Hormuz and the Weighting Game • Part 2: The Weights

A Race, Not a Gift

DeepSeek didn't copy the American playbook. It forced the American playbook to exist. What the race it started actually subsidizes, and who cashes the subsidy fastest, is a different, harder question than who gave the first present.

Autumn 2026 Part 2 of 6: the horizontal axis (talent, weights, routing)

On January 27, 2025, Nvidia lost $590 billion of market value in a single trading session — the largest one-day loss for any company in the history of public markets — because a Chinese lab most of Wall Street had never heard of a week earlier had published, for free, the weights to a model that looked like it belonged in the same conversation as America's best. The popular telling of what happened next runs backward. It says American open-source culture (Meta's Llama, the old Silicon Valley instinct to give the tools away and monetize the platform) was already a gift to the world, and China simply took it. The filings, the timestamps, and the people who were in the room that week tell a cleaner and less flattering story: the gift-giving framing has the order of operations exactly backward. DeepSeek didn't accept an invitation. It sent one.

Where this sits

Part 2 of Hormuz and the Weighting Game. Part 1 followed a researcher's return ticket to China and asked what tacit knowledge travels with a person that no export-control regime can seize. This part follows the thing that travels even faster than a person (a checkpoint, published once and downloaded forever) and asks who actually wins the race that publishing starts. The vertical axis (chips, power, equity) belongs to The Stack, cross-linked, not re-argued. This is the horizontal axis: who holds the weights, and on whose electricity bill they get retrained.

/ 01The Shock, By the Numbers

Start with the timeline, because the sequence is the entire argument. On December 26–27, 2024, DeepSeek released V3, an open-weight base and chat model, claiming a $5.6 million cost for its final training run: a figure that describes one run, not the lab's total infrastructure spend, and one that SemiAnalysis would later estimate against roughly $1.6 billion in total server capex.1 That distinction (final run versus total build-out) is the actual mechanism of what happened three and a half weeks later, because it is precisely the distinction the market did not make.

EventDetailDate
DeepSeek-V3 releasedOpen-weight base/chat model; claimed $5.6M final-run cost, contested (SemiAnalysis: ~$1.6B total capex)Dec 26–27, 2024
DeepSeek-R1 releasedReasoning model rivaling OpenAI's o1, MIT license, technical paper disclosing method — not a black-box breakthroughJan 20–21, 2025
Nvidia's single-day loss−17% share price, −$590B market cap — largest single-day loss for any company, everJan 27, 2025
OpenAI's first open weights since 2019gpt-oss (120B/20B, Apache 2.0); coverage explicit it was “driven by competitive pressure from models such as Llama and DeepSeek”Aug 5, 2025
PRC models' share of global AI workloadsPer the House Select Committee on the CCP's own framing, in its letters opening a joint investigation into US firms' use of Chinese models~30% by end of 2025
Compiled from DeepSeek's own changelog, contemporaneous market data (Windows Central, Forbes, TradingView), SemiAnalysis, and the House Select Committee on the CCP / House Homeland Security Committee's joint press release on the Airbnb/Anysphere investigation.128

Two reactions from that week are worth quoting directly, because they're the cleanest evidence of how the shock actually registered among the people whose job is to price it. Marc Andreessen, on the day of the crash: “DeepSeek R1 is AI's Sputnik moment.”3 Ben Thompson, writing at Stratechery the same week, sharpened the analogy in the direction that matters for this series: “this is like Sputnik if the Russians had made it to space and then explained in great detail how they got there.”1 The open publication (the technical paper, the MIT license, the fact that anyone could read exactly how DeepSeek did it) is what turned a single competitive model into a commoditization event. A closed frontier model that merely matched GPT-4-class performance would have been alarming. An open one, with the method disclosed, was structurally destabilizing, because it did not just prove the capability existed. It handed every other lab, American and Chinese, a working recipe for how to get there cheaper.

