeepSeek) are forced or purely strategic reactions. He argues China has embraced open source for over a decade, viewing it positively as a way to counter long-standing accusations of IP theft and ensure access to foundational technologies (like Linux) independent of potential Western export controls. This established comfort with open source informs the current generation of AI entrepreneurs.
• Bill Gurley explains: "they've been accused of stealing tech IP for years and so when something like open source comes along this looks like the best thing possible right there's no one that can accuse us of IP theft because there is no IP ownership in an open-source world."

Open Source as a Defensive Strategy: Kubernetes, Llama, and DeepSeek

• Bill further explains how major US tech companies use open source defensively to level the playing field when they aren't market leaders. He cites Google's use of Kubernetes (an open-source container orchestration system) to counter AWS's cloud dominance and Meta's release of Llama to disrupt the closed AI model space. This strategy fosters competition and benefits consumers but can be used to commoditize competitors' advantages.
• DeepSeek Concerns: DeepSeek's success is noted (hosted on AWS/Google, widely forked), alongside rising concerns in Washington D.C. about the proliferation of Chinese open-source AI models, potentially leading to regulatory limits on its use in the US.

OpenAI's Shift Towards Open-Weight Models

• OpenAI's recent announcement about releasing a powerful new open-weight language model (a model whose parameters/weights are publicly released, allowing inspection and modification, though potentially with usage restrictions) is analyzed. Sam Altman's public statements indicate a commitment to this, potentially even exceeding Meta's Llama in openness by not imposing usage restrictions for large-scale deployment, a direct jab at Meta's Llama license terms.
• Sam Altman quote (paraphrased from Brad): When asked about licenses like Meta's, Sam replied "No," indicating a potentially more permissive open model.

Competitive Dynamics: OpenAI, Meta (Llama 4), and the Open Source Race

• The timing of OpenAI's announcement is significant, preceding Meta's upcoming LlamaCon developer event where the launch of Llama 4 is widely anticipated. Rumored Llama 4 specs include 400B parameters, a mixture-of-experts architecture, and a massive 10M token context window. OpenAI's move intensifies the competition for dominance in the open-source AI space, particularly as a US-based alternative should regulators restrict models like DeepSeek.
• Strategic Importance: Bill emphasizes that the next 3-6 months are critical, as companies' decisions on model openness could determine market leadership for the next 5-10 years. The competition isn't just domestic; Western models must also compete globally with offerings like DeepSeek.

OpenAI's Strategic Rationale and Business Model Evolution

• Bill suggests OpenAI's open-source move is strategically clever: it builds developer ecosystems, potentially offloads massive compute costs (inference runs on others' hardware), and pressures competitors focused on closed API models, protecting OpenAI's flank as it focuses more on consumer products (ChatGPT). Brad concurs, adding that Sam Altman views models as commoditizing and the real battleground shifting to products/services. While ChatGPT is a consumer focus, the consumerization of the enterprise (where popular consumer tools drive enterprise adoption, like iPhone adoption) fuels OpenAI's enterprise growth.
• Model Commoditization: Sam Altman believes foundational AI models are becoming widely available, shifting the competitive focus to differentiated products and user experiences built on top.

OpenAI's Landmark Funding Round and Valuation Analysis

• The discussion covers OpenAI's recently announced funding round led by SoftBank, often headlined as $40 billion but structured in tranches (approx. $10B then $30B). The round reportedly values OpenAI at $260 billion pre-money ($300B post if fully funded). Brad contextualizes this valuation, noting market leaders rarely look cheap, referencing past investments in Google and Meta.
• Valuation Comparison: Brad compares OpenAI's implied forward revenue multiple (around 20x based on $13B projected revenue) favorably against Anthropic (est. 50x) and the X/X.AI merger (est. 80x), suggesting OpenAI's valuation, while high, is relatively lower than its direct peers.
• Brad Gerstner notes: "...relative to their peers um it certainly appears to me like uh you know 20 versus 50 versus 80 it's hard to say that this would be more expensive on a multiple basis than Anthropic or uh or or X.AI."

