PCAST 2026: Cross-Provider Analysis
Trump's Science Advisory Council — A Definitive Research Synthesis
Executive Summary
-
This is not a science council — it's a CEO council with a government advisory mandate. Across all 7 providers, there is near-unanimous consensus that the 2026 PCAST represents an unprecedented departure from historical norms: roughly 90%+ of members are C-suite executives, venture capitalists, or billionaires, compared to ~60-70% PhD academics on Obama-era and Biden-era panels [3]. Only 4-5 of 15 total members (including co-chairs) hold research doctorates.
-
The combined net worth of this council is approximately $930 billion to $1.1 trillion, making it the wealthiest advisory body in U.S. history. The top five members alone (Ellison ~$247-258B, Zuckerberg ~$225-248B, Brin ~$140-230B, Huang ~$160-164B, Dell ~$121-140B) account for the vast majority of that figure — exceeding the GDP of countries like the Netherlands, Sweden, and Poland [6].
-
At least 11 of 15 members have active, material business before the federal government — including antitrust cases (Meta, Google), export control battles (Nvidia, AMD), federal cloud contracts (Oracle, Dell), NRC licensing (Oklo), DOE fusion funding (Commonwealth Fusion Systems), and SEC/CFTC crypto regulation (Coinbase/Paradigm). The conflict-of-interest density is without precedent in PCAST history [5].
-
The political lean is approximately 70-80% Republican in recent cycles, driven by mega-donors Andreessen, Ellison, Sacks, Dell, and Ehrsam — several of whom were historically Democratic or politically neutral before 2022-2024 realignments [4]. Sergey Brin and Mark Zuckerberg remain the most Democratic-leaning members, though both have made recent gestures toward the Trump administration.
-
The exclusion of Musk and Altman is structurally significant: it signals the administration prefers commercially loyal, institutionally embedded Big Tech over independent, potentially disruptive AI voices — and may reflect specific interpersonal and competitive dynamics that will shape the council's AI policy outputs [3].
Cross-Provider Consensus
The following findings were independently confirmed by multiple providers and represent the highest-confidence conclusions of this analysis.
FINDING 1: The 2026 PCAST is overwhelmingly composed of industry executives, not scientists.
- Providers: Gemini, OpenAI, OpenAI-Mini, Grok, Grok-Premium, Gemini-Lite, Perplexity (all 7)
- Confidence: HIGH
- Sources: [5]- Every provider independently characterized this council as a "CEO Council" rather than a science advisory body. The Register headline "Trump staffs science and technology panel with non-scientists" [60] and Ars Technica's coverage [60] are cited by multiple providers. This is the single most robustly confirmed finding in the dataset.
FINDING 2: Only 2 of 15 members are women (Safra Catz and Lisa Su).
- Providers: Gemini, OpenAI, OpenAI-Mini, Grok, Grok-Premium, Perplexity (6 of 7)
- Confidence: HIGH
- Sources: [4]- All providers agree on the exact count and the identity of the two women. The council is 87% male.
FINDING 3: The council has zero Black, Hispanic, or Indigenous members.
- Providers: Gemini, OpenAI, OpenAI-Mini, Grok, Perplexity (5 of 7)
- Confidence: HIGH
- Sources: [4]- Racial composition is described as overwhelmingly white and Asian-American, with no representation from Black, Latino, or Indigenous communities.
FINDING 4: The combined net worth exceeds $900 billion and likely approaches or exceeds $1 trillion.
- Providers: Gemini, OpenAI, OpenAI-Mini, Grok, Grok-Premium, Gemini-Lite (6 of 7)
- Confidence: HIGH (with variance in exact figures — see Contradictions section)
- Sources: [8]- All providers confirm the figure is historically unprecedented for an advisory body. Perplexity's lower estimate ($600-800B) is the outlier, likely reflecting older data.
FINDING 5: The aggregate political lean is approximately 70-80% Republican in recent cycles.
- Providers: OpenAI, OpenAI-Mini, Grok, Grok-Premium, Perplexity (5 of 7)
- Confidence: HIGH
- Sources: [5]- Multiple providers independently converge on this range. The council is not bipartisan in any meaningful sense, though it includes some historically Democratic donors (Brin, Zuckerberg) who have recently moderated or shifted.
FINDING 6: Several members have recently "flipped" from Democratic to Republican giving.
- Providers: Gemini, OpenAI, Grok, Grok-Premium (4 of 7)
- Confidence: HIGH
- Sources: [4]- Marc Andreessen, David Sacks, and to a lesser extent Mark Zuckerberg and Michael Dell are identified as having made notable rightward shifts between 2022 and 2024.
FINDING 7: At least 10-11 members have active business before the federal government.
- Providers: OpenAI, Grok, Grok-Premium, Perplexity, Gemini-Lite (5 of 7)
- Confidence: HIGH
- Sources: [6]- The specific conflicts named consistently across providers: Oracle (federal cloud contracts, TikTok), Meta (FTC antitrust), Google (DOJ antitrust), Nvidia (export controls), AMD (export controls), Dell (federal hardware contracts), Coinbase (SEC/CFTC), Oklo (NRC licensing), Commonwealth Fusion (DOE funding).
FINDING 8: John Martinis, Lisa Su, Jacob DeWitte, and Bob Mumgaard hold research PhDs.
- Providers: Gemini, OpenAI, OpenAI-Mini, Grok, Grok-Premium, Perplexity (6 of 7)
- Confidence: HIGH
- Sources: [5]- Martinis (Physics, UCSB), Su (EE, MIT), DeWitte (Nuclear Engineering, MIT), Mumgaard (Applied Plasma Physics, MIT). Sergey Brin has an MS but not a PhD. This gives the council 4 confirmed research doctorates out of 15 members.
FINDING 9: The council lacks cybersecurity specialists, AI ethicists, climate scientists, biotech experts, and academic AI researchers.
- Providers: OpenAI, OpenAI-Mini, Grok, Grok-Premium, Perplexity, Gemini-Lite (6 of 7)
- Confidence: HIGH
- Sources: [3]- This is the most consistently cited structural blind spot across all providers.