/ 02Sequencing the Blame Correctly

Here is the causal correction this part exists to make, and it matters because the wrong sequence is the more comfortable one to tell. It is comfortable to say American open-source culture (Zuckerberg's stated bet that Meta should “own the ecosystem instead of the access control”4) was already running before DeepSeek showed up, and that China's labs simply rode the wave America built. Llama's download numbers support part of that story: over a billion downloads by March 2025, 1.2 billion by April, 65,000-plus community fine-tunes.4 But the sequence of what happened in the seven months after the DeepSeek shock is not a story about an American gift being accepted. It is a story about an American retreat being forced.

1

The shock (Jan 20–27, 2025)

DeepSeek-R1 ships open-weight with a disclosed method. Nvidia loses $590B in a day on the theory that frontier-comparable performance no longer requires frontier-scale compute spend.

2

The admission (early Feb 2025)

Sam Altman, on Reddit: “I personally think we have been on the wrong side of history here and need to figure out a different open-source strategy.” OpenAI's own CEO concedes the closed-weight posture is losing, in public, within two weeks of the shock.5

3

The war rooms (same week)

Zuckerberg convenes four internal engineering “war rooms” to reverse-engineer DeepSeek's efficiency gains: two on training-cost technique, two on data composition. Meta's own infrastructure director reportedly warns colleagues DeepSeek could outperform not the current Llama, but the next one.6

4

The policy response (July 2025)

America's AI Action Plan reverses the prior administration's open-model caution and directs the government to actively promote American open-weight models as a geopolitical tool: explicitly aiming to make US models “the gold standard… worldwide” so that “allies are building on American technology.”7

5

The concession, made flesh (Aug 5, 2025)

OpenAI ships gpt-oss, its first open weights since GPT-2 in 2019. Six years of closed-weight strategy ended in six months, with contemporaneous coverage naming the exact competitive pressure that forced it.1

6

The retreat (Apr 2026)

Meta, the standard-bearer of American open weights, reverses course entirely: its Superintelligence Labs ships Muse Spark, Meta's first proprietary closed-weight flagship, breaking from the Llama lineage. The reported reason: Llama 4 Maverick had been overtaken on benchmarks by Chinese open models. The lab that founded the American open-weight lane exited it because China now sets the pace inside it.19

Read that sequence in order and the ideology-as-origin story doesn't survive it. An administration does not write a national AI strategy in reaction to its own country's prior generosity. A CEO does not concede publicly, on Reddit, that his company was “on the wrong side of history” because his own strategy was working too well. Every American open-weight move that followed January 27, 2025 (the Action Plan, gpt-oss, the doubled-down Llama roadmap) reads correctly as a participant's response to a race someone else started, not as the race's opening move. DeepSeek didn't inherit the American open-source ethic. It ambushed a market that thought it still controlled the pace of its own commoditization, and the pace question is the one that actually decides who benefits.

/ 03Who Absorbs Whom

Once the race is correctly sequenced, the next question is who wins it. And “wins” here does not mean whose model tops a given month's leaderboard. It means who converts an absorbed capability, of any origin, into a shipped product fastest and cheapest. On that question the record is asymmetric, but not in the simple “China copies, America invents” shape the original framing risked. It is asymmetric in velocity.

