JD Vance's AI Speech, Techno-Optimists vs Doomers, Tariffs, AI Court Cases with Naval Ravikant

It is too early to judge this prediction as right or wrong.

Key parts of Friedberg’s claim:

1. It is explicitly conditional and long‑term:

   • “If” the EU and U.S. adopt a techno‑pessimistic, highly regulatory stance on AI and
   • “over the coming years” China’s GDP will scale while U.S. GDP stagnates relative to China.

   As of November 30, 2025, less than one year has passed since the February 15, 2025 podcast. That is far too short to evaluate multi‑year structural GDP divergence.

2. Policy condition is not clearly resolved yet:

   • The EU has indeed been moving toward heavy AI regulation via the EU AI Act, often cited as one of the world’s most restrictive AI regulatory frameworks.
   • The U.S. has taken a more fragmented approach: an AI Executive Order, sectoral proposals, and some state‑level rules, but not yet a comprehensive, clearly “techno‑pessimistic” regime. Policy direction is still evolving and contested.
   • China has been promoting rapid AI deployment within a state‑controlled framework, combining industrial policy support with content and safety regulations; whether this is best characterized as "embracing" vs. "constraining" AI is itself debated.

   Because the exact scenario he described (strongly pessimistic, highly regulatory U.S. and EU versus a clearly more liberal, growth‑oriented China) is still in flux, the antecedent of the “if…then…” prediction is not firmly in place yet.

3. GDP outcomes have not had time to manifest in the structural way implied:

   • Real GDP levels and growth for 2024–2025 for the U.S. and China can be estimated or projected by IMF/World Bank/OECD data, but a single year or two of growth data cannot confirm a multi‑year structural stagnation of U.S. GDP relative to China driven specifically by AI regulatory stance.
   • Even if early data showed China growing somewhat faster or slower than the U.S., attributing that to AI regulatory philosophy (rather than real estate cycles, demographics, trade, fiscal policy, etc.) would be speculative at this stage.

Because:

• the time horizon (“over the coming years”) has not elapsed,
• the policy precondition is not fully realized or stable, and
• GDP trends and causal attribution cannot yet be robustly assessed,

this prediction cannot be declared right or wrong at this time and is best classified as inconclusive.

The prediction has a 10‑year horizon ("in the next ten years"). From the podcast release on 15 February 2025, that implies outcomes by roughly 2035, while today is 30 November 2025—less than one year into the window.

To evaluate it fully, we would need to know by ~2035 whether:

1. The U.S. lost millions of driving‑related jobs (truck drivers, delivery drivers, taxi/TNC drivers, etc.) primarily due to AI/self‑driving technologies, and
2. That labor‑market shift caused U.S. immigration policy to become more restrictive, with fewer people allowed to immigrate.

As of late 2025:

• Human driving jobs remain widespread; self‑driving tech is being piloted and deployed in limited commercial settings, but there is no evidence yet of multi‑million‑person job loss attributable to it at national scale.
• U.S. immigration levels and policies in 2024–2025 are shaped mainly by politics, border enforcement debates, and legislative/executive actions, not by documented large‑scale displacement of drivers.

Because the specified 10‑year period has not remotely elapsed, the prediction cannot yet be judged right or wrong, even if early indicators are modest. It is therefore too early to evaluate.

This prediction is about a conditional and multi‑year outcome:

If the U.S. hobbles its AI industry with unnecessary regulations while China does not, then China will win the AI race over the ensuing years.

As of 30 November 2025, both sides of that conditional and the long‑term outcome are not clearly settled:

1. Has the U.S. “hobbled” its AI industry with heavy regulations while China stayed lax?

   • Federally, the U.S. actually moved toward deregulation in 2025: Trump’s Executive Order 14179 is explicitly titled “Removing Barriers to American Leadership in Artificial Intelligence” and rescinds Biden’s more restrictive Executive Order 14110, aiming to roll back policies seen as constraints on AI development. (en.wikipedia.org)
   • Some U.S. states, notably California, have introduced targeted frontier‑model safety and transparency rules (e.g., SB‑53 / Transparency in Frontier Artificial Intelligence Act), but these are still relatively new, and there is an ongoing political fight over whether federal law will pre‑empt stricter state regulation. (en.wikipedia.org)
   • China, for its part, has also rolled out substantial AI regulations, including 2023–2025 generative‑AI rules requiring algorithm registration, security reviews, and mandatory labeling of AI‑generated content, with new labeling measures taking effect in September 2025. (cimplifi.com)
     Given this, it is not accurate to say the U.S. is clearly hobbling itself while China refrains from similar or stronger controls; both are regulating, but in different ways.