FINDING 10: Musk and Altman are notably absent, and their exclusion is deliberate.
- Providers: Gemini, OpenAI, Grok, Grok-Premium, Perplexity, Gemini-Lite (6 of 7)
- Confidence: HIGH
- Sources: [3]- All providers treat the exclusion as intentional and analytically significant, though they differ on the precise reasons (see Contradictions section).
FINDING 11: The council's most likely policy outputs are AI deregulation, crypto-friendly frameworks, CHIPS Act expansion, and nuclear/fusion fast-tracking.
- Providers: OpenAI, OpenAI-Mini, Grok, Grok-Premium, Perplexity (5 of 7)
- Confidence: HIGH
- Sources: [3]- Consistent across providers regardless of their framing of the council's legitimacy.
FINDING 12: The biggest structural risk is regulatory capture.
- Providers: OpenAI-Mini, Grok, Grok-Premium, Gemini-Lite, Perplexity (5 of 7)
- Confidence: HIGH
- Sources: [3]- The council advises on the exact regulatory domains where its members have the most to gain financially.
Unique Insights by Provider
Gemini
- David Sacks's specific early Democratic giving history: Gemini is the only provider to cite Sacks's $58,000+ donation to Gavin Newsom in 2017 and his 2016 Clinton/DNC contributions [4], providing the most granular documentation of his political flip. This matters because it establishes the magnitude of his ideological journey — from California Democrat megadonor to Trump's AI czar in under a decade.
- Michael Dell's "Trump Accounts" commitment: Gemini uniquely highlights Dell and his wife Susan's $6.25 billion philanthropic commitment to seed Trump Accounts in December 2025 [2] — a figure that, while not a direct political donation, represents an extraordinary alignment of personal wealth with a signature Trump policy initiative.
OpenAI
- Marc Andreessen's board seat at Meta as a specific conflict: OpenAI uniquely emphasizes that Andreessen has been a Meta board member since 2008 [2], meaning he simultaneously advises the government on AI policy while sitting on the board of a company facing FTC antitrust action. This is the most concrete single-person conflict of interest identified in the dataset.
- The "All-In Podcast" network as a pre-existing social infrastructure: OpenAI notes that Sacks and Friedberg co-host the All-In podcast [13], which functions as an informal ideological coordination mechanism among council-adjacent figures. This social network predates and likely shapes the council's intellectual agenda.
OpenAI-Mini
- Specific FEC-traceable donation data for Lisa Su: OpenAI-Mini is the only provider to cite a specific, FEC-traceable donation for Lisa Su — $2,800 to Democratic Rep. Anna Eshoo in 2020 [40]. This is the most granular data point for a member otherwise described as politically neutral.
- Larry Ellison's election-denier funding via Opportunity Matters Fund: OpenAI-Mini provides the most detailed account of Ellison's 2022 giving, noting his $20 million accounted for 78% of the Opportunity Matters Fund's total receipts [37], and that this PAC specifically backed election-denier Senate candidates — a more politically charged characterization than other providers offer.
Grok
- Andreessen Horowitz's investment in both Coinbase and Oklo: Grok uniquely maps the a16z investment portfolio to specific council members [59], noting that Andreessen's firm has invested in Coinbase (Ehrsam's company) and Oklo (DeWitte's company) — creating a financial web where the council's VC representative has equity stakes in two other council members' companies. This is the most specific business entanglement identified.
- The council has 9 seats still vacant: Grok-Premium notes that the announced 13 members represent an incomplete council — a typical PCAST has ~24 members, meaning 9 seats remain unfilled [54]. This is analytically important: the council's composition could shift significantly, and the vacant seats represent potential future political appointments.
- Specific net worth figures with sourcing: Grok provides the most granular individual net worth estimates with source attribution, including Sacks at $1.5B [2], Kratsios at $500M, DeWitte at $500M [44], and Martinis at $50M [15] — figures not provided by other providers.
Grok-Premium
- Oracle's TikTok involvement as a specific conflict: Grok-Premium most explicitly connects Oracle's past TikTok bid [2] to the council's potential influence over tech platform regulation, noting Oracle invested millions in government influence before winning a major stake in TikTok [74]. This is the most specific example of a council member's company having benefited from prior White House access.
- Marc Andreessen's $2.5M+ to "Right for America" PAC: Grok-Premium provides the most specific breakdown of Andreessen's 2024 giving [2], citing the "Right for America" PAC specifically — a level of granularity not matched by other providers.
Gemini-Lite
- The council functions "less like a scholarly advisory panel and more like an industry trade lobby": While other providers make similar observations, Gemini-Lite is the most explicit in characterizing the structural function of the council as analogous to a trade association rather than an advisory body — a framing with significant implications for how its recommendations should be weighted by policymakers and the public.
- Trump's first-term PCAST maintained ~50% academic scientists: Gemini-Lite provides the most specific comparison to Trump's first-term PCAST, noting it maintained roughly 50% academic presence [62] — meaning the 2026 council represents a dramatic departure even from Trump's own prior standard, not just from Obama/Biden norms.
Perplexity
- Explicit acknowledgment of knowledge cutoff limitations: Perplexity is the only provider to explicitly flag that its knowledge cutoff (April 2024) limits the accuracy of its 2025-2026 data — a methodologically honest disclosure that should inform how readers weight its specific figures versus providers with more current data access.
- The most comprehensive "missing voices" list: Perplexity provides the most systematic enumeration of absent expertise, naming specific individuals: Dario Amodei, Stuart Russell, Fei-Fei Li, Andrew Ng, Jennifer Doudna — and specific institutional gaps: DARPA leadership, Linux/Apache open-source community, labor unions. This is the most actionable gap analysis in the dataset.