The clearest single case: in November 2024, Reuters reported that six researchers across three Chinese institutions (two under the PLA's Academy of Military Science) had built “ChatBIT,” a military intelligence and decision-support chatbot, on top of Meta's Llama 2 13B. Meta confirmed the use was unauthorized and called the underlying model “outdated.” That's a fair defense of the specific version, and also beside the actual point: an open American model, released under a permissive license Meta could not enforce at the point of use, became infrastructure for Chinese military AI with no mechanism for Meta to detect or prevent it.9 Running the same dynamic in the other technical direction: on January 30, 2025, OpenAI told reporters it had evidence DeepSeek-linked accounts used “obfuscated third-party routers” to query its models at scale and harvest outputs for training, a direct violation of terms of service that explicitly bar training a competing model on OpenAI's outputs. OpenAI's own language accused DeepSeek's effort of “free-riding on the capabilities developed by OpenAI and other US frontier labs”10, a genuinely notable irony from a company that itself trained on a scale of scraped web data it never fully disclosed, now the aggrieved party alleging IP theft in reverse. Both directions of absorption are real. Neither is symmetric in scale. And the extraction channel did not stay a one-week news story: a year later both major US labs escalated it into quantified federal complaints. In February 2026, OpenAI formalized its accusation in a memo to the House Select Committee on the CCP, and Anthropic filed its own technical disclosure naming DeepSeek, Moonshot AI, and MiniMax, counting roughly 24,000 fraudulent accounts and more than 16 million exchanges used to query Claude for apparent distillation purposes. Part 4 carries the full treatment of what those harvested exchanges contain; what matters here is the scale, which is no longer alleged in the abstract but counted.20

FlowFigure
Qwen-based derivative models, Hugging Face151,448 — 2.6× Meta-derived models, 4.7× Llama-family specifically
Chinese-origin models' share of top new-model downloads, 2025>45% (Qwen + DeepSeek + GLM + Kimi combined)
Chinese model inference pricing (MiniMax, Zhipu)~$0.30 per million input tokens, vs. ~$5 for Claude Opus — roughly 16.7×
US commercial adopters of Chinese open weights, namedMicrosoft (Azure/GitHub, R1), Amazon (Bedrock, R1), Cursor/Anysphere (Kimi), Airbnb (Qwen — CEO's stated reason: “fast and cheap”)
Hugging Face's own 2025 year-end report and Alibaba's own claims (noted discrepancy: Alibaba cites 3B+ cumulative downloads against Hugging Face's ~2.05B count); named-adopter disclosures via company statements and reporting; token-pricing comparison sourced to a single industry roundup — directionally solid, treat exact multiple as indicative pending re-verification.1112

The congressional response is itself evidence of how fast and how deep this went. In spring 2026, the House Select Committee on the CCP and the House Homeland Security Committee opened a joint investigation into Airbnb and Anysphere specifically over their use of DeepSeek, Alibaba, Moonshot AI, and MiniMax models; the committees' own framing, not an outside estimate, put PRC models' share of global AI workloads at roughly 30% by the end of 2025.8 Weights that leak out of the United States mostly travel through unenforceable licenses and alleged terms-of-service violations: leakage the originating lab cannot fully see or stop. Weights that flow into the United States travel through completely legal, fully disclosed commercial adoption, because they are released under more permissive terms and, increasingly, because they are simply cheaper to run. Neither path required anyone to steal anything. The second path is just faster, and it is faster because of what section 06 gets to.

/ 04The Honest Steelman, and Where It Runs Out

A fair accounting of America's open-source strategy has to grant its architects a real argument, made in good faith, by people with no reason to be wrong about their own logic. David Sacks, the sitting White House AI and crypto czar, has made the ecosystem-lock-in case as plainly as anyone: “A meaningful segment of the global market will prefer the cost, customizability, and control that open source offers. We want the U.S. to win this category too”; and, reaching for the analogy that has anchored Silicon Valley's self-understanding for three decades, “Permissionless innovation is how America won the internet.”13 America's AI Action Plan codifies exactly this logic as official policy, not just industry rhetoric: give the world American open weights, and the world builds its habits, its tooling, and its standards on an American substrate, regardless of which country's model tops a given month's benchmark. It is not a naive argument. It mirrors how Android and Linux won platform share by losing the license fee, and there is a real, defensible case that Llama's early lead in downloads and derivatives proves the logic wasn't fantasy: American open models really did seed the current global open-weight norm, and the distribution infrastructure underneath nearly all of it (Hugging Face, the safetensors format, the transformers library) is still, quietly, American-built plumbing that Chinese labs ship their own weights through.

