The Browser Revolutionary Turned AI Kingmaker

On April 8, 2025, Andreessen Horowitz (a16z) announced plans to raise $20 billion for what would become the largest AI-focused venture fund in history—surpassed only by SoftBank's Vision Funds. The fund's sole mission: deploy capital into growth-stage American artificial intelligence companies at a scale and velocity unprecedented in venture capital.

Leading this audacious bet stands Marc Andreessen, the 54-year-old co-founder who created the first widely-used web browser at age 22, sold Netscape to AOL for $4.3 billion, and now commands one of Silicon Valley's most influential investment firms with $46 billion in assets under management as of July 2025.

But the $20 billion megafund represents more than capital deployment. It crystallizes Andreessen's evolved thesis: if "software is eating the world"—his famous 2011 prediction—then artificial intelligence is devouring software itself. And he's betting that whoever controls AI capital controls the future.

The timing is strategic. In 2024, a16z participated in 100 reported funding rounds, ranking as the most active global post-seed investor. The firm invested $200 million in OpenAI, $150 million in Anthropic, $100 million in Cohere, and $75 million in Inflection AI. It led Mistral AI's €385 million Series A at a $2 billion valuation and positioned itself to lead former OpenAI CTO Mira Murati's Thinking Machines Lab at a reported $10 billion valuation—despite the startup having no product.

Yet Andreessen's AI empire-building faces mounting scrutiny. His October 2023 "Techno-Optimist Manifesto"—which called AI safety advocates a "cult" and declared AI development restrictions "a form of murder"—ignited fierce backlash. Anthropic CEO Dario Amodei publicly rebuked Andreessen's argument that AI shouldn't be regulated because it's "just math." Safety researchers accused him of "underhanded insults" and "survivorship bias" in dismissing existential AI risks.

Then came the political pivot. After meeting with Biden administration officials about AI policy in May 2024, Andreessen emerged "very scared" and described the conversations as "absolutely horrifying." He subsequently endorsed Donald Trump, joined a $100 million lobbying effort to shape AI regulation, and in November 2025, the Trump White House drafted an executive order—championed by Silicon Valley titans including a16z—to sue states attempting AI regulation and withhold federal funding from noncompliant jurisdictions.

This investigation examines how Marc Andreessen transformed from internet pioneer to AI's most polarizing venture capitalist. Based on analysis of a16z's $45+ billion portfolio, Andreessen's published writings, regulatory filings, and interviews with industry insiders, we trace his journey from techno-utopian ideals to political hardball—and ask whether his concentrated AI bets will create broadly shared prosperity or concentrate unprecedented power in the hands of a few.

From Cedar Falls to Silicon Valley: The Making of a Browser Revolutionary

Marc Lowell Andreessen was born July 9, 1971, in Cedar Falls, Iowa, and raised in New Lisbon, Wisconsin—population 2,554. While still in grammar school, Andreessen taught himself BASIC programming to write computer games. This self-directed technical education would define his career trajectory: identify emerging platforms early, build foundational tools, capture market share before incumbents react.

In 1993, as a computer science student at the University of Illinois at Urbana-Champaign, Andreessen spent much of his time at the National Center for Supercomputing Applications (NCSA). Working with a team of students and staff, he co-authored Mosaic—the first web browser to display inline images rather than requiring separate windows.

Mosaic was announced in January 1993. Within one year, over two million copies were downloaded, significantly increasing web usage from a niche academic tool to mainstream technology. The timing was exquisite: the internet existed, but accessibility remained limited to technically sophisticated users. Mosaic's graphical interface democratized access.

After graduating in 1993 with a bachelor's degree in computer science, Andreessen moved to California's Silicon Valley to work at Enterprise Integration Technologies. Then came the pivotal meeting: Jim Clark, founder of Silicon Graphics who had recently exited the firm, contacted Andreessen. In April 1994, the duo founded Mosaic Communications Corporation.

The University of Illinois objected to the company's use of the Mosaic name. Mosaic Communications rebranded to Netscape Communications, and its flagship product became Netscape Navigator. The browser was commercially launched and, almost overnight, captured over 75 percent market share by mid-1996.

Netscape's August 9, 1995 initial public offering made Andreessen a public figure at age 24. The company's stock price doubled on the first day of trading, valuing Netscape at $2.9 billion despite minimal revenue. Andreessen appeared on the cover of Time magazine, barefoot and sitting on a golden throne—an image that would define the dotcom era's youthful excess and ambition.