Contradictions and Disagreements
Contradiction 1: Combined Net Worth — Wide Variance Across Providers
| Provider | Estimate |
|---|---|
| OpenAI | $1.0–1.1 trillion |
| Grok | ~$930 billion |
| Grok-Premium | $900B+ (top 5 alone) |
| Gemini | Exceeds $1 trillion |
| Perplexity | $600–800 billion |
| Gemini-Lite | "Hundreds of billions" |
Analysis: The variance is primarily driven by different data vintages. Perplexity's lower estimate likely reflects pre-AI-boom figures (pre-April 2024 cutoff). The most current providers (Grok, OpenAI, Grok-Premium) converge on $930B–$1.1T. The discrepancy for individual members is most pronounced for Sergey Brin (Grok: $140B vs. OpenAI: $230B) and Jensen Huang (Grok: $164B vs. Perplexity: $60-65B), reflecting the extraordinary volatility of AI-era wealth. Do not treat any single figure as definitive; use the $930B–$1.1T range as the working estimate.
Contradiction 2: Number of PhD Holders — 2 vs. 4 vs. 5
| Provider | PhD Count |
|---|---|
| OpenAI-Mini | 2 of 15 |
| Perplexity | 2-3 of 15 |
| OpenAI | 4 of 15 |
| Grok | 5 of 15 |
| Grok-Premium | ~4 of 15 |
Analysis: The discrepancy stems from different treatment of Sergey Brin (MS, not PhD — most providers correctly exclude him) and whether co-chairs are counted. The defensible count is 4 confirmed research PhDs among the 13 members: Martinis (Physics), Su (EE), DeWitte (Nuclear Engineering), Mumgaard (Plasma Physics). Brin has an MS. Jensen Huang has an MS. The co-chairs (Sacks, Kratsios) have no research doctorates. OpenAI-Mini's count of 2 appears to exclude DeWitte and Mumgaard, possibly due to their lower public profile. The correct answer is 4 PhDs among 13 members, or 27%.
Contradiction 3: Sergey Brin's Political Lean — 80% D vs. 90% D, and Recent Shift
| Provider | Estimate | Notes |
|---|---|---|
| OpenAI | ~90% Democrat | Historical giving |
| Grok | 80% D / 20% R | Notes "minimal recent R donations" |
| OpenAI-Mini | "Predominantly Democratic-leaning" | Notes no Trump booster activity |
| Gemini | 90% Democrat | Notes $100K RNC donation in 2025 |
| Perplexity | "Historically Democratic, no recent major donations" |
Analysis: Gemini uniquely cites a $100,000 RNC donation in 2025 [10] and a $20 million donation to "Building a Better California" in early 2026 [2] — the latter being a Democratic-leaning cause. If the RNC donation is accurate, Brin represents a genuine recent shift, though at a much smaller scale than his historical Democratic giving. This specific claim (RNC donation) should be independently verified via FEC records before being treated as confirmed.
Contradiction 4: Mark Zuckerberg's Political Lean — 80% D vs. 85% D vs. "Centrist"
| Provider | Estimate |
|---|---|
| OpenAI-Mini | ~80% D / 20% R (best guess) |
| Grok | 85% D / 15% R |
| Perplexity | "Historically centrist/bipartisan, increased Republican engagement post-2024" |
| OpenAI | "Avoided direct candidate donations; major giving was philanthropic" |
Analysis: The disagreement reflects genuine ambiguity in Zuckerberg's giving history. His $400 million in 2020 election infrastructure donations [2] were formally nonpartisan but widely perceived as benefiting Democrats. Meta's $1 million to Trump's inaugural fund [41] was corporate, not personal. Zuckerberg's personal FEC record is thin. The honest answer is that Zuckerberg's personal political lean is genuinely difficult to categorize, and the "80-85% D" estimates are inferences rather than documented giving records.
Contradiction 5: Why Were Musk and Altman Excluded?
| Provider | Primary Explanation |
|---|---|
| Grok | "Preference for loyal Big Tech over independents" |
| Grok-Premium | Musk: "too chaotic or conflicting interests"; Altman: "safety emphasis misalignment, prior Democratic ties, rivalry with council members" |
| Perplexity | Musk: "too toxic or unpredictable"; Altman: "administration favors rapid deployment over caution" |
| OpenAI | Musk/Altman feud with each other and with council members as a factor [47] |
| Gemini-Lite | Structural: administration prefers "commercially loyal, institutionally embedded Big Tech" |
Analysis: No provider has definitive sourcing for the exclusion rationale — all are inferential. The most plausible synthesis: Musk's exclusion reflects his existing formal role (DOGE) making a separate advisory seat redundant or politically complicated, plus his increasingly erratic public behavior. Altman's exclusion likely reflects his ongoing legal conflict with Musk (which would create council dysfunction), his company's competitive relationship with several council members' AI investments, and possible administration skepticism about OpenAI's safety-first positioning. Neither exclusion is definitively explained by available sources.
Contradiction 6: Marc Andreessen's Total 2024 Giving — $6M vs. $25M vs. $42M+
| Provider | Figure | Context |
|---|---|---|
| Gemini | $6M to MAGA Inc. (with Horowitz) + $25M to Fairshake PAC | Separate figures |
| Grok | $42M+ total to Trump PACs and super PACs | Aggregate |
| Grok-Premium | $2.5M+ to "Right for America" | Specific PAC |
Analysis: These figures are not necessarily contradictory — they may represent different PACs and time periods being aggregated differently. The Fairshake PAC figure ($25M from a16z) [22] is the most consistently cited and likely the most reliable. The $42M+ figure from Grok may be an aggregate across multiple vehicles. The $25M Fairshake figure has the strongest sourcing [2]; the $42M+ figure should be treated as unverified until cross-referenced with FEC records.