The Coreweave ($CRWV) IPO: AI Infrastructure Goes Public

• The speakers discuss the recent IPO (Initial Public Offering) of Coreweave ($CRWV), a specialized cloud provider focused on GPU compute for AI. Brad (whose firm Altimeter is an investor) acknowledges the challenging market timing ("category 5 hurricane" of tariff news) and initial stock dip below the offering price. However, recent customer announcements, including a significant deal with Google for Nvidia Grace Blackwell GPUs, have boosted the stock and diversified Coreweave's customer base beyond Microsoft.
• Customer Diversification: Coreweave now counts Microsoft, Meta, Google, OpenAI, Nvidia, Cohere, and Mistral among its clients, solidifying its position as a key AI infrastructure provider.

Analyzing Coreweave's Business Model and Depreciation Debate

• A key debate surrounding Coreweave involves the depreciation schedule for its GPUs (expensing the cost of an asset over its estimated useful life). Nvidia CEO Jensen Huang's comments suggesting older GPUs quickly lose value fueled criticism of Coreweave's potentially longer depreciation timeline (e.g., 6 years vs. a suggested 2-4 years). Brad counters that GPUs retain value for inference long after they are cutting-edge for training, citing the continued use of older A100s (2020) and V100s (2017). He argues Coreweave's model relies on multi-year contracts (e.g., 4 years) designed to recoup costs within 3 years, making year 4 profitable even without assuming longer GPU lifespans.
• Unit Economics: Brad outlines Coreweave's model: secure multi-year contracts, pay back capex/opex/GPU costs in ~3 years, with the final contract year(s) representing profit margin, plus potential upside from GPU use beyond the contract term.
• IPO Performance Context: Bill provides a list of major tech companies (Salesforce, Netflix, Amazon, Facebook, etc.) that also "broke issue" (traded below their IPO price) initially, cautioning against judging a company's quality solely on early post-IPO trading.

The TikTok Saga: Potential Deal Structure and Geopolitical Entanglement

• The discussion turns to the ongoing TikTok situation, with a potential deal deadline approaching (April 5th). Brad (a shareholder in parent company ByteDance since 2015) outlines the rumored structure for TikTok US: a new entity ("NewCo") would be formed, with ByteDance retaining a minority stake (under 20%), existing ByteDance investors (like Altimeter) receiving shares, and new US investors (rumored names include Amazon, Andreessen Horowitz, Oracle) contributing fresh capital (~50% ownership). This NewCo would license the TikTok algorithm from ByteDance and manage US data/operations, likely involving Oracle due to existing hosting arrangements.
• Strategic Breakthrough: A key shift enabling this potential deal is the move away from forcing algorithm separation towards a licensing model, allowing TikTok US and global TikTok to share the core algorithm under US oversight.
• Unlocking Value: Brad emphasizes the deal's importance for unlocking the massive unrealized value in ByteDance (estimated fair value $800B-$1T, 60% US-owned) for its venture investors (pension funds, endowments), potentially triggering a huge return of capital (DPI) upon a ByteDance IPO or exit, which hinges on resolving the TikTok US issue.
• Geopolitical Hurdle: Despite progress, the newly announced tariffs on China create uncertainty, as the Chinese government could still block the deal as part of broader trade negotiations.

Broader Implications for US-China Relations

• Brad concludes on an optimistic note, believing President Trump ultimately wants a comprehensive trade deal with China's President Xi. He views the high tariffs and TikTok negotiations as interconnected parts of a larger strategic dialogue. Resolving these issues is crucial for global economic stability and growth, despite the underlying competitive tensions between the two superpowers.

Conclusion

• This episode highlights the critical intersection of macroeconomic policy (tariffs) and the rapidly evolving AI sector, impacting everything from market sentiment to open-source strategy and infrastructure investment. Crypto AI investors and researchers must closely monitor both geopolitical negotiations, particularly US-China relations, and the strategic platform moves within AI, especially concerning compute access and model openness.