But Netscape's dominance proved fleeting. Microsoft integrated Internet Explorer into Windows, leveraging its operating system monopoly to undercut Netscape's market position. By late 1998, Netscape's browser share had plummeted. AOL acquired the company in 1999 for $4.3 billion—a substantial exit, but far below Netscape's peak valuation and a cautionary tale about defensibility against platform owners.

The Opsware Education: Learning to Build Enterprise Software

After AOL acquired Netscape, Andreessen and Ben Horowitz—whom he'd met at Netscape in the mid-1990s where Horowitz worked as product manager—co-founded Loudcloud in 1999 with Tim Howes and In Sik Rhee. The company offered cloud computing infrastructure and managed services for enterprise customers.

Loudcloud went public in March 2001 at the height of the dotcom crash, raising $162.5 million at a $700 million valuation. The timing was catastrophic. Within months, the company's stock collapsed as the bubble burst. Facing insolvency, Andreessen and Horowitz made a desperate pivot: they sold Loudcloud's hosting services to EDS for $63.5 million in July 2002, retaining only the Opsware software business.

Rebranded as Opsware, the company focused on data center automation software. Over the next five years, Andreessen and Horowitz rebuilt the business through painful layoffs, product pivots, and relentless enterprise sales. The experience taught Andreessen lessons that would shape his venture philosophy: persistence through adversity, the value of enterprise contracts, and the difficulty of competing against incumbents without technical moats.

In September 2007, Hewlett-Packard acquired Opsware for $1.6 billion. The sale provided financial returns but, more importantly, operational credentials. Andreessen and Horowitz had now built, scaled, nearly killed, revived, and successfully exited two companies. They understood founder psychology, product-market fit challenges, and the brutal realities of enterprise software sales.

Founding Andreessen Horowitz: The Super Angels Institutionalize

Between 2006 and 2010, Marc Andreessen and Ben Horowitz actively invested in technology companies as "super angels." Separately and together, they deployed $4 million across 45 startups, including Twitter. This period established their reputation for backing ambitious founders early and providing hands-on operational support.

On July 6, 2009, Andreessen and Horowitz announced Andreessen Horowitz (a16z) with an initial capitalization of $300 million. The firm's founding thesis challenged conventional venture capital wisdom on multiple dimensions.

Operational support over board governance: Rather than simply providing capital and strategic oversight, a16z built extensive support functions—recruiting, PR, business development, customer introductions—treating portfolio companies as clients rather than investments.

Equal partnership with founders: The firm's "founder-friendly" approach granted entrepreneurs voting control, multiple board seats, and protection against investor interference. This contrasted sharply with traditional VC governance models emphasizing investor control.

Software focus: In 2009, as the economy recovered from the financial crisis, most VCs remained cautious about consumer internet companies after the dotcom crash. a16z bet aggressively on software companies, anticipating the cloud computing and mobile revolutions.

The timing proved prescient. Within three years, a16z grew to $2.7 billion under management. Early investments in Facebook, Foursquare, GitHub, Pinterest, and LinkedIn delivered exceptional returns, validating the firm's thesis and founder-friendly approach.

By 2011, Andreessen codified his investment philosophy in the Wall Street Journal essay "Why Software Is Eating the World." The 2,500-word manifesto argued that software companies would dominate every industry—retail (Amazon), entertainment (Netflix), telecommunications (Skype), even finance and healthcare. At the time, enterprise software spending stood at $269 billion annually according to Gartner. Andreessen predicted explosive growth.

He was correct. By 2025, companies worldwide were on pace to spend nearly $600 billion on enterprise software—a 123 percent increase. The "software is eating the world" thesis became Silicon Valley's defining framework for understanding technology's economic impact.

The AI Pivot: From Software to Superintelligence

Andreessen recognized artificial intelligence's transformative potential earlier than most venture capitalists. While competitors dismissed AI as hype or focused on narrow applications, a16z began building a comprehensive AI portfolio spanning foundation models, infrastructure, and applications.

Foundation model bets (2024-2025): The firm invested $200 million in OpenAI, $150 million in Anthropic, $100 million in Cohere, and $75 million in Inflection AI. In December 2023, a16z led Mistral AI's €385 million Series A at a $2 billion valuation—the firm's first bet on an open-source large language model creator and a strategic hedge against OpenAI and Anthropic's closed models.

