E139: Recapping Chamath's wedding, VC surplus, unions vs Hollywood, room-temp superconductors & more

As of late 2025, there isn’t clear evidence that the 2023 Hollywood writers’ and actors’ strikes have backfired in the way Chamath predicted.

1. What the unions actually achieved on AI
Both the Writers Guild of America (WGA) and SAG‑AFTRA made AI a core bargaining issue and secured explicit contractual guardrails:

• The 2023 WGA contract states that AI cannot write or rewrite literary material, AI‑generated text cannot be treated as “source material,” and companies cannot require writers to use AI tools; AI use must be disclosed and remains under union control on covered projects. (en.wikipedia.org)
• The 2023 SAG‑AFTRA TV/Theatrical deal requires clear consent and compensation for digital replicas, including pay tied to the days an actor would have worked and residuals for certain AI uses. These protections apply even when digital doubles or background replicas are used. (thewrap.com)
  These outcomes strengthened formal bargaining power on AI in union‑covered film/TV work, which is the opposite of an immediate erosion.

2. AI use has grown, but mostly as augmentation, not wholesale replacement
Since the strikes, Hollywood has seen a surge of interest in generative AI across VFX, editing, pre‑visualization and experimentation with synthetic performers, and industry coverage routinely describes an "AI freak‑out" and deep anxiety about the technology. (vanityfair.com) Netflix, for example, has used generative AI to accelerate special effects on shows, emphasizing speed and cost savings but framing it as enhancing human crews rather than replacing them. (theguardian.com) New ventures like Staircase Studios explicitly market AI as a way to lower budgets while still employing union talent at standard rates. (nypost.com)

3. Early attempts at synthetic actors have met union pushback
Projects like the AI‑generated “actor” Tilly Norwood have triggered strong condemnation from SAG‑AFTRA, which argues such uses may violate the 2023 contract and threaten livelihoods. (washingtonpost.com) The fact that unions can credibly threaten contractual and legal consequences suggests they retain substantial leverage, at least within the studio system that signed those agreements.

4. Ongoing organizing has extended, not reduced, AI protections
SAG‑AFTRA’s 2024–2025 video game strike—explicitly focused in part on AI—ended with new consent and disclosure rules around digital replicas and guarantees against certain forms of AI substitution for performers in games. (en.wikipedia.org) That’s further evidence that, post‑2023, unions are using their bargaining power to shape how AI is deployed rather than being structurally sidelined by it.

5. Why the prediction is still hard to judge
Chamath’s normalized claim is about long‑term dynamics: that the strike will accelerate AI adoption to the point of reducing or eliminating the need for human writers and actors, thereby undermining their bargaining power and job security. In the short run (2023–2025):

• Union contracts in core Hollywood film/TV have added AI constraints and economic gains, not visible erosion of formal power.
• AI adoption is clearly accelerating across media, but so far it is primarily augmenting workflows, marketing, and niche or experimental projects, with union‑covered writers and actors still central to mainstream production.
• Any eventual structural displacement of human talent by AI will likely play out over a much longer horizon than two years, and it’s not yet clear how much of that will be caused by the 2023 strikes versus broader technological and economic trends.

Because of this mix—stronger contract protections now, rising but mostly complementary AI use, and the inherently long‑term nature of the prediction about “disintermediation” and job security—the available evidence does not yet let us say the prediction is clearly right or clearly wrong.

So the fairest assessment as of November 30, 2025 is: it’s too early to tell whether the strikes will ultimately weaken or strengthen writers’ and actors’ long‑term bargaining power in the face of AI.