Steelman: the ecosystem bet wasn't fantasy

Take Sacks's argument on its own terms and it is genuinely strong: mindshare compounds even when a rival tops this month's leaderboard, and a developer who has built five projects on Llama's tooling conventions doesn't necessarily migrate to Qwen's the day a new benchmark drops. If the payoff to openness is measured in multi-year ecosystem control rather than this quarter's win, the strategy could still be correct even while losing on points right now. That distinction (strategy versus current scoreline) is the honest form of the steelman, and it deserves to be stated before it's set aside.

Here is where the honest version of the steelman has to confront its own author. In July 2026, after Moonshot AI's Kimi K3 (a 2.8-trillion-parameter open-weight model) took the top spot on the Frontend Code Arena and scored at or near the frontier on multiple other benchmarks, Sacks himself called the result “concerning,” warned that defensive American policy reactions (data-center permitting bans, pre-approval requirements for model releases) meant the US was “tying itself in knots,” and said plainly: “this is how you lose the AI race.”14 This is not a critic's assessment. It is the architect of the ecosystem-lock-in steelman, eighteen months after making the case, conceding on the record that the strategy is currently losing on the merits, while still defending permissionless innovation as the right approach, just one being executed too cautiously and too late. The steelman survives as a claim about what the right strategy was. It does not survive as a claim about what is happening in 2026. Reported download and adoption numbers (Qwen ahead of Llama on Hugging Face, Chinese models above 45% of top new-model downloads, roughly 30% of global AI workload share) back Sacks's own July assessment more than they back his February one.

/ 05A Move, Not a Present

It helps here to name what an open-weight release actually is, in the vocabulary international-relations scholars use for exactly this kind of act, rather than treat it as either charity or theft. Thomas Schelling's account of strategic commitment supplies the sharper frame. An actor becomes more credible, not less, when it deliberately gives up the option to reverse course: burning a bridge, tying its own hands, manufacturing an irreversibility that a purely discretionary actor could never signal.15 Publishing a model's weights is precisely this kind of move. Once a checkpoint is public, it cannot be unpublished; no future policy shift, boardroom decision, or export-control tightening can claw it back. DeepSeek's R1 release, and every American open-weight release that followed it, functions less like a gift with strings attached and more like a commitment device: each lab signaling, to rivals and to its own government simultaneously, that it is in this race and cannot credibly threaten to leave it. That single-shot irreversibility sits inside a much longer game, though, and that's where Robert Axelrod's work on the iterated Prisoner's Dilemma becomes the more useful lens for the race as a whole. Axelrod's central finding (that cooperation becomes rational in a repeated game because of the “shadow of the future,” the durable expectation of retaliation or reward) only holds if that shadow is long and credible on both sides.16 China's five-year-plan continuity gives its labs and regulators a planning horizon that doesn't reset with an election. American AI policy, by contrast, has visibly reversed at each administration change over the past decade, which shortens the credible time horizon any single US actor can commit to, and, by Axelrod's own logic, should make the US side structurally less able to sustain a patient, iterated strategy and more prone to reactive, single-cycle moves. That is not an accusation of bad faith. It is a structural reading of why the same open-weight instinct that might have paid off over twenty years of steady execution instead produced a scramble.

This is also the honest answer to the more comfortable rival hypothesis: that the whole US-China AI relationship is really a symmetric prisoner's dilemma, a mutual hostage exchange dressed up in Sputnik headlines. The frameworks above don't refute that reading; they sharpen it into a truer, more specific one. Both governments are behaving rationally inside a shared, iterated game, and both feel the exchange: that much of the symmetric framing is correct. What the game-theoretic literature on this exact case (Zeng's 2025 documentation of the ChatGPT-to-DeepSeek action-reaction spiral is the clearest empirical version of it17) actually supports is not mutual advantage but mutual reaction on unequal footing: a real, mirrored security-dilemma spiral sitting on top of an asymmetric ability to convert what gets absorbed into deployed advantage. Feeling the exchange and benefiting from it are not the same thing, and the distillation numbers in section 03 are the tally for which side currently does more of the second.