2. Has China clearly “won” the global AI race by late 2025?
   Available indicators point to a competitive, mixed picture rather than a decisive Chinese win:

   • On investment and market leadership, the U.S. remains far ahead in private AI investment (about $109B vs. China’s $9.3B in 2024) and is widely projected to remain the front‑runner in global AI market share and frontier model development through 2030, with China as a fast‑growing second pole. (allaboutai.com)
   • On overall capability indices, a 2025 comparative AI index finds the U.S. composite AI score (68.1) still exceeds China’s (59.4), even though China is catching up quickly and shows strong regional clusters. (arxiv.org)
   • On research output and specific niches, China leads the world in share of AI publications and is surging in some model benchmarks and in the global market for open‑weight models, where recent work shows Chinese models slightly surpassing U.S. models in download share (17% vs. 15.8%) and performance on some tests. (arxiv.org)

3. Timeframe: “over the ensuing years”
   The statement was made in February 2025; only about nine months have elapsed. The prediction explicitly concerns a multi‑year trajectory (“ensuing years”) of global leadership. Current data show an ongoing, evolving competition rather than a settled outcome, and many forecasts still treat 2025–2030 as the decisive window rather than a period whose results are already known. (allaboutai.com)

Because (a) the antecedent conditions (U.S. uniquely hobbling itself while China does not) have not clearly materialized, and (b) the long‑term outcome (China definitively winning the AI race) cannot reasonably be judged this early and is not reflected in present indicators, the prediction cannot yet be evaluated as right or wrong.

So the appropriate classification is “inconclusive (too early to tell).”

Multiple major tech companies launched and marketed AI agents as a distinct, commercially available product category throughout 2025, matching Sacks’s prediction of “a whole new wave” of agentive products.

Evidence of a broad 2025 agent wave:

• Microsoft security agents (early 2025): At Microsoft Secure and RSA Conference 2025, Microsoft rolled out Security Copilot agents—described as autonomous AI designed to tackle high‑volume security tasks—and began phased public preview to paying enterprise customers.(techcommunity.microsoft.com)
• Microsoft 365 Copilot agents & Agent 365 (late 2025): At Ignite 2025, Microsoft announced Word/Excel/PowerPoint “agents in chat,” an “Agent Mode” in Office apps, and Agent 365 as a control plane for managing large numbers of agents across an organization. These capabilities are positioned as a new agent ecosystem layered on top of the already widely deployed Microsoft 365 Copilot, which Microsoft says is used by over 90% of the Fortune 500.(news.microsoft.com)
• Salesforce Agentforce 360 (Oct 2025): Salesforce launched Agentforce 360, an AI agent platform that automates routine workflows across its cloud products. Reuters reports that Agentforce 360 already had about 12,000 customers at global launch, indicating substantial commercial uptake rather than a niche experiment.(reuters.com)
• Consumer/SMB autonomous agents (Manus): Butterfly Effect’s Manus AI agent, launched March 6, 2025, is marketed as an autonomous agent capable of independent reasoning and multi‑step task execution, available as a web, iOS, and Android product—one of the first general-purpose consumer agent apps.(en.wikipedia.org)
• Cross‑industry framing of a new “agentic” era: Industry coverage in 2025 explicitly talks about the “age of agentic AI,” describing agents as a shift from simple copilots or chatbots to systems that plan, reason, and execute multi‑step workflows across tools and data sources.(itpro.com) In parallel, analysts like IDC (via Microsoft) forecast over a billion AI agents by 2028, reinforcing that vendors and enterprises regard “agents” as a distinct, emerging product category rather than a mere UI tweak.(microsoft.com)
• Commerce and vertical agents: Financial analysis from Morgan Stanley in 2025 discusses “agentic commerce” and AI shopping agents used by major retailers (e.g., Amazon’s Rufus and other shopping assistants), treating agents as a new way consumers will interact with e‑commerce by autonomously handling search, comparison, and purchase flows.(businessinsider.com)

Across enterprise security, productivity suites, CRM, and consumer/mobile, 2025 saw a clear wave of explicitly branded AI “agents” and “agentic” products moving beyond simple chat to delegated, multi‑step task execution. These products were commercially available (often to large paying customer bases) within the 2025 calendar year, and industry commentary treats them as a notable new category. That aligns closely with Sacks’s forecast of “a whole new wave” of agentive products arriving in 2025, so the prediction is right.

Multiple substantial new U.S. import tariffs were in fact implemented in 2025, going well beyond the early‑2025 status quo and forming a central pillar of economic policy, not just rhetoric.