Infrastructure layer: a16z positioned itself to lead Thinking Machines Lab's funding round at a reported $10 billion valuation in April 2025, despite former OpenAI CTO Mira Murati's stealth startup having no product. The astronomical valuation demonstrates a16z's willingness to pay premium prices for elite AI talent, even pre-product.

Application layer: The firm backed over 20 AI startups across verticals—Cursor (AI coding), Harvey AI (legal), Hippocratic AI (healthcare), Character.AI (conversational AI), and ElevenLabs (voice synthesis). This diversified approach hedges against winner-take-all outcomes in specific categories.

In April 2025, a16z announced plans to raise $20 billion for its largest-ever fund focused on growth-stage AI companies. The fund would primarily target follow-on investments in existing portfolio companies like Databricks and xAI, addressing AI startups' capital-intensive requirements. By October 2025, revised reports indicated the firm was targeting $6 billion for growth-stage investments, $3 billion for AI, and $1 billion for defense technology—totaling $10 billion, with an additional $1.5 billion each for AI applications and AI infrastructure.

The deployment strategy reveals a16z's market position. Much of the new fund would be reserved for follow-on investments, ensuring continued support for high-cost ventures already in the portfolio. This approach locks in a16z's relationships with category leaders while preventing dilution from competitor investors in subsequent rounds.

As of July 2025, Andreessen Horowitz ranks first among venture capital firms by assets under management, with $46 billion. The firm's AI portfolio positioning—spanning foundation models, infrastructure, and applications—makes it the most comprehensively diversified AI investor in Silicon Valley.

The Techno-Optimist Manifesto: AI as Civilizational Imperative

On October 16, 2023, Marc Andreessen published a 5,200-word essay to the a16z website titled "The Techno-Optimist Manifesto." The document set out to defend a strong conception of techno-optimism, arguing that technological advancement is the primary driver of human progress and well-being.

The manifesto's core thesis: growth is progress—leading to vitality, expansion of life, increasing knowledge, higher well-being. Everything good is downstream of growth. Andreessen argued that because AI might save lives, any pause on AI development that limits preventable deaths "is a form of murder."

The essay named explicit "enemies" of progress: sustainability, ESG (environmental, social, governance), trust and safety, tech ethics, risk management, and precautionary principles. Andreessen characterized these movements as obstacles to technological advancement and human flourishing.

On AI specifically, the manifesto declared: "AI is quite possibly the most important – and best – thing our civilization has ever created, certainly on par with electricity and microchips, and probably beyond those." He predicted AI would protect humanity from harms like climate crisis, provide avenues for new medicines and space exploration, and usher in an era of abundance.

The manifesto's rhetorical style borrowed from early 20th-century futurist manifestos—declarative, absolutist, dismissive of critics. It listed philosophical influences including Nietzsche, Rand, and Hayek, signaling Andreessen's libertarian-accelerationist intellectual framework.

Reaction was swift and polarized. Supporters praised Andreessen's defense of technological progress against what they perceived as excessive precaution. Critics condemned the essay's dismissal of legitimate safety concerns, its conflation of technology skepticism with anti-progress ideology, and what some called "techno-supremacy."

In a nearly 7,000-word follow-up essay "Why AI Will Save the World" published in June 2023, Andreessen expanded his argument. He called AI safety critics "a cult" and dismissed the profession of "AI safety expert" and "AI risk researcher," saying "They are paid to be doomers." He characterized concerns about existential AI risks as "hysterical fear and paranoia."

Anthropic CEO Dario Amodei publicly rebuked Andreessen's argument that AI shouldn't be regulated because it's "just math." In a November 2024 Fortune interview, Amodei stated that reducing AI to mathematics ignored its real-world impacts and the need for governance frameworks.

Effective Altruism researchers published detailed critiques arguing Andreessen's logic relied on survivorship bias—grouping AI with a skewed sample of only safe technologies while ignoring failed or harmful innovations. Critics noted he "failed to present a compelling case against the potential risks associated with unmanaged AI."

The Political Turn: From Biden to Trump

By late 2023, Marc Andreessen—formerly a supporter of the Democratic Party—grew politically disillusioned with the Biden administration. Sources cited volatility in the technology space attributed to antitrust enforcement on the tech sector and renewed cryptocurrency regulation. According to The New York Times, Biden's proposed "billionaire minimum income tax" accelerated Andreessen's shift toward the Trump campaign.