Detailed Synthesis
Section 1: Political Donation Profiles — Member by Member
David Sacks (Co-Chair) The most dramatic political conversion on the council. [Gemini] documents that Sacks donated to Hillary Clinton's campaign and the DNC in 2016 and gave over $58,000 to Gavin Newsom in 2017 [4] — making him, at that point, a reliable California Democratic donor. His trajectory shifted after January 6, 2021, when he initially condemned Trump, then reversed course. By 2024, he had co-hosted a $300,000-per-ticket Trump fundraiser in San Francisco [2] and become what [OpenAI-Mini] describes as Trump's self-styled "AI and crypto czar" [31]. [Grok] estimates his current giving at approximately 90% Republican [3]. His Craft Ventures fund has pro-crypto, pro-Trump alignment, and [Grok] notes associations with the Fairshake PAC ecosystem [6]. Estimated lean: 85-90% Republican (recent cycles); historically 60-70% Democratic.
Michael Kratsios (Co-Chair) [OpenAI] and [Perplexity] both describe Kratsios as a protégé of Peter Thiel who served as Trump's Chief Technology Officer in the first term [2]. His donation record is thin — [Gemini] cites $12,400 to Blake Masters across two cycles and $2,650 to Will Scharf [20], both Trump-aligned Republicans. [Grok] estimates approximately 80% Republican [3]. No major Democratic giving is documented by any provider. Estimated lean: ~80% Republican.
Marc Andreessen The most significant financial convert on the council. [Grok] notes he supported Democrats including Obama and Clinton through 2016 [2], but by 2024 had become one of the largest individual donors to Trump-aligned causes. [Grok-Premium] cites $2.5M+ to "Right for America" [2]; [Gemini] cites $6M to MAGA Inc. (jointly with Ben Horowitz) [21]; [OpenAI-Mini] cites $25M from a16z to the Fairshake PAC [22]. [Grok]'s aggregate estimate of $42M+ to Trump PACs and super PACs [2] may reflect the total across all vehicles. The Guardian reported that tech billionaires poured $394.1M into the 2024 election [32], with Andreessen among the top contributors. His board seat at Meta since 2008 [2] creates a specific, documented conflict: he advises on AI regulation while sitting on the board of a company facing FTC antitrust action. Estimated lean: 90-95% Republican (recent cycles); historically 50-60% Democratic.
Sergey Brin The most genuinely ambiguous member politically. [OpenAI] and [Grok] both describe his historical giving as approximately 80-90% Democratic [2], including support for Obama and environmental causes. [OpenAI-Mini] notes a $200K donation to a California environmental ballot measure and $1M to a clean-energy PAC [33]. [Gemini] uniquely reports a $100,000 RNC donation in 2025 and a $20M donation to "Building a Better California" in early 2026 [2] — the latter suggesting continued Democratic-cause alignment even as he makes token Republican gestures. His presence on this council is the most surprising given his giving history. [Perplexity] notes he has "no recent major donations disclosed," suggesting he may be deliberately keeping a lower political profile. Estimated lean: 80% Democratic (lifetime); recent shift toward neutrality with token Republican gestures.
Safra Catz The most consistently Republican member among the women on the council. [OpenAI-Mini] documents $125K to Trump's campaign and tens of thousands more to GOP committees in 2020 [34], with [OpenAI] noting she was on Trump's 2016 transition team [27]. [Grok] estimates $1M+ to Trump-aligned causes [3]. [OpenAI-Mini] states Forbes data confirms donations to Trump Victory, the NRCC, and RNC [11]. [Grok] estimates 70% Republican / 30% Democrat [3], while [OpenAI] estimates 75% Republican / 25% Democrat [11]. Oracle's business interests — $10B+ in government cloud contracts [12], the TikTok stake [2], and ongoing antitrust probes — make her one of the most conflicted members on the council. Estimated lean: 70-75% Republican.
Michael Dell [OpenAI-Mini] traces Dell's early-2000s FEC record to $1,000 contributions to Kay Granger and the NRCC, with one contribution to Democrat Patrick Leahy [35]. [Grok] estimates 75% Republican / 25% Democrat [5]. The most striking recent data point, cited by [Gemini], is Dell and his wife Susan's $6.25 billion philanthropic commitment to seed Trump Accounts in December 2025 [2] — not a direct political donation but an extraordinary alignment of personal wealth with a signature Trump policy. Dell Technologies has $20B+ in federal contracts [2], making his presence on a council that advises on federal technology procurement a textbook conflict of interest. Estimated lean: 75-80% Republican.
Jacob DeWitte The least politically visible member. [Grok] and [OpenAI-Mini] both note minimal public FEC data [44]. His significance on the council is substantive rather than political: as CEO of Oklo, the advanced nuclear micro-reactor company, he represents a genuine emerging technology sector. However, Oklo is actively seeking NRC licenses [44] and DOE funding [44] — both of which could be influenced by PCAST recommendations. [Grok] notes a16z has invested in Oklo [59], creating a financial link between DeWitte and Andreessen. Estimated lean: Insufficient data; likely politically neutral or minimally engaged.
Fred Ehrsam Co-founder of Coinbase and founder of Paradigm, a crypto-focused investment firm. [Grok] estimates 85% Republican / 15% Democrat [2], driven primarily by his association with the Fairshake PAC, which received $40M+ in pro-crypto/Trump support [6]. [OpenAI] estimates 70% Republican / 30% Democrat [2]. Coinbase faces ongoing SEC and CFTC regulatory battles [2], making Ehrsam's presence on a council that could influence financial technology regulation a direct conflict. Estimated lean: 70-85% Republican (recent cycles).
Larry Ellison The most unambiguously Republican megadonor on the council. [OpenAI-Mini] documents $20M to the Opportunity Matters Fund in 2022, which accounted for 78% of that PAC's total receipts and backed election-denier Senate candidates [37]. [OpenAI] notes a $5M donation to a pro-Lindsey Graham super PAC in 2020 [37] and [Gemini] cites $35M to Tim Scott's Opportunity Matters Fund [6]. [Grok] estimates 90% Republican / 10% Democrat [45]. At 81 years old [2], he is the eldest council member. Oracle's federal footprint — $10B+ in government cloud contracts, the JWCC Pentagon contract [2], and the TikTok stake [2] — makes him among the most conflicted members. Estimated lean: ~90-95% Republican.