Key developments:

• On February 1, 2025, President Trump signed executive orders imposing 25% tariffs on all goods from Mexico and Canada (with a 10% rate on Canadian energy) and an additional 10% tariff on Chinese imports on top of existing duties; these measures began taking effect in early February and March 2025. (en.wikipedia.org)
• On April 2, 2025 (“Liberation Day”), Trump declared a national emergency over the U.S. trade deficit and signed Executive Order 14257. This imposed a universal 10% baseline tariff on nearly all imports starting April 5, 2025, plus a tiered schedule of higher “reciprocal” tariffs (up to roughly 50%) on countries with large trade surpluses with the U.S. These orders were explicitly framed as a core economic policy to address trade deficits and generate revenue. (en.wikipedia.org)
• Additional, targeted tariffs were layered on: expanded steel and aluminum tariffs to all countries effective March 12, 2025; a 25% global tariff on automobiles and many auto parts starting April–May 2025; and further 2025 tariff hikes on copper, lumber, and furniture, all of which significantly increased effective import taxation relative to early 2025. (whitecase.com)

Because these tariffs:

1. were new and material compared with the pre‑2025 regime, and
2. were actually implemented as major elements of the administration’s 2025 economic policy (not just talked about), Chamath’s prediction that “the tariffs will be a plug… Are they coming? Absolutely.” is borne out by subsequent events.

Available data show that by roughly 6–9 months after February 2025 (August–November 2025), the U.S. federal funds rate target range was well below the 5.0–5.5% band Chamath speculated about.

Key facts:

• By late 2024, the FOMC had already cut the target range from its 5.25–5.50% peak (reached July–September 2023) down to 4.25–4.50%, and that range was still in place as of early February 2025. (federalreserve.gov)
• The Fed’s June 2025 Monetary Policy Report and contemporaneous analysis confirm that from January through June 2025 the federal funds target range was held at 4.25–4.50%, below 5%. (federalreserve.gov)
• On September 17, 2025, the FOMC cut the target range to 4.00–4.25%. (haver.com)
• On October 29, 2025, the Fed cut again, to 3.75–4.00%, where it remained through at least mid‑November 2025. (pcbb.com)
• Time‑series data summarizing the "federal funds target range – upper limit" show values of 4.50% on July 30, 2025, 4.25% on September 17, 2025, and 4.00% on October 29, 2025, with no return to 5% or higher during that period. (tradingeconomics.com)

Since during the relevant 6–9 month window (August–November 2025) the federal funds target range never was and did not return to the 5.0–5.5% region Chamath described, the prediction that rates in that timeframe could be around 5–5.5% (implying they would remain there or move back up to that range) turned out to be incorrect in light of actual policy outcomes.

Key parts of Friedberg’s scenario have already occurred, but the decisive "tens of billions" in new tariff‑driven farm transfers are not yet clearly in place, and Trump’s second term is still in its first year.

• A future Trump administration did in fact impose a new round of broad tariffs on China in 2025, including a 10% baseline tariff on almost all imports and a higher “reciprocal” rate on China of about 34%, on top of earlier 2025 measures that doubled tariffs on Chinese goods from 10% to 20%. (en.wikipedia.org)
• China responded exactly as predicted in direction: it imposed additional 10–15% retaliatory tariffs on a wide range of U.S. agricultural and food products (covering hundreds of tariff lines) and also suspended soybean import licenses for several U.S. firms and halted some U.S. log imports, clearly targeting U.S. agriculture. (fas.usda.gov)
• However, while there are substantial government payments to farmers in 2025, most are from pre‑existing or broadly framed programs (e.g., a $10 billion Emergency Commodity Assistance Program for low prices, of which over $8 billion has been paid, plus large disaster payments projected at about $35.7 billion) rather than a clearly identified, new, tariff‑specific bailout on the scale of the 2018–19 trade‑war programs. (reuters.com)
• A new trade‑related bailout in the $10–15 billion range has been reported as planned, and over $3 billion in CCC‑funded aid tied to trade disputes is being released, but reporting as of late October–November 2025 treats these packages as pending or partial; analysts describe tens of billions in tariff‑funded farm aid as a likely outcome if the trade war continues, not as an accomplished fact. (reuters.com)

Because (1) the tariff and Chinese‑retaliation components of the prediction are already borne out, but (2) the scale and explicit tariff‑linkage of new farm transfer payments have not yet clearly reached “tens of billions” attributable to the 2025 China episode, and (3) Trump’s term runs through January 2029, it is too early to say definitively whether the full prediction has come true. Hence the result is classified as inconclusive rather than right or wrong at this time.