In May 2024, Andreessen met with Biden administration officials to discuss AI policy. After the meetings, he publicly stated he was "very scared" and described the conversations as "absolutely horrifying." Though Andreessen didn't detail specific concerns, industry observers interpreted his reaction as opposition to proposed AI safety regulations and oversight mechanisms.

Shortly afterward, Andreessen endorsed Donald Trump for president. The endorsement signaled a broader Silicon Valley realignment as tech executives grew frustrated with Democratic administration policies on antitrust, cryptocurrency, and AI regulation.

In August 2025, Andreessen Horowitz joined other major Silicon Valley investors in backing the American Innovators Network—a new $100 million political organization seeking to influence US artificial intelligence policy under President Trump. The network spent over $350,000 on lobbying and digital advertising against New York's RAISE Act, which would mandate safety testing for AI models.

The political investments paid dividends. In November 2025, the Trump administration drafted an executive order that would thwart states' AI laws by launching legal challenges and withholding federal funding. The draft order gave Attorney General Pam Bondi 30 days to establish an "AI Litigation Task Force" whose sole mission would be challenging state AI regulations.

The executive order represented a major win for the AI industry, whose leaders—including Sam Altman's OpenAI, Andreessen Horowitz, and other Silicon Valley titans—opposed inconsistent state-by-state policy approaches. The order would preempt California's SB 1047 (requiring safety testing for large AI models) and New York's RAISE Act, among other state initiatives.

At the February 2025 Paris Global AI Summit, both the United States and UK declined to sign a follow-up declaration on ethics and safety. Vice President J.D. Vance publicly pushed back against "excessive" safety-oriented regulations, echoing Andreessen's techno-optimist arguments.

As of November 22, 2025, the Trump administration's order targeting state-level AI regulation was reportedly on hold amid pushback from state officials and safety advocates. But the episode demonstrated Andreessen's political influence and his willingness to deploy capital and relationships to shape regulatory outcomes favorable to a16z portfolio companies.

The Investment Strategy: Full-Stack AI Dominance

Andreessen Horowitz's AI investment strategy differs fundamentally from competitors like Sequoia Capital and Thrive Capital. While Sequoia emphasizes companies with clear revenue paths and sustainable competitive advantages, and Thrive focuses on concentrated bets in category-defining companies, a16z pursues comprehensive coverage across the AI stack.

Foundation models: By investing in OpenAI, Anthropic, Cohere, Inflection, and Mistral, a16z hedges against uncertainty about which model provider will dominate. This "spray and pray" approach at the foundation layer ensures a16z benefits regardless of which models win consumer and enterprise adoption.

Infrastructure: Investments in Groq (inference chips), Cerebras (wafer-scale AI processors), and Databricks (lakehouse architecture) position a16z in the critical compute and data layers beneath applications. These picks-and-shovels bets capture value regardless of which applications succeed.

Applications: The firm backs over 20 vertical AI applications—Cursor achieving $500 million ARR in coding, Harvey AI reaching $100 million ARR in legal, Hippocratic AI completing 115 million patient interactions in healthcare. This diversified approach assumes multiple billion-dollar vertical applications will emerge, not winner-take-all concentration.

Defense technology: a16z allocated $1 billion for defense tech ventures, including investments in Anduril. This positions the firm to benefit from government AI procurement and aligns with Republican administration priorities on military AI applications.

The strategy contrasts with Sequoia's more cautious approach. David Chan, a Sequoia partner, has expressed concerns about AI infrastructure costs and questioned whether massive investments yield sufficient profits. Angela Strange, an a16z partner, takes a more optimistic view, believing AI is already causing significant changes and will drastically reduce productivity costs despite high implementation expenses.

Thrive Capital achieved the highest aggregate investment value in 2024, largely due to leading two massive rounds: $10 billion for Databricks and $6.6 billion for OpenAI. Thrive's concentrated strategy focuses on high-conviction investments in category-defining companies, often taking board seats and actively participating in strategic decisions.

a16z's approach differs: participate in 100+ rounds annually, provide extensive operational support through the firm's platform teams, and reserve massive follow-on capital to support winners. This volume-based strategy assumes a16z's superior pattern recognition and support infrastructure will identify and accelerate category leaders.