David Friedberg Co-host of the All-In podcast with Sacks [13], founder of The Production Board (a food and agriculture tech company). [Grok] notes limited public political donation data [38] with "minor pro-Trump support" and crypto/energy ties. [OpenAI-Mini] provides no specific donation data. His net worth is estimated at $1B [38]. His primary significance on the council may be his role in the Sacks social network rather than his independent policy expertise. Estimated lean: Insufficient data; likely Republican-leaning given All-In network affiliation.
Jensen Huang The most politically cautious member of the council. [Grok] and [Perplexity] both describe him as having "no large FEC donations" and being "largely neutral" [2]. [Gemini] notes that Huang and Su "intentionally avoid partisan crosshairs because their companies rely on global supply chains and federal CHIPS Act funding" [2], citing Huang's statement: "Whatever the tax rate is, we're going to support it." Nvidia's entanglement with federal policy is profound — AI chip export controls to China [39], potential antitrust issues, and $100B+ in DoD-adjacent contracts [12] — but Huang has managed to remain politically neutral while navigating these pressures. Estimated lean: Genuinely neutral; minimal partisan giving.
John Martinis A UCSB physics professor and Nobel Prize winner (2025, Physics, for quantum computing work) [3]. [Grok] states he has "no political gifts" [15]. He is the council's most credentialed pure scientist and the member whose presence most legitimizes the "science" framing of the council. His net worth is estimated at $50M [15] — the lowest among members with disclosed figures, and consistent with an academic career. Estimated lean: No documented partisan giving; likely politically neutral.
Bob Mumgaard CEO of Commonwealth Fusion Systems (CFS), which is developing high-temperature superconducting magnet technology for fusion energy. [Grok] describes him as "neutral politically" [25]. CFS has received $1B+ in DOE fusion grants [25] and is partnering with the DOE on a fusion pilot plant. His PhD in Applied Plasma Physics from MIT [53] makes him one of the council's genuine scientific voices. Estimated lean: Politically neutral; no documented partisan giving.
Lisa Su CEO of AMD and one of the most respected semiconductor executives in the world. [OpenAI-Mini] cites a specific FEC record: $2,800 to Democratic Rep. Anna Eshoo in 2020 [40], with no comparable Republican gifts documented. [Grok] describes her as having "low-profile Democratic ties" [2]. Like Huang, she appears to deliberately avoid partisan entanglement given AMD's dependence on federal CHIPS Act funding and government contracts. Her MIT PhD in Electrical Engineering [2] makes her one of the council's four credentialed scientists. Estimated lean: Lightly Democratic; minimal partisan giving overall.
Mark Zuckerberg The most complex political figure on the council. [OpenAI] notes he "avoided direct candidate donations" with major giving being philanthropic [5] — specifically the $400M in 2020 election infrastructure funding [2] that was formally nonpartisan but widely perceived as benefiting Democrats. [OpenAI-Mini] estimates his personal giving as approximately 80% Democratic / 20% Republican [41]. Meta's $1M to Trump's inaugural fund [41] was a corporate, not personal, contribution. His recent public gestures toward Trump — including visiting Mar-a-Lago and appearing on Joe Rogan — suggest a deliberate repositioning, but his FEC record remains thin. Meta faces FTC antitrust action over the WhatsApp and Instagram acquisitions [2], and a California jury found Meta liable for social media addiction harm [19]. Estimated lean: Historically 70-80% Democratic; recent repositioning toward neutrality/Republican accommodation.
Section 2: Aggregate Political Lean and Historical Comparison
[Grok] estimates the aggregate recent-cycle lean at approximately 75% Republican [7]. [OpenAI-Mini] estimates 70-80% Republican by contribution weight [41]. [Perplexity] estimates 65-70% Republican-aligned, 20-25% neutral, 5-15% Democratic or mixed. The convergence across providers suggests a working estimate of 70-80% Republican in recent cycles, driven primarily by the mega-donations of Andreessen, Ellison, Sacks, Dell, and Ehrsam.
This is a stark departure from historical PCAST compositions. [Grok] and [Gemini-Lite] provide the most useful comparisons [3]:
- Obama PCAST: ~70% academics, dominated by university presidents, Nobel laureates, and top-tier researchers. Politically diverse but institutionally centrist-liberal. Co-chaired by Eric Lander (genomics) and John Holdren (energy/climate) [42].
- Trump First Term PCAST: ~50% academics, ~50% industry. Maintained significant scientific credibility. Co-chaired by Kelvin Droegemeier (atmospheric science) — an actual scientist.
- Biden PCAST: 28 members [2], approximately 50-60% PhDs or MDs [Perplexity], heavy academic-industry balance. Included AI safety researchers, climate scientists, and biotech experts.
- Trump Second Term PCAST (2026): ~90%+ industry practitioners, 4 PhDs out of 15 members, 0 academic AI researchers, 0 climate scientists, 0 cybersecurity specialists. Co-chaired by a venture capitalist and a political appointee.
[Gemini-Lite] makes the sharpest observation: even Trump's first-term PCAST maintained ~50% academic scientists [62], meaning the 2026 council represents a departure from Trump's own prior standard, not just from Democratic-era norms.
Section 3: Net Worth and Wealth Concentration
Using the most current provider estimates [Grok][OpenAI][Grok-Premium]:
| Member | Estimated Net Worth (2026) |
|---|---|
| Larry Ellison | $247–258 billion |
| Mark Zuckerberg | $225–248 billion |
| Sergey Brin | $140–230 billion |
| Jensen Huang | $160–164 billion |
| Michael Dell | $121–140 billion |
| Marc Andreessen | $1.9–2 billion |
| Safra Catz | $2–3.4 billion |
| Fred Ehrsam | $1–3 billion |
| David Friedberg | ~$1 billion |
| Lisa Su | $1.4–1.5 billion |
| Bob Mumgaard | ~$200 million |
| Jacob DeWitte | ~$500 million |
| John Martinis | ~$50 million |
| David Sacks (co-chair) | $1.5–3 billion |
| Michael Kratsios (co-chair) | $500 million |
| TOTAL | ~$930B–$1.1 trillion |
[OpenAI] notes this exceeds the GDP of over 170 countries [2]. [Grok] specifies that the total exceeds the GDP of the Netherlands, Sweden, and Poland [2]. To put this in context: the Netherlands' GDP is approximately $1.1 trillion; Sweden's is approximately $600 billion; Poland's is approximately $800 billion. The council's combined wealth is roughly equivalent to the GDP of the Netherlands — a developed nation of 18 million people.