/ 06The Meter Underneath

One question this part can only gesture at, because it belongs properly to Part 3: why does absorption convert to advantage faster on one side of the Pacific than the other? The short answer is electricity. American household power runs roughly $0.18 per kilowatt-hour against a Chinese household rate near $0.08. Chinese industrial rates, especially in the western provinces where much of the country's newest compute is sited near wind and solar, run lower still.18 Distillation and fine-tuning (the retraining step that turns an absorbed model, of any origin, into a shippable competitive product) is a compute-intensive, power-hungry process in its own right. Cheap, reliable domestic power doesn't just subsidize the original training run. It subsidizes the entire downstream loop of absorbing, retraining, and redeploying whatever capability shows up on either side of the export-control line, which is the actual mechanism connecting this part's argument to the energy axis Part 3 covers in full.

“The chip embargo had removed the hardware. But the hardware had already been mapped by a mind that no customs officer could detain.”

/ 07The Race, Correctly Timestamped

Put the sequence back together. DeepSeek-R1 shipped open-weight, with its method disclosed, on January 20, 2025, and by January 27 it had erased more market value in a single day than any company loss in history. Every American open-weight move that followed (Altman's public reversal, Meta's war rooms, the AI Action Plan, gpt-oss) is dated after that week, and each one is documented, in its own contemporaneous coverage, as a response to competitive pressure rather than an independent act of generosity. The steelman for America's open-source strategy is real and was argued honestly here, by its own architect, before it was set against his own later assessment that the strategy is currently losing. And the frameworks that describe why (Schelling's commitment device for what a published checkpoint actually is, Axelrod's shadow of the future for why an electorally discontinuous strategy struggles against a continuously planned one) land on the same conclusion the raw distillation numbers do: openness didn't originate in Menlo Park and travel outward as a gift. It originated in Hangzhou as a shock, and the race it forced runs on whoever's electricity bill is smaller. Part 3 follows the tokens themselves — where American inference demand actually routes, and why.