The Critics Strike Back: Is Andreessen Dangerous?

Marc Andreessen's influence extends beyond capital deployment. His essays, podcast appearances, and political advocacy shape Silicon Valley's ideological consensus on AI development. But critics argue his techno-optimist framework dismisses legitimate risks and concentrates dangerous power.

AI safety researchers: Effective Altruism researchers published detailed rebuttals to Andreessen's "Why AI Will Save the World" essay, arguing he employs "underhanded insults about the AI Safety community" and conflates safety concerns with "woke 'trust and safety' people." They note his logic relies on survivorship bias—grouping AI with technologies that succeeded while ignoring failed innovations that caused harm.

Anthropic leadership: Dario Amodei, whose company received $150 million from a16z, publicly criticized Andreessen's argument that AI shouldn't be regulated because it's "just math." Amodei argued this reductionist view ignores AI's real-world impacts and governance needs.

Academic critics: Stuart Russell (UC Berkeley, founder of Center for Human-Compatible AI) and Yoshua Bengio (2018 Turing Award winner) have articulated concerns about AI existential risks that directly contradict Andreessen's dismissals. Russell's technical work on value alignment and Bengio's leadership of the 2025 International AI Safety Report represent the research consensus Andreessen characterizes as "cult-like."

Regulatory advocates: Helen Toner, former OpenAI board member who participated in Sam Altman's removal, testified to Congress about AI safety governance failures. Her work at Georgetown's Center for Security and Emerging Technology directly contradicts Andreessen's opposition to AI regulation.

The criticism centers on several concerns:

Dismissing existential risks: Rather than acknowledging documented risks of AI systems, Andreessen claims AI could be "a way to make everything we care about better" without grappling with failure modes, misalignment, or dual-use dangers.

Opposition to all regulation: Andreessen argued that placing limits on AI development "would cripple industry and academia, and could even lead to war between superpowers." Critics see this as absolutist thinking that precludes reasonable safety standards.

Financial conflicts: As a16z deploys billions into AI companies, Andreessen's advocacy against regulation directly benefits his portfolio. Critics question whether his positions reflect genuine philosophical convictions or financial self-interest.

Concentration of power: By backing OpenAI, Anthropic, and multiple foundation models while opposing regulation, Andreessen positions a16z to influence AI development without democratic accountability or safety constraints.

The Competitive Landscape: Where a16z Stands Among AI Investors

Andreessen Horowitz's $46 billion in assets under management and $20 billion planned AI megafund position it as Silicon Valley's most ambitious AI investor. But how does it compare to competitors?

Sequoia Capital focuses on sustainable business models and clear paths to profitability. Co-stewards Pat Grady and Alfred Lin delivered the 2025 AI Ascent keynote emphasizing AI's trillion-dollar opportunity, but Sequoia's portfolio construction emphasizes execution and revenue traction over pure technology bets. Sequoia backed OpenAI, Snowflake, and UiPath, but maintains smaller check sizes and more diversified exposure across sectors.

Thrive Capital, led by Joshua Kushner, takes concentrated high-conviction bets. Thrive led OpenAI's $40 billion round with a $1 billion commitment and co-led Cursor's $900 million Series C. The firm's strategy: identify category-defining companies early, take large positions, and support through multiple growth stages. Thrive's $1 billion+ OpenAI investment at $300 billion valuation represents the most aggressive single bet on AI's winner-take-most dynamics.

Founders Fund, co-founded by Peter Thiel, emphasizes contrarian AI applications beyond chatbots. The firm led Anduril's $2.5 billion Series G with a $1 billion commitment (FF's largest ever) and backs Anthropic and Crusoe Energy. Founders Fund's libertarian ideology and defense tech focus differentiate it from consumer-focused competitors.

Lightspeed Venture Partners led Anthropic's $3.5 billion Series E at $61.5 billion valuation, one of 2025's largest AI rounds. Partner Jeremy Liew's early Snap investment and consumer product instincts inform Lightspeed's AI application theses.

a16z's competitive advantages:

• Platform support: The firm's recruiting, PR, and business development teams provide portfolio companies operational assistance beyond capital.
• Network effects: With 100+ portfolio companies, a16z facilitates partnerships, customer introductions, and talent sharing across its ecosystem.
• Follow-on capital: The $20 billion megafund ensures a16z can support winners through multiple growth stages without dilution.
• Brand recognition: Andreessen's public profile and thought leadership attract entrepreneurs seeking marquee investors.