The wealth is extraordinarily concentrated: the top 5 members (Ellison, Zuckerberg, Brin, Huang, Dell) account for approximately $893B–$1.04T of the total, meaning the remaining 10 members collectively hold less than 10% of the council's wealth.
Section 4: Conflicts of Interest — The Business-Before-Government Problem
[OpenAI] estimates that "at least 10 of the 15 council members have active financial interests in matters likely to come before federal regulators or agencies" [6]. [Grok] puts the figure at 11 of 15 [12]. The specific conflicts, organized by domain:
Antitrust (DOJ/FTC):
- Sergey Brin/Google: Landmark DOJ antitrust trial over search monopoly [2]. PCAST could advise on tech competition policy.
- Mark Zuckerberg/Meta: FTC antitrust suit over WhatsApp and Instagram acquisitions [2]. PCAST could advise on social media regulation.
- Safra Catz/Oracle: Ongoing antitrust probes [12]. PCAST could advise on cloud computing competition.
Export Controls and China Policy (Commerce/BIS):
- Jensen Huang/Nvidia: AI chip export controls to China are an existential business issue [39]. PCAST could advise on export control policy.
- Lisa Su/AMD: Same export control dynamics as Nvidia [3]. Direct competitor to Huang on the same council.
Federal Contracts (DoD/GSA/IC):
- Safra Catz & Larry Ellison/Oracle: $10B+ in government cloud contracts, JWCC Pentagon contract [3]. PCAST could advise on federal cloud procurement.
- Michael Dell/Dell Technologies: $20B+ in federal hardware contracts [2]. PCAST could advise on federal IT procurement.
Nuclear Regulation (NRC/DOE):
- Jacob DeWitte/Oklo: Seeking NRC licenses for advanced micro-reactors [44]. PCAST could advise on nuclear regulatory reform.
- Bob Mumgaard/Commonwealth Fusion Systems: $1B+ in DOE fusion grants, partnering with DOE on fusion pilot plant [25]. PCAST could advise on fusion energy policy.
Crypto/Financial Regulation (SEC/CFTC):
- Fred Ehrsam/Coinbase & Paradigm: Ongoing SEC and CFTC regulatory battles [2]. PCAST could advise on digital asset regulation.
Platform/Content Regulation:
- Mark Zuckerberg/Meta: Content moderation regulations, Section 230 reform [41]. PCAST could advise on platform governance.
TikTok/Foreign Technology:
- Safra Catz & Larry Ellison/Oracle: Oracle won a major stake in TikTok's U.S. operations [2], a deal that required White House approval. PCAST could advise on foreign technology acquisition policy.
VC Portfolio Conflicts:
- Marc Andreessen/a16z: Invested in Coinbase (Ehrsam's company) [59] and Oklo (DeWitte's company) [59]. Sits on Meta's board [2]. PCAST recommendations on AI regulation, crypto, and nuclear energy directly affect his portfolio.
The structural problem, as [Gemini-Lite] frames it, is that "the council advises on the exact regulatory domains where its members have the most to gain financially" — and there is no disclosed recusal mechanism or ethics framework governing these conflicts.
Section 5: The Web of Business Relationships
[Grok] provides the most systematic mapping of inter-member financial relationships [59]:
Board Relationships:
- Marc Andreessen → Meta board (since 2008) [22] → Zuckerberg is Meta CEO
- This means Andreessen and Zuckerberg have a direct fiduciary relationship on the same advisory council
Investment Relationships:
- a16z (Andreessen) → invested in Coinbase → Ehrsam co-founded Coinbase [59]
- a16z (Andreessen) → invested in Oklo → DeWitte is Oklo CEO [59]
- Craft Ventures (Sacks) → co-invests with Friedberg ventures [2]- Sacks and Friedberg co-host All-In podcast [13] — a pre-existing social and ideological coordination mechanism
Competitive Relationships:
- Jensen Huang (Nvidia) vs. Lisa Su (AMD): Direct chip competitors [3] — both on the same council advising on semiconductor policy
- Oracle (Catz/Ellison) vs. Google (Brin): Cloud computing competitors, both facing antitrust scrutiny
- Oracle, Google, and Meta: All compete in cloud/AI infrastructure [2]Shared Regulatory Exposure:
- Nvidia and AMD: Both subject to the same AI chip export control regime
- Oracle and Dell: Both major federal IT contractors competing for the same government contracts
The Fairshake PAC Network:
- Andreessen, Ehrsam, and Sacks are all connected to the pro-crypto Fairshake PAC ecosystem [3], creating a shared political infrastructure that predates the council
This web of relationships means the council is not 15 independent voices — it is a network of financially entangled actors with overlapping interests, shared investors, and pre-existing social bonds. The All-In podcast network (Sacks, Friedberg, and adjacent figures) effectively functions as a pre-deliberation forum where council positions may be informally coordinated before formal meetings.
Section 6: Scientific Credentials — A Science Council in Name Only
[Grok] identifies 5 PhD holders; [OpenAI] and [Grok-Premium] identify 4; [OpenAI-Mini] and [Perplexity] identify 2-3. The defensible count is 4 confirmed research PhDs among the 13 members:
- John Martinis — PhD, Physics, UCSB; 2025 Nobel Prize in Physics for quantum computing [3]. The council's most credentialed scientist.
- Lisa Su — PhD, Electrical Engineering, MIT [2]. A genuine engineer-scientist who has published research.
- Jacob DeWitte — PhD, Nuclear Engineering, MIT [2]. Applied scientist in a critical emerging energy domain.