Sources & Citations (20)
[1] DeepSeek-V3 changelog and release notes (Dec 26–27, 2024): claimed $5.6M final-training-run cost, corroborated across BentoML and Analytics Vidhya coverage. DeepSeek-R1 release (Jan 20–21, 2025), MIT license, technical paper disclosing method. Ben Thompson, “Stratechery,” Jan 27, 2025: “this is like Sputnik if the Russians had made it to space and then explained in great detail how they got there.” SemiAnalysis's ~$1.6B total-server-capex counter-estimate; Demis Hassabis calling the $5.6M figure “exaggerated and a little bit misleading.” OpenAI gpt-oss release, Aug 5, 2025 (CNBC, Sherwood News, Built In, all Aug 5 2025 coverage explicit on competitive framing vs. Llama/DeepSeek).
[2] Nvidia single-day loss, Jan 27, 2025: ~15–17% share-price decline (sources vary within this range), ~$589–590B market-cap loss, widely reported as the largest single-day market-cap loss for any company in market history. Windows Central, Forbes, TradingView, Jan 27–28, 2025.
[3] Marc Andreessen (@pmarca), X/Twitter, Jan 27, 2025: “Deepseek R1 is AI's Sputnik moment.” Widely quoted same day across Fortune, Yahoo Finance, and other outlets.
[4] Llama download/derivative figures (1B by March 2025, 1.2B by April 2025, 65,000+ community fine-tunes) and Meta's own LlamaCon 2025 framing (“own the ecosystem instead of the access control”), as reported contemporaneously across coverage of the event.
[5] Sam Altman, Reddit AMA, early Feb 2025: “I personally think we have been on the wrong side of history here and need to figure out a different open-source strategy… not everyone at OpenAI shares this view, and it's also not our current highest priority.” Corroborated across VentureBeat, Futurism, TechStartups.
[6] Meta's “four war rooms” reporting, Jan 27, 2025, sourced to unnamed Meta insiders across Futurism, Fortune, Cybernews, Windows Central, Yahoo Finance; corroborated across outlets but not company-confirmed on the record. Zuckerberg's Q4 earnings-call commitment to ~$60B in 2025 AI capex and the stated goal that Llama 4 should “lead the market rather than follow it.”
[7] “Winning the Race: America's AI Action Plan,” White House, July 2025: verified against the published PDF via CSET, AEI, Brookings, and Fenwick summaries. Quoted language on making US models “the gold standard… worldwide” and ensuring “allies are building on American technology” appears in the plan itself; the plan's causal link to the DeepSeek shock specifically is inferred from timing and framing, not a line in the document itself.
[8] House Select Committee on the CCP (Chairman John Moolenaar) and House Homeland Security Committee (Chairman Andrew Garbarino) joint investigation into Airbnb and Anysphere/Cursor over use of DeepSeek, Alibaba, Moonshot AI, and MiniMax models; letters and press release dated around Apr 29, 2026, with Semafor/CNBC/Nextgov/Slashdot coverage through July 2026. PRC-models'-30%-of-global-workload figure per the committee's own framing.
[9] Reuters, Nov 2024: PLA/Academy of Military Science-affiliated researchers' “ChatBIT” built on Meta's Llama 2 13B; Meta confirmed the use as unauthorized and called the underlying model version “outdated.” Corroborated by TechCrunch, The Information, Gizmodo. Jamestown Foundation's Sunny Cheung on this as the first substantial evidence of systematic PLA research leveraging Meta's open models specifically.
[10] OpenAI's distillation accusation against DeepSeek, reported Jan 30, 2025 (SCMP, TechRadar, Malay Mail): alleged use of “obfuscated third-party routers” to query OpenAI models and harvest outputs for training, in violation of OpenAI's terms of service; OpenAI's own framing of this as DeepSeek “free-riding on the capabilities developed by OpenAI and other US frontier labs.” The accusation is well-documented; the underlying technical claim remains OpenAI's assertion, not independently adjudicated.
[11] Qwen download/derivative counts: Hugging Face's own 2025 year-end blog report (151,448 Qwen-based derivative models: 2.6× Meta-derived, 4.7× Llama-family specifically; ~2.05B cumulative 2025 downloads per Hugging Face's own count, against Alibaba's separately claimed 3B+ figure, discrepancy noted, not reconciled). Chinese-origin models' >45% share of top new-model downloads in 2025, corroborated across China Daily, TheNextWeb, AIWorld, Dealroom with some variance in exact figures.
[12] Named US commercial adopters of Chinese open models: Microsoft (Azure AI Foundry and GitHub, R1, Feb 2025), Amazon (AWS Bedrock, R1), Perplexity (R1, cost-motivated), Nvidia (R1 on its own dev platforms), Cursor/Anysphere (“Composer 2,” disclosed as built on Moonshot AI's Kimi), Airbnb (AI customer-service agent built on Alibaba's Qwen; CEO Brian Chesky's stated reason: “fast and cheap”). Token-pricing comparison (~$0.30/M input tokens for MiniMax/Zhipu vs. ~$5/M for Claude Opus, ~16.7× gap) sourced via a single industry-roundup chain (Huxiu, aggregated by a Chinese-tech-portal roundup), directionally solid, exact multiple not independently re-verified against a primary pricing index.
