On Sand Hill Road, mistakes are usually composted into mythology. A bad pass becomes a vague story about timing; a lost auction becomes proof of discipline. Bessemer Venture Partners does something stranger. It publishes the names. Apple, Google, Airbnb, FedEx, Intuit - all appear in the firm's Anti-Portfolio, a cabinet of opportunities declined, mishandled or simply misunderstood. The page is funny because venture capital is not known for public self-abasement. It is useful because the joke carries an operating principle: memory matters more when it includes the wrong answers.
That candor is the liveliest entrance to an institution that might otherwise look imposing. Bessemer reports roughly $20 billion under management, more than 450 portfolio companies and over 155 public listings. Its current and former investments range from LinkedIn, Pinterest and Shopify to Twilio, Toast, Canva, ServiceTitan, Abridge and Anthropic. It invests globally, from seed rounds to growth checks, and now buys substantial stakes in established technology businesses through BVP Forge.
The customers are founders and management teams, but the product is not one thing. Capital is the admission ticket. Around it sits a mesh of operating advisers, recruiting help, market research, executive coaching, peer networks and practical courses on pricing, product-market fit and go-to-market design. The firm's own limited partners are the other customer: they supply the pools of money and expect returns from acquisitions, secondary sales and public offerings. Bessemer lives in the middle, translating long-duration capital into a series of concentrated judgments about young companies.
01 / The origin
A technology firm with steel in the family
Bessemer's lineage begins before venture capital had a name. Henry Phipps made his fortune alongside Andrew Carnegie and Henry Clay Frick, commercializing steel at a scale enabled by the Bessemer process. Phipps placed wealth in a trust for his descendants. The family office that managed it eventually developed a taste for high-risk, high-growth investments. In 1975, that operation focused on high technology, opened in Silicon Valley and formally became Bessemer Venture Partners.
The history supplies more than a handsome origin story. Family capital could tolerate cycles that punish impatient investors. Projects might take five, 10 or 20 years. It also made the institution less dependent on a charismatic founder whose preferences become unwritten law. Bessemer's name honors a process, not a personality - an accidental bit of governance branding that still suits the place.
Bessemer lists 1911 as its founding year, while its modern venture identity dates to 1975. Both dates describe something real: the continuity of the investment institution and the arrival of the technology partnership founders recognize today. That dual clock helps explain the firm's pitch. It wants to be early in a company's life and durable enough to remain relevant through the uncomfortable middle.
“The focal point was always the technology, not the individual.”Bessemer partner Jeremy Levine, on the firm's inherited ethos
02 / The machine
Conviction without a corner office
Many venture partnerships say they welcome debate. Bessemer's design gives debate a specific consequence: a partner who develops conviction can pursue an investment without having to flatter a founder-chairman or protect an inherited house view. The model is deliberately disaggregated. Sector roadmaps turn individual research into firm knowledge; investment memos preserve how a decision looked before hindsight arrived. Partners own their calls, while the wider platform supplies pattern recognition and support.
That independence solves a familiar venture problem. Breakout companies rarely arrive looking like consensus. A committee optimized to avoid embarrassment can reject the oddest founders and newest categories - precisely the candidates capable of producing an outlier return. Bessemer does not eliminate committees or errors. Its structure tries to keep a strong, informed minority opinion alive long enough to write a check.
An investor studies a market, publishes a thesis and builds a network before a deal appears.
Partners can champion idiosyncratic investments instead of waiting for universal enthusiasm.
Memos, portfolio experience and visible misses inform the next decision without pretending certainty.
For a founder, the practical benefit is speed and clarity. A genuine internal champion matters more than a room full of polite maybes. Once invested, the firm can bring in functional experts who have scaled sales, marketing, people operations, finance and product teams. Atlas, its public knowledge platform, packages part of that expertise into founder playbooks. Bessemer's early-stage page includes courses and guides for demand generation, pricing, the first $1 million in annual recurring revenue and the climb from $1 million to $10 million.
The content serves two markets at once. Founders can use it without taking Bessemer's money. The firm, in turn, demonstrates how it thinks before asking for a board seat. In an industry where every investor promises a network, a published framework is inspectable evidence. A founder may disagree with it - disagreement is considerably more useful than a brochure.
