Marc Andreessen

Key takeaways from Marc Andreessen’s June 2023 essay, “What Is To Be Done?” on how governments should approach artificial intelligence:

• Let major AI companies build quickly: Andreessen argues that large AI companies should be allowed to develop their systems aggressively, while being prevented from using regulation to create protected cartels or achieve regulatory capture. In his view, government-granted insulation from competition would reduce the benefits of companies that are among the most important technological businesses of the modern economy.
• Keep the market open to start-ups: AI start-ups should be free to compete without either special protection from larger incumbents or government assistance. Even when a start-up does not ultimately succeed, Andreessen argues that its presence can motivate larger companies to improve, leaving consumers and society better off.
• Remove barriers to open-source AI: He proposes that open-source AI should be allowed to proliferate and compete with both large companies and start-ups. Beyond commercial competition, he presents open source as an educational and access mechanism that could help people around the world learn to build and use AI regardless of their financial resources.
• Use AI defensively: To address harmful uses, Andreessen calls for governments and private companies to work together on defensive capabilities. He extends this beyond AI-specific threats to problems such as malnutrition, disease and climate change, arguing that AI should be treated as a tool for solving major societal challenges rather than only as a source of risk.
• Compete globally: The essay also frames AI leadership as a geopolitical contest. Andreessen advocates close coordination between Western governments, scientific institutions and private companies to achieve global leadership over China, while ending with a broader call to build rather than slow technological progress.

The argument combines a strong pro-innovation position with opposition to incumbent protection. Its central policy distinction is between allowing rapid technical development and allowing companies to use public policy to suppress competition.

Read more: Marc Andreessen on X · Wikimedia Commons image

Key takeaways from Marc Andreessen’s September 2026 essay on Cognition and the next phase of software automation:

• From human-speed to compute-speed software: Andreessen argues that software has already reshaped the global economy despite being written almost entirely by hand. His comparison is that technology’s share of the market capitalisation of the world’s ten largest businesses rose from 31.5% to 94.4% over 15 years. The next step is software development itself becoming increasingly automated.
• Agent leverage: Cognition says Devin grew from writing 13% of the company’s production code to more than 90% in one year. Andreessen’s thesis is that this changes the engineer’s role from manually producing code to directing and reviewing fleets of coding agents, potentially multiplying individual output by an order of magnitude or more.
• Early enterprise evidence: The essay cites an eight-month COBOL migration completed in eight days at Mercedes-Benz, a 10x increase in test-generation velocity at Rivian, and automatic remediation of 70% of security vulnerabilities at Itaú. These are company-reported examples rather than independently audited benchmarks, but they illustrate the breadth of tasks being delegated beyond code completion.
• Demand, not substitution, is the central bet: Andreessen revisits the argument that every productivity leap will eliminate software engineering. In his framing, compilers, open source and cloud computing lowered the cost of software and expanded demand for it; coding agents could do the same at a much larger scale by making previously uneconomical software projects viable.
• Strategic implication: If software continues to eat the world, automating software creation could accelerate change across every other industry. That is why Andreessen presents Cognition as more than a single AI application: it is an expression of a16z’s broader belief in technical founders, small teams with powerful tools and software expanding the boundaries of addressable markets.
• Backing the team: Andreessen says a16z’s conviction is grounded in a long relationship with Cognition founder Scott Wu, whom the firm first backed at Lunchclub. He highlights Wu’s competitive-programming record, speed, judgement, customer focus and ambition as reasons for backing Cognition again.

The essay is an explicitly bullish investor thesis, not a neutral market forecast. Its strongest claim is that the bottleneck in software will move from writing code to defining systems, setting goals and exercising judgement over increasingly capable agents.

Read more: Marc Andreessen on X · The Wall Street Journal

Marc Andreessen has been appointed to the US Defense Policy Board, the advisory panel that counsels the Pentagon's civilian leadership on force structure, modernisation and military strategy. The reconstituted board, announced by Defense Secretary Pete Hegseth in June 2026, names thirteen members, chaired by former trade representative Robert Lighthizer.

The appointment has drawn scrutiny because Andreessen Horowitz, the firm he co-founded, holds stakes in several defence-tech companies whose prospects track Pentagon procurement decisions, including Anduril, Skydio, Shield AI, Saronic and Flock Safety. Anduril reached a $61bn valuation in a $5bn round after the US Army awarded it a $20bn enterprise agreement in March 2026.