- Bob Mumgaard — PhD, Applied Plasma Physics, MIT [53]. Active researcher in fusion science.
Notable non-PhDs:
- Sergey Brin: MS, Computer Science, Stanford (dropped out of PhD program) [7]
- Jensen Huang: MS, Electrical Engineering, Stanford [8]
- Marc Andreessen: BS, Computer Science, Illinois
- Larry Ellison: No degree (dropped out of University of Illinois and University of Chicago)
- Mark Zuckerberg: No degree (dropped out of Harvard)
- Michael Dell: No degree (dropped out of UT Austin)
For comparison [Grok][Perplexity][Gemini-Lite]:
- Obama PCAST: ~60-70% PhDs, dominated by Nobel laureates, university presidents, and research scientists [42]
- Biden PCAST: ~50-60% PhDs or MDs, including AI safety researchers and climate scientists
- Trump First Term PCAST: ~50% academics with research credentials [62]
- Trump Second Term PCAST (2026): ~27% PhDs (4 of 15), 0 academic AI researchers, 0 climate scientists, 0 biomedical researchers
[Ars Technica]'s headline — "Trump staffs science and technology panel with non-scientists" [60] — is the most accurate single-sentence description of this credential gap.
Section 7: Diversity Analysis
Gender: 2 women (Catz, Su) out of 15 members = 13% female, 87% male [3]. For context, women make up approximately 26% of the computing workforce and 28% of STEM workers. The Biden PCAST was approximately 40% female.
Race/Ethnicity: The council is overwhelmingly white and Asian-American. [OpenAI], [OpenAI-Mini], [Grok], and [Perplexity] all confirm zero Black, Hispanic, or Indigenous members [4]. Black Americans make up approximately 9% of the STEM workforce; Hispanic Americans approximately 8%. Neither group has representation on this council.
Age: [Grok] estimates average age at approximately 55 [16]. The range spans from Zuckerberg (41) [2] to Ellison (81) [3]. [OpenAI-Mini] estimates median age in the mid-50s [27]. The council skews toward established, senior executives rather than emerging voices in technology.
Geographic: Overwhelmingly Silicon Valley and Bay Area-based. No representation from the Midwest, South, or non-coastal technology hubs (Austin, Seattle, Boston, Research Triangle).
Institutional Background: 0 representatives from academia, 0 from civil society, 0 from organized labor, 0 from consumer protection advocacy, 0 from national laboratories [2]. The council is exclusively drawn from the private sector.
This composition is consistent with [Grok]'s observation that it "ties to Trump's anti-DEI push" [17] — the council's demographic profile mirrors the administration's broader skepticism toward diversity initiatives.
Section 8: Strengths and Blind Spots
Genuine Strengths [Grok][Grok-Premium][Perplexity]:
- Unmatched operational experience: The council collectively runs the companies that build, deploy, and scale AI infrastructure. Huang (Nvidia), Su (AMD), Ellison/Catz (Oracle), Zuckerberg (Meta), and Brin (Google) have direct, real-time knowledge of AI compute constraints, data center energy demands, and deployment bottlenecks that no academic panel could match.
- Capital deployment expertise: Andreessen, Sacks, Ehrsam, and Friedberg understand venture capital dynamics, startup ecosystems, and the conditions that produce technological breakthroughs.
- Energy transition knowledge: DeWitte (nuclear) and Mumgaard (fusion) represent genuine expertise in the energy infrastructure that AI's compute demands will require.
- Semiconductor supply chain: Huang and Su have direct knowledge of the global semiconductor supply chain, TSMC dependencies, and the strategic implications of chip manufacturing geography.
- Quantum computing: Martinis is a world-class quantum computing researcher whose expertise is directly relevant to next-generation computing policy.
Critical Blind Spots [OpenAI][OpenAI-Mini][Grok][Perplexity][Gemini-Lite]:
- AI safety and alignment: No member has a background in AI safety research. The council includes no one from Anthropic, DeepMind's safety team, or academic AI safety programs. This is the most consequential gap given that AI safety is arguably the most important policy question the council will face.
- Cybersecurity: No cybersecurity specialist. As AI systems become critical infrastructure, the absence of expertise in adversarial robustness, critical infrastructure protection, and national security applications of AI is a serious gap [2].
- Biotech and life sciences: No biologist, physician, or biotech researcher. AI's most transformative near-term applications may be in drug discovery, genomics, and healthcare — domains entirely unrepresented on this council.
- Climate science: No climate scientist or environmental researcher [2]. AI's energy consumption is a major climate issue; the council has no one to represent this perspective.
- Academic AI research: No academic computer scientist working on foundational AI research. The council has no Fei-Fei Li, no Yoshua Bengio, no Stuart Russell [Perplexity].
- Open-source AI: No representative of the open-source AI community (Hugging Face, Linux Foundation, Apache). The council's composition strongly favors proprietary, closed AI systems.
- Labor and economics: No labor economist, no worker advocate, no sociologist studying AI's impact on employment. The council will advise on "workforce impacts" with no one representing workers.
- International perspective: No non-American technology leader. AI governance is inherently global; the council has no perspective from European, Asian, or Global South technology ecosystems.
- Defense and national security: No DARPA leadership, no former intelligence community official, no defense technology specialist [Perplexity].
Section 9: Why Musk and Altman Were Excluded
Elon Musk: [Grok-Premium] offers the most nuanced analysis [28]: Musk already has direct White House access through DOGE, making a separate PCAST seat potentially redundant or creating jurisdictional confusion. His increasingly erratic public behavior (X platform controversies, DOGE overreach) may have made him a liability for a council meant to project credibility. [Perplexity] suggests he is "too toxic or unpredictable." There is also a competitive dynamic: Musk's xAI (Grok) competes directly with Google (Brin), Meta (Zuckerberg), and Nvidia (Huang) — his presence would have created immediate tensions. Finally, [OpenAI] notes the Musk-Altman feud [47] — including Musk's lawsuit against OpenAI — means including either one without the other would have been seen as taking sides in a high-profile tech dispute.