03 / The product
One relationship, several pools of capital
Bessemer's core venture practice can invest while a company is still pre-revenue; the firm says 62 percent of its investments were made before revenue as of September 2025. It can continue through later rounds as the problem changes from proving demand to building an organization. The growth practice, formalized in 2019, concentrates on expansion-stage businesses. BVP Forge, introduced in 2021, moves further along the maturity curve.
Forge targets self-sustaining software and technology-enabled services companies seeking liquidity through majority or significant-minority investments. Its second fund closed at $1 billion in November 2025, following a $780 million inaugural vehicle. ForgeEdge adds a structured business-scaling program. The proposition is aimed at CEOs who need a private-equity transaction but still value access to a venture network trained on product cycles and emerging technology.
The business model is the familiar investment-management bargain. Limited partners commit capital to funds. Bessemer deploys it, supports the companies and earns management fees plus a share of profits when investments produce liquidity. The exact economics differ by vehicle and are private. What Bessemer has added is coverage: a founder can encounter the firm before revenue, during rapid scale or when an established business considers a strategic owner.
Its expertise is broad but not shapeless. Cloud and software are central. The annual Cloud 100, produced with Forbes and Salesforce Ventures, has become both a ranking and a data set. In 2025 its 100 companies surpassed $1.1 trillion in aggregate value. Bessemer reported that AI companies on the list reached $100 million in annual recurring revenue in an average 5.7 years, compared with 7.5 years for cloud companies overall. The numbers feed new theses in AI infrastructure, developer tools and vertical applications.
Healthcare, fintech, consumer technology, cybersecurity, marketplaces and defense widen the aperture. The second dedicated India fund, announced at $350 million in 2025, targets early-stage founders across AI-enabled services and software, fintech and insurance, digital health, consumer brands and security. Investment teams in North America, London, Bangalore and Tel Aviv give the global claim some physical substance.
The firm sells money. The harder product to copy is permission to think in decades - while keeping a record of yesterday's bad ideas.
04 / The edge
Why the mistakes stay on the wall
Bessemer competes with a formidable alphabet of multi-stage firms: Accel, Andreessen Horowitz, General Catalyst, Greylock, Index, Insight, Lightspeed, NEA and Sequoia among them. Capital itself is abundant at the moments when everyone wants the same company. Brand, sector access, price, partner chemistry and concrete operating help decide who wins. No historical anecdote can substitute for a founder liking the person who will attend the board meeting.
The Anti-Portfolio helps because it makes an abstract cultural claim visible. Consider the incentives. Successful investors are rewarded for narrating foresight. A public list of errors says outcomes do not retroactively make every judgment brilliant. It also tells a rejected founder something healthy: the investor may be wrong. That small redistribution of authority makes the relationship feel less ceremonial.
The page contains genuine clangers. Bessemer declined an early chance at FedEx after doubts about the air-freight market. It passed on Google. A partner heard about Airbnb and did not pursue it. Along the way there were less digital experiments too - a wig company, a french-fry business and a Hawaiian railroad. The archive does not prove superior returns, and humility can become marketing like anything else. Its value is behavioral: it lowers the reputational cost of recording a miss.
That is the stealable idea. A company does not need a century of capital to maintain an anti-roadmap: the hires not made, customers misunderstood, products killed too early and assumptions that survived too long. Record the contemporary reasoning. Revisit it without staging a trial. Make the lessons searchable. The result is not a shrine to failure; it is protection against a confident organization quietly rewriting its own history.
05 / The next turn
From software screens to operational systems
Bessemer's recent activity suggests AI is spreading from software categories into the physical and administrative systems beneath them. Its 2026 announcements include DriveNets for AI networking, Illoca for building design, Amperos for healthcare denial management, Dome Systems for agentic-enterprise operations, Upwind for cloud security and investments touching mobility, maritime systems and energy. These are not one neat sector. They share a belief that models become valuable when attached to expensive workflows, infrastructure or decisions.
For founders deciding whether Bessemer can help, the test is practical. Does a partner have a specific thesis about the market? Can the operating network introduce the executive, customer or benchmark required at the next stage? Is the proposed fund aligned with the company's maturity and desired control? Atlas is a free preview; the portfolio and Anti-Portfolio are reference checks hiding in plain sight.
Bessemer fits the market as a global, full-lifecycle technology investor with an adjacent private-equity arm. Its longevity is unusual, but longevity alone is not an advantage. The useful part is the system built around it: independent partners, written theses, multiple pools of capital and an institutional willingness to keep embarrassing evidence. Steel made the fortune. Memory is what keeps the machine from hardening.