As board members who do not otherwise work for the government are appointed as special government employees, Andreessen would file a confidential OGE Form 450 disclosure rather than a public filing, and federal conflict-of-interest rules require recusal from matters touching his financial interests. There is no suggestion of improper conduct; the appointment is consistent with a broader migration of Silicon Valley figures into federal roles.

Read more: The Next Web

Image credit: TechCrunch

Key takeaways from the a16z panel "The Media Game Has Changed", with co-founders Marc Andreessen and Ben Horowitz alongside partners Gabby Benort and Eric (19 June 2026):

Old media vs new media. Old media meant restricted channels and short formats where the brand was the company; new media is unlimited formats and channels where the brand is the person. Old media is defensive — its one rule is "don't be interesting" — while new media is offensive and demands you be interesting or get drowned out.

Authenticity is the playbook. Andreessen recounts 1990s media training from a former 60 Minutes producer whose advice was simply to say in public the things you would say to a friend at lunch — talk only about what you know intimately. The same approach is visible in how Elon Musk, Palmer Luckey, Alex Karp and Jensen Huang communicate today via long-form podcasts.

The press has changed. Andreessen argues that legacy media's "speak truth to power" mission overtook its objective-journalism mission around 2017, making it impossible for founders to reliably land favourable coverage. The answer is to "go direct" — tell your own story through your own channels and through allied new-media voices.

The founder is the brand. Companies winning at marketing are fronted by a person, not an abstract corporate brand — Musk for his ventures, Karp for Palantir. Horowitz notes the brand-bearer must be a permanent fixture, not a marketing VP on a three-year run. He ties the abstract-corporate-brand era to centralised media from the 1930s, which is now unwinding.

Outside-in storytelling. The cheat code, per Andreessen, is not to talk about yourself but to tell the most interesting story happening in the world and plug your company into it — Karp talks about US military and superintelligence, never Palantir's product; Ryan Peterson of Flexport reframed freight as the collapsing global supply chain.

Pick the right fights. Responding to attacks can boost a brand — Horowitz cites an Instagram/New York Times clash that drove his biggest post at the time. But avoid amplifying low-follower critics. In new media you want people to both love and hate you; being lukewarm is fatal to being interesting.

Hiring for new media. People trained in old media struggle to switch — it is "opposite world". Look for proof of work: have they built an audience or a brand themselves? a16z hires obsessive storytellers (Alex Denko, Henry, Brent) over traditional marketers.

Common founder mistakes. Gabby Benort warns against over-indexing on distribution and tactics before getting the message right — distribution is just a multiplier, so amplifying a wrong or boring message backfires. Work backwards from the outcome you want (a customer, a hire) rather than dumping everything true about the company.

Source: a16z — 19 June 2026.

Historical Background

• In 1993, while at the University of Illinois at Urbana-Champaign, Marc Andreessen was part of the team that developed Mosaic, the first widely used graphical web browser.
• Andreessen co-founded Netscape Communications in 1994, which commercialized the browser technology. In 2009, he co-founded the venture capital firm Andreessen Horowitz (a16z) with Ben Horowitz.

Source of Capital

• As a general partner in Andreessen Horowitz, the firm's capital is raised from outside investors, which typically include institutional clients and high-net-worth individuals.
• Specific limited partners of Andreessen Horowitz are not publicly disclosed.

Major Events

• In 1999, Netscape was acquired by AOL for $4.2 billion.
• Andreessen later co-founded and served as chairman of Loudcloud, which was rebranded as Opsware and subsequently acquired by Hewlett-Packard for $1.6 billion in 2007.
• In 2019, Andreessen Horowitz changed its structure, registering its general partners as financial advisers rather than venture capitalists to allow for more flexibility in its investment strategies, such as holding larger stakes in public companies or cryptocurrencies.

Allocation Strategy

• Andreessen's investment activities are primarily conducted through Andreessen Horowitz, a venture capital firm.
• The firm is stage-agnostic, investing in companies from seed and venture to later growth stages.
• a16z maintains a broad technology focus, with stated interests in artificial intelligence, biotechnology and healthcare, consumer technology, cryptocurrency, enterprise software, financial technology, and gaming.

Sources

• https://en.wikipedia.org/wiki/Marc_Andreessen
• https://www.britannica.com/biography/Marc-Andreessen
• https://a16z.com/about/
• https://www.forbes.com/sites/alexkonrad/2019/04/02/andreessen-horowitz-is-blowing-up-the-venture-capital-model-again/
• https://en.wikipedia.org/wiki/Andreessen_Horowitz
• https://tradersunion.com/persons/ben-horowitz/