Friedberg relays a model in E139 that projects, for the leading public tech companies, that over the next 10 years their combined market cap will rise from about $10T to $25T, with roughly 70% of tech value creation accruing to public markets and 30% to private markets.(podscripts.co) The episode was released July 27, 2023, so the 10‑year window runs roughly to mid‑2033.(allinchamathjason.libsyn.com) As of November 30, 2025, only a bit more than two years of that period have elapsed, so the forecast horizon has not completed. Current data show that the usual Big Tech names already have very large combined market caps (Apple ≈$4.1T, Microsoft ≈$3.6T, Alphabet ≈$3.9T, Amazon ≈$2.5T, Meta ≈$1.6T, putting just those five around the mid‑teens in trillions), which is consistent with significant growth from the earlier ~$10T baseline but does not yet tell us whether they will reach ~$25T by 2033 or what the final public‑vs‑private split will be.(companiesmarketcap.com) Because the specified 10‑year period has not yet ended, and the public/private allocation is defined over that full decade, the prediction cannot currently be judged as right or wrong.

Friedberg’s prediction was that as AI tools for scripting, directing, and rendering are adopted, eventually we could see “100 times more films” being made at about “1/100 the cost” of today’s studio films. In the episode he frames this as a forward-looking structural change (“we could see…”) rather than attaching any specific time horizon like “within a few years.” (podscripts.co)

As of late 2025, neither part of that quantitative claim has actually materialized:

• Volume of films: Global feature-film output remains in the thousands per year, not hundreds of thousands. For example, in 2022 India produced 1,691 feature films and the U.S. 1,361; 2024 data and box-office reporting show a film landscape of similar order of magnitude, with some markets (e.g., China) releasing fewer features in 2024 than in 2023 (612 vs. 792). This is nowhere near a 100× jump in the number of films. (en.wikipedia.org)
• Costs: Big studio projects still routinely have production budgets in the tens to hundreds of millions of dollars, with many recent tentpoles in the $130–300M range, and industry analyses still put the average major-studio film around ~$65M—far from a 100× reduction. (axios.com)

At the same time, AI tooling is clearly spreading into film and video workflows:

• Text-to-video and generative video models (Runway Gen-2/3, Sora, Pika, Lumiere, Dream Machine, etc.) have become commercially available and are increasingly used for pre‑visualization, concepting, VFX prototypes, and marketing content. (en.wikipedia.org)
• Major tools are being integrated into pro software (e.g., Adobe Premiere Pro adding Sora/Runway/Pika plug‑ins) and studios are signing direct deals (e.g., Lionsgate–Runway), while Netflix and AI-focused studios like Wonder report using generative AI to cut specific VFX or production costs by multiples, particularly for limited sequences or short-form content. (venturebeat.com)

So the directional thesis—that AI will increase leverage and lower some production costs—is being borne out in early ways, but the specific magnitude (100× more films at 1/100 cost) is very far from current reality.

However, because Friedberg did not specify a time frame, this was effectively a long‑run scenario rather than a claim about what would be true by 2025. Only a bit more than two years have passed since the prediction, and AI video capabilities are still evolving rapidly. Given that:

• The literal numbers he gave are not yet realized, but
• He did not say by when they should be realized,

there isn’t enough basis to call the prediction definitively “right” or “wrong” at this point. It remains a speculative long‑term forecast whose truth value can’t yet be determined.

Therefore the appropriate rating is “inconclusive (too early to tell).”

Sacks’s statement is explicitly conditional: if the South Korean LK‑99 claim were real (valid and replicable), then it would end up being regarded as the most important discovery in physics of the 21st century.

Follow‑up work in 2023–2024 showed that LK‑99 is not a room‑temperature, ambient‑pressure superconductor. Multiple experimental and theoretical studies found it to be an insulator or, at best, a non‑superconducting correlated material, and replication attempts failed to show zero resistance or a Meissner effect.(techcrunch.com) Reviews in Nature and other outlets now treat the original claim as refuted.(researchgate.net) As of late 2025, there is still no accepted room‑temperature, ambient‑pressure superconductor.(theguardian.com)

Because the condition (“their result is real and replicable”) turned out to be false, the world in which Sacks’s forecast would be tested never occurred. We therefore cannot say whether, had LK‑99 been genuine, it actually would have become regarded as the most important physics discovery of the century. The prediction’s truth value cannot be determined from real‑world outcomes, so it is best classified as ambiguous rather than right or wrong.