[13] David Sacks (White House AI & crypto czar), X posts: “A meaningful segment of the global market will prefer the cost, customizability, and control that open source offers. We want the U.S. to win this category too” and “Permissionless innovation is how America won the internet.” Corroborated by Benzinga and Axios coverage.
[14] David Sacks on Moonshot AI's Kimi K3 (2.8T-parameter open-weight model) topping the Frontend Code Arena and scoring near-frontier on multiple benchmarks: called the result “concerning,” warned the US was “tying itself in knots” with defensive policy reactions (data-center permitting bans, pre-approval requirements for model releases), said “this is how you lose the AI race.” Reported by Axios, Jul 17, 2026; direct Axios fetch returned a 403 in research for this piece, so exact quote wording is corroborated via WebSearch synthesis and three independent outlet headlines describing the same statement, not a verbatim primary-source pull; flagged for re-verification against Axios's original text or Sacks's own post before any further reuse.
[15] Thomas Schelling, The Strategy of Conflict (1960), ch. 8–9, on commitment devices and the strategic value of deliberately relinquished discretion; digitized original 1959 memo via RAND (HDA1631-1).
[16] Robert Axelrod, The Evolution of Cooperation (1984), on tit-for-tat and the “shadow of the future” as the condition for rational cooperation in an iterated Prisoner's Dilemma. The application of this framework to US electoral-cycle discontinuity versus China's five-year-plan planning continuity, specifically in the AI-competition context, is this series' own analytical synthesis of Axelrod's general theory; not a claim lifted from a published China-AI-specific application of Axelrod, which this series' research did not locate in the peer-reviewed IR literature.
[17] Jinghan Zeng, “ChatGPT as a security threat: US–China security dilemma in the generative AI race,” British Journal of Politics and International Relations (SAGE, Nov 2025), DOI 10.1177/13691481251389415: documents the 2022–25 action-reaction spiral: ChatGPT's success triggering Chinese regulatory and capability response, DeepSeek's breakthrough triggering a symmetric spike in US strategic insecurity and tighter controls.
[18] US household electricity ~$0.18/kWh (Mar 2024) vs. Chinese household ~$0.08/kWh (The Diplomat, Feb 2025); Chinese industrial rates cited lower still, especially in western provinces (0.15–0.28 yuan/kWh, roughly $0.02–0.04), sourced via a single Chinese-tech-portal roundup (citing Huxiu and CAICT analyst Shi Yuxia), directionally corroborated but not independently verified against a primary Chinese government or IEA source. Detailed grid-capacity and pricing figures (PJM, US data-center draw) belong to Part 3 and Michael Cembalest's verified J.P. Morgan figures, reused there rather than duplicated here.
[19] VentureBeat, "Goodbye, Llama? Meta launches new proprietary AI model Muse Spark," April 2026, with corroborating coverage of Meta Superintelligence Labs' formation (summer 2025, under Alexandr Wang) and the reported rationale that Llama 4 Maverick had been overtaken on benchmarks by Chinese open models (GLM-5, Qwen 3.6 Plus named in reporting). Documented at fuller length in this platform's sibling series, The Stack, Part 4 ("The Frontier Models"), which also carries the June 2026 benchmark data on how narrow the frontier band had become: the best Chinese open-weight models within a few dozen Elo points of closed frontier models at 10 to 50 times lower per-token cost, per Artificial Analysis and OpenRouter data cited by J.P. Morgan.
[20] OpenAI memo to the U.S. House Select Committee on the CCP, Feb 12, 2026 (Rest of World; Foundation for Defense of Democracies, Feb 13, 2026); CNBC, "Anthropic accuses DeepSeek, Moonshot and MiniMax of distillation attacks on Claude," Feb 24, 2026 (~24,000 fraudulent accounts, 16M+ exchanges per Anthropic's own technical disclosure). The first time both major US labs made quantified, named, public distillation allegations against specific Chinese competitors inside the same month. Part 4 of this series treats the same events from the data angle; The Stack, Part 4, treats them from the frontier-lab-economics angle.
A note on method. The anchor sequence here (DeepSeek's release dates, Nvidia's crash, Altman's public reversal, gpt-oss) is high-confidence and independently corroborated across multiple outlets. Figures that remain estimate, single-sourced, or unreconciled (the exact token-pricing multiple, the Chinese industrial electricity range, the precise wording of Sacks's July 2026 remarks) are flagged in the footnotes above rather than smoothed into certainty. The Axelrod application to US electoral discontinuity is original analysis grounded in a real published framework, not a citation of a paper that doesn't exist; it is labeled as such rather than dressed up as settled scholarship.