But a16z faces challenges:

• Valuation discipline: Paying $10 billion pre-product for Thinking Machines Lab suggests potentially excessive valuations driven by FOMO (fear of missing out).
• Concentration risk: Heavy exposure to OpenAI, Anthropic, and foundation models creates portfolio correlation—if the model layer commoditizes, multiple a16z bets fail simultaneously.
• Political backlash: Andreessen's Trump endorsement and regulatory lobbying may alienate Democratic policymakers and create regulatory risks.

The Future: AI's Power Broker or Overextended Bull?

Marc Andreessen's influence on artificial intelligence extends far beyond capital deployment. Through a16z's $46 billion portfolio, his public essays reaching millions, and political lobbying shaping regulatory outcomes, Andreessen has positioned himself as AI's ideological and financial kingmaker.

The bull case for Andreessen's strategy:

Comprehensive AI coverage: By investing across foundation models, infrastructure, and applications, a16z captures value regardless of which layer commoditizes or which specific companies win.

Follow-on firepower: The $20 billion megafund ensures a16z can support category winners through multiple growth stages, preventing dilution and maintaining influence.

Operational support: The firm's platform teams provide recruiting, PR, and business development assistance that accelerates portfolio company growth beyond pure capital.

Ideological alignment: Andreessen's techno-optimist manifesto and regulatory opposition create favorable policy environment for a16z portfolio companies under Trump administration.

Track record: The firm's early bets on Facebook, GitHub, and Airbnb demonstrate pattern recognition for transformative platforms.

The bear case:

Valuation excess: Paying $10 billion pre-product for Thinking Machines Lab and backing multiple foundation models at astronomical valuations suggests bubble dynamics and insufficient discipline.

Concentration risk: Heavy exposure to OpenAI and Anthropic creates correlated outcomes—if foundation models commoditize or face regulatory constraints, multiple a16z positions suffer simultaneously.

Regulatory backlash: Andreessen's political advocacy and opposition to AI safety measures may trigger regulatory responses that harm portfolio companies, especially if Democratic administrations return to power.

Talent exodus: As AI founders launch new ventures (Mira Murati, Ilya Sutskever, Mira Murati), competition for elite teams intensifies. a16z's willingness to pay premium prices may not guarantee access to best founders.

Philosophical opposition: Andreessen's dismissal of AI safety concerns alienates researchers, academics, and safety-focused entrepreneurs who may choose competing investors aligned with responsible development.

Conclusion: The Techno-Optimist's Gamble

Marc Andreessen's journey from 22-year-old browser creator to 54-year-old AI kingmaker reveals Silicon Valley's evolving power structure. His early recognition that "software is eating the world" proved prescient. His current bet—that AI will eat software and create trillions in value—may prove equally visionary.

But the stakes are higher now. Mosaic and Netscape democratized internet access. AI systems, if Andreessen's critics are correct, could pose existential risks requiring governance frameworks he vehemently opposes.

The $20 billion AI megafund represents more than capital deployment. It's an ideological statement: that technological acceleration should proceed unfettered by safety constraints, that market competition will solve alignment problems, and that concentrated capital in the hands of a few elite VCs will produce better outcomes than democratic governance.

Andreessen's political evolution—from Democratic donor to Trump supporter to $100 million lobbying effort opponent of state AI regulation—demonstrates his willingness to deploy all available resources to shape outcomes favorable to his portfolio.

Whether history judges Andreessen as visionary or reckless depends on questions we can't yet answer: Will AI systems remain controllable as capabilities increase? Will concentrated capital in foundation models create monopolistic dynamics or competitive abundance? Will Andreessen's opposition to safety research prove justified optimism or catastrophic hubris?

What's certain: as a16z deploys $20 billion across hundreds of AI companies, Marc Andreessen's influence over artificial intelligence's trajectory rivals that of any individual outside foundation model labs. His techno-optimist manifesto isn't merely philosophical positioning—it's the operating system for Silicon Valley's AI elite.

The debate over Andreessen's legacy has begun. His supporters see a defender of progress against excessive caution and regulatory capture. His critics see dangerous concentration of power in the hands of an ideologue dismissive of legitimate safety concerns.

Both may be right. And that's what makes Marc Andreessen's AI empire so consequential—and so unsettling.