Evidence strongly suggests Chamath was referring to Mitra Future Technologies (Mitra Chem), an AI‑enabled battery materials startup that Social Capital led the Series A for and where Chamath joined the board. The company focuses on iron‑based cathode materials such as LFP for EVs and energy storage. (prnewswire.com)

Looking at Mitra Chem’s public announcements around the prediction window:

• Sept 16/19, 2022 (before the podcast): Mitra Chem announced shipment of U.S.-made LFP cathode samples whose performance "exceed[s]" industry‑standard Chinese LFP, achieved in four months using proprietary ML algorithms. This is framed as a technical breakthrough, but it occurred about a year before July 27, 2023, so it cannot satisfy a prediction about an announcement in the following few weeks. (prnewswire.com)
• July 12, 2023 (before the podcast): The World Materials Forum highlighted Mitra Chem’s “breakthroughs in next generation cathode material” (including rapid LFP and LMFP development), but again this recognition predates the July 27 episode. (batteriesnews.com)
• Aug 16–17, 2023 (~3 weeks after the episode): Mitra Chem announced a $40M first close of a $60M Series B round led by GM. The release and related coverage describe the company’s AI‑enabled "groundbreaking" technology and its plan to develop advanced iron‑based cathodes (e.g., LMFP) for affordable EV batteries, but they do not disclose a new, specific LFP performance result or clearly labeled scientific/engineering “breakthrough” achieved at that time; the focus is financing and partnership, not a newly achieved technical milestone. (mitrachem.com)
• Sept 13–14, 2023 (~7 weeks after the episode): Another press release announced new investment and a lithium‑procurement partnership with TechMet‑Mercuria, again framed as supporting Mitra Chem’s mission and "groundbreaking work" but with no concrete, new LFP improvement claim. (mitrachem.com)

Mitra Chem’s own site and news archive show that its post‑July‑27, 2023 announcements through September 2023 are about funding rounds and partnerships, not about a newly achieved, publicly quantified LFP or LMFP performance advance. The only clearly documented LFP performance "breakthrough" (materials outperforming Chinese state‑of‑the‑art) is from September 2022, long before the prediction date. (mitrachem.com)

Given this, there is no public evidence that Chamath’s materials/battery company announced a new major LFP‑related breakthrough within a few weeks after July 27, 2023. While a large GM‑led Series B was indeed announced in that timeframe, it does not clearly match the normalized prediction of a breakthrough announcement about improved LFP performance. Under that strict reading, the prediction did not come true.

Sacks predicted that following the July 2023 LK‑99 preprints, other research groups would attempt to replicate the reported room‑temperature, ambient‑pressure superconductivity by reproducing the material and measurements.

This clearly occurred:

• A widely cited August 2023 paper "Synthesis and characterisation of LK-99" explicitly notes that the original LK‑99 preprints "str[uck] worldwide experimental research efforts in replicating the results" and then describes an independent synthesis and characterization that found no superconductivity. (arxiv.org)
• Another group published "On the synthesis methodologies to prepare Pb₉Cu(PO₄)₆O" detailing their "various attempts to prepare" LK‑99 using multiple synthesis routes, again as a replication effort. (arxiv.org)
• A separate preprint "Absence of superconductivity in LK-99 at ambient conditions" reports continued efforts "to synthesize phase pure LK-99" following the procedure of Lee et al., explicitly as a replication and test of the superconductivity claim. (arxiv.org)
• Summary articles and timelines note that, after the July 2023 preprints, many different researchers around the world attempted to reproduce the synthesis and measurements of LK‑99 within weeks, with more than 15 notable labs publishing (mostly negative) replication results. (en.wikipedia.org)

Sacks did not predict that the replication attempts would succeed, only that “it will be replicated. People will try and do what they are now claiming they did.” That prediction—about the strong, rapid global effort to reproduce LK‑99—is unambiguously borne out by the record.