Sam Altman: [Grok-Premium] suggests Altman's exclusion reflects "misalignment on safety emphasis, prior Democratic ties, and rivalry with council members" [2]. OpenAI's safety-first positioning — its support for AI regulation, its advocacy for compute governance, its cautious public messaging — is philosophically at odds with the deregulatory agenda this council will likely pursue. Altman also has a complicated relationship with the Trump administration: he participated in the Stargate announcement but has maintained positions (on AI safety, on international cooperation) that conflict with the council's likely direction. His prior Democratic donor history and his public support for AI governance frameworks make him a poor fit for a council whose primary mandate appears to be removing regulatory barriers. The Musk-Altman legal conflict [47] also makes including Altman politically complicated given Musk's DOGE role.
What Their Absence Signals: The exclusion of both figures signals that the administration is not seeking the most prominent AI voices — it is seeking the most commercially aligned and politically loyal ones. Musk and Altman are both too independent, too publicly prominent, and too likely to generate headlines that distract from the council's work. The administration appears to prefer a council that will produce recommendations quietly and efficiently, without the drama that either figure would bring.
Section 10: Who Is Missing
[Perplexity] provides the most comprehensive list of absent expertise. Beyond the obvious names:
Academic AI:
- Fei-Fei Li (Stanford, AI4ALL, former Google Cloud AI chief) — the most prominent academic AI researcher in the U.S.
- Yoshua Bengio (Mila, Montreal) — Turing Award winner, AI safety advocate
- Stuart Russell (Berkeley) — foundational AI safety researcher, author of Human Compatible
- Andrew Ng (Coursera, Landing AI) — AI education and deployment expert
AI Safety:
- Dario Amodei (Anthropic) — the most prominent AI safety-focused CEO
- Paul Christiano (ARC Evals) — leading AI alignment researcher
- No one from DeepMind's safety team
Cybersecurity:
- No former NSA/CISA director
- No academic cybersecurity researcher
- No critical infrastructure protection specialist
Biotech/Life Sciences:
- Jennifer Doudna (UC Berkeley, CRISPR pioneer) — Nobel laureate whose work is directly relevant to AI-biology intersections
- No synthetic biology representative
- No physician or healthcare AI expert
Climate/Energy:
- No climate scientist
- No renewable energy expert (the council's energy focus is exclusively nuclear/fusion)
- No NOAA or EPA scientific representative
Open Source/Public Interest:
- No Linux Foundation or Apache representative
- No Electronic Frontier Foundation or civil liberties voice
- No consumer protection advocate
Defense/National Security:
- No former DARPA director
- No former intelligence community official
- No defense technology specialist
Labor/Economics:
- No labor economist
- No worker advocate
- No sociologist studying AI's employment impacts
Geographic/Demographic Diversity:
- No representative from HBCUs or minority-serving institutions
- No representative from non-coastal technology hubs
- No international technology leader
Section 11: The Sacks Effect — How This Council Differs from Traditional PCAST
[OpenAI-Mini] describes Sacks as "a libertarian-turned-Trump ally" [2] who represents "a profit-driven, pro-innovation agenda rather than a cautionary, evidence-first approach" [2]. [Gemini-Lite] notes that under Sacks, the council is likely to "prioritize aggressive deregulation" and "reduce federal oversight of AI and crypto" [13].
A traditional PCAST under a scientist co-chair would:
- Commission independent research before making recommendations
- Seek input from academic researchers, civil society, and international experts
- Weigh risks and benefits of emerging technologies with precautionary consideration
- Produce reports with extensive peer review and public comment periods
- Maintain independence from the industries being regulated
Under Sacks, the council is more likely to:
- Start from a deregulatory presumption and work backward to justification
- Treat industry consensus as equivalent to scientific consensus
- Prioritize speed of recommendation over rigor of analysis
- Use PCAST's advisory role to provide political cover for predetermined policy positions
- Coordinate informally through existing social networks (All-In podcast, a16z ecosystem) before formal deliberation
[Grok-Premium] notes that Sacks has already "dropped the 'AI czar' label" while maintaining policy influence [26] — suggesting a preference for informal power over formal accountability. His PCAST co-chairmanship extends this pattern: maximum influence, minimum institutional constraint.
The most significant structural difference from traditional PCAST is the absence of any mechanism for independent scientific review. Previous PCAST reports were credible in part because they were produced by scientists who had no financial stake in the outcomes. This council's recommendations will be produced by people who collectively stand to gain or lose hundreds of billions of dollars based on the policies they recommend.
Section 12: Policy Outlook — Next 12 Months
Most Likely Outputs [OpenAI][Grok][Grok-Premium][Perplexity]:
-
AI Deregulation Framework: A PCAST report recommending against mandatory AI safety requirements, favoring voluntary industry standards. Likely to cite competitiveness with China as justification [2].
-
CHIPS Act Expansion/Acceleration: Recommendations to expand semiconductor manufacturing subsidies, streamline permitting for fab construction, and reduce export control complexity [2].
-
Nuclear/Fusion Fast-Track: Recommendations to streamline NRC licensing for advanced reactors (directly benefiting Oklo) and accelerate DOE fusion funding (directly benefiting CFS) [2].
-
Crypto/Digital Asset Framework: Pro-crypto legislative guidance, likely recommending light-touch SEC/CFTC oversight of digital assets [2].
-
China AI Competition Strategy: Recommendations on AI chip export controls, likely recommending targeted rather than broad restrictions to protect Nvidia and AMD revenues [39].
-
Workforce AI Training: Recommendations on AI education and workforce reskilling, likely favoring private-sector-led initiatives over federal programs [2].
Biggest Risks [OpenAI-Mini][Grok][Grok-Premium][Gemini-Lite][Perplexity]:
- Regulatory Capture: The council's recommendations will be produced by the people most financially affected by those recommendations. This is the textbook definition of regulatory capture, and it creates a structural legitimacy problem that will undermine public trust in any output the council produces [2][