In 2025 the U.S. did enact exactly the kind of new, broad tariff rounds Friedberg was talking about: Trump layered a 10% “fentanyl” tariff on all Chinese imports, added sweeping reciprocal tariffs of 10–50% on many countries, and imposed 25% duties on Canada, Mexico, and others under various executive orders. (reuters.com) China initially retaliated with steep tariffs but then agreed to a series of negotiated truces: a May 2025 Geneva deal slashed reciprocal rates from well over 100% to 10% for both sides, followed by an October accord that locked in lower tariffs for a year and bundled Chinese commitments on fentanyl control and resumed U.S. farm purchases, i.e., clear partial concessions within a negotiated settlement rather than sustained maximal retaliation. (china-briefing.com) Canada first imposed broad 25% counter-tariffs, then in August 2025 dropped most of them on USMCA‑qualifying goods to “match” U.S. exemptions, a move widely described as easing or retreating from its hard line, while sectoral tariffs on U.S. steel, aluminum, and autos remained – again, a partial concession embedded in ongoing talks rather than indefinite full retaliation. (cnbc.com) Mexico, similarly, secured exemptions for USMCA‑compliant exports from the global tariff hike and later obtained a 90‑day reprieve from additional planned U.S. tariffs to negotiate a broader arrangement, instead of implementing or sustaining maximum counter‑tariffs of its own. (reuters.com) The EU suspended planned 25% countermeasures, then agreed to a July–August 2025 framework where it will eliminate tariffs on U.S. industrial goods and grant preferential access for many U.S. farm and seafood exports, while accepting a 15% U.S. tariff on most EU goods and pausing its retaliation for at least six months – a textbook negotiated compromise with significant European concessions. (theguardian.com) Some disputes, notably with India, remain in flux, but even there officials on both sides now openly expect a tariff truce and trade deal that would adjust the punitive U.S. rates, not lock in India’s maximum retaliatory posture. (reuters.com) Taken together, the major trading partners targeted in this new tariff round have indeed moved toward partial concessions and negotiated settlements rather than maintaining maximal retaliatory tariffs indefinitely, matching Friedberg’s prediction in substance.

As of November 30, 2025, the key pieces of Jason’s prediction have not resolved in a way that lets us clearly mark it right or wrong.

1. NYT v. OpenAI decisive loss + injunctions (given 5–10% probability)

   • The New York Times’ copyright lawsuit against OpenAI and Microsoft is still ongoing. In April 2025, Judge Sidney Stein allowed the core copyright claims by the Times and other newspapers to proceed while dismissing some other claims; this was a procedural ruling, not a final judgment on the merits, and no broad injunctions on OpenAI’s practices have been issued. (apnews.com)
   • Docket activity from October–November 2025 shows the parties still litigating discovery disputes, with no final liability ruling or settlement recorded. (dockets.justia.com)
   • Because the case is unresolved and no “decisive win with injunctions” has occurred (nor been ruled out), we can’t evaluate whether assigning a 5–10% probability was well‑calibrated. Probabilistic forecasts of ongoing events generally can’t be scored until the outcome is known.

2. Industry endgame: Napster→Spotify-style regime with 50–66% revenue share to rightsholders

   • Since early 2024, OpenAI and other AI companies have signed a series of licensing deals with major publishers (e.g., Axel Springer, Dotdash Meredith, News Corp, TIME, Financial Times, Vox, Le Monde, Prisa, Reuters). These deals involve lump‑sum or relatively modest annual payments, not a dominant share of AI providers’ total revenue. For example, reporting on OpenAI’s Axel Springer and Dotdash Meredith deals suggests figures on the order of $10–16 million per year, amounting to roughly ~1% or less of those publishers’ revenues and an even smaller fraction of OpenAI’s. (niemanlab.org)
   • Other arrangements, like Perplexity’s plan to share some advertising revenue with publishers whose content is cited, are still early-stage and framed as incremental revenue streams rather than a Spotify‑style commitment of half or more of platform revenue. (techcrunch.com)
   • Meanwhile, OpenAI’s major revenue‑sharing obligation is to Microsoft (around 20% of OpenAI revenue under their partnership, with discussion of reducing this over time), not to content rightsholders. (reuters.com)
   • The broader legal and business landscape for AI copyright (including suits by authors, music labels, and news organizations) remains in flux, with some settlements (e.g., in AI music) and many cases still active. (apnews.com)

Overall, while there is clear movement toward more licensing and some revenue sharing, nothing like a stable, Napster→Spotify-style regime where large, especially closed-source, AI model providers pay 50–66% of their revenue to content rightsholders has emerged by late 2025. At the same time, the ultimate structure of the market and the final outcomes of key lawsuits (including NYT v. OpenAI) are still undecided. Because both the specific legal outcome and the long‑run industry structure Jason is predicting are unresolved, the fairest status is **“inconclusive (too early)” rather than clearly right or wrong.