Cognition AI Raises $2 Billion at $48 Billion Valuation in Andreessen Horowitz- and Accel-Backed

Cognition AI Raises $2 Billion at $48 Billion Valuation in Andreessen Horowitz- and Accel-Backed Round

Artificial intelligence coding startup Cognition has raised more than $2 billion in a Series E funding round at a $48 billion valuation, nearly doubling its valuation in just a few months as demand for autonomous software-engineering agents accelerates across large enterprises.

The financing was led by new investors Andreessen Horowitz and Accel, with existing backers Founders Fund, General Catalyst and Avenir also participating.

The round comes only months after Cognition raised more than $1 billion at a $26 billion post-money valuation in May 2026.

Since that previous financing, Cognition says its run-rate revenue has climbed from:

$492 million

to:

nearly $900 million.

That rapid revenue expansion has helped support a fresh valuation of:

$48 billion,

placing Cognition among the world's most valuable privately held artificial intelligence software companies.

The company is best known for Devin, an autonomous AI software-engineering agent designed to plan, write, test, debug and deploy code with limited human intervention.

Cognition Raises More Than $2 Billion in Series E

Cognition announced the new financing on:

September 8, 2026.

The company said it had raised:

more than $2 billion

in its:

Series E.

The transaction values Cognition at:

$48 billion.

This represents one of the largest private funding rounds completed by an AI application company focused specifically on software engineering.

Andreessen Horowitz and Accel Lead the Round

The Series E was led by:

Andreessen Horowitz

and:

Accel.

Both are new investors in Cognition's latest financing.

Their participation adds two of Silicon Valley's most prominent venture-capital firms to a shareholder base that already contains several major technology investors.

Existing investors participating in the round included:

Founders Fund,

General Catalyst,

and:

Avenir.

Investor Group Extends Across Major Venture Firms

The financing also attracted a broad group of institutional investors.

Participants disclosed by Cognition include firms such as:

Benchmark,

Bessemer Venture Partners,

Kleiner Perkins,

Greylock,

Lightspeed,

Altimeter,

Bond Capital,

Meritech,

T. Rowe Price,

Lux Capital,

8VC,

D1,

DST,

Bain Capital Ventures,

Alpha Wave,

Alkeon,

and NVIDIA.

The breadth of participation highlights the intense investor demand surrounding AI software companies that are demonstrating large and rapidly growing commercial revenue.

Valuation Nearly Doubles From May

Cognition's valuation has risen exceptionally quickly.

In:

May 2026,

the company raised more than:

$1 billion

at a:

$26 billion post-money valuation.

The latest $48 billion valuation represents an increase of approximately:

85%

in only a few months.

The scale of that repricing reflects the pace at which Cognition says customers are adopting its software-engineering products.

Revenue Run Rate Approaches $900 Million

One of the most important figures supporting the new valuation is:

nearly $900 million in run-rate revenue.

Cognition said its revenue run rate was approximately:

$492 million

at the time of its May funding round.

That means the company has added roughly:

$400 million

of annualised revenue run rate in only a few months.

The increase represents growth of more than:

80%.

Few enterprise-software companies reach this revenue scale so quickly after founding.

Cognition Was Founded Only in 2024

Cognition was founded in:

2024.

The company was created around the idea that software engineers could increasingly act as:

architects and supervisors

while autonomous AI agents perform more of the underlying implementation work.

CEO:

Scott Wu

and his team built Cognition around this thesis.

The company's rapid rise demonstrates how quickly AI-native software businesses can scale when they gain enterprise adoption.

Devin Is Cognition’s Flagship Product

Cognition's best-known product is:

Devin.

The company describes Devin as an:

AI software engineer.

Unlike conventional coding assistants that mainly suggest lines of code, Devin is designed to work on complete engineering tasks.

It can:

analyse requirements,

create a plan,

write code,

run tests,

debug failures,

investigate issues,

and implement changes.

The objective is to move AI coding from:

code completion

toward:

task completion.

Autonomous Coding Is Becoming a Major Software Category

Traditional coding assistants have generally operated alongside developers.

A programmer writes code while AI suggests:

functions,

lines,

or fixes.

Agentic coding systems attempt to take this much further.

A developer can assign:

a bug,

a feature,

a migration,

or an investigation

to an AI agent and allow it to work independently for an extended period.

This changes the economics of software engineering because one human engineer could potentially supervise several AI agents simultaneously.

Cognition Sees Engineers Becoming Architects

Cognition's long-term vision is that software engineers will increasingly spend their time defining:

goals,

architecture,

product priorities,

and technical constraints.

AI agents can then handle more of the repetitive implementation work.

This does not necessarily eliminate human engineers.

Instead, it changes their role.

The most valuable engineering skill could shift from writing every line of code toward:

designing systems

and:

orchestrating autonomous agents.

Devin Can Take More Proactive Actions

Cognition has been expanding Devin beyond individual coding assignments.

The platform now includes capabilities such as:

Devin Auto-Triage,

which can perform an initial investigation of incidents.

It also includes:

Devin Security Swarm,

designed to identify and triage software vulnerabilities.

These features move Devin closer to continuous software operations rather than purely developer-requested tasks.

Automations Connect Devin to Enterprise Tools

Cognition has also introduced:

Devin Automations.

These allow companies to configure work to begin when events occur inside software such as:

Slack,

GitHub,

Linear,

and other enterprise systems.

This reduces the need for a developer to manually start every AI session.

A future software-engineering environment could therefore contain agents continuously responding to:

bugs,

security events,

code changes,

and operational workflows.

NVIDIA Is Among Cognition’s Customers

Cognition says Devin is being used by major organisations across several industries.

Its customers include:

NVIDIA

for engineering work related to chip design.

The company also cites:

GE Aerospace

in aviation,

Citi

in financial services,

Mercedes-Benz

in automotive,

and:

Modal

in AI infrastructure.

These customers demonstrate Cognition's focus on enterprise adoption rather than only individual programmers.

Large Enterprises Are Increasing AI Coding Adoption

Enterprise software engineering represents a particularly large market.

Large corporations employ thousands of developers working across:

internal applications,

customer platforms,

data systems,

security,

infrastructure,

and legacy technology.

Even modest improvements in developer productivity can therefore create substantial economic value.

This is one reason enterprises are experimenting aggressively with AI coding tools.

Software Demand Continues to Exceed Engineering Capacity

Cognition's investment thesis rests on a simple constraint:

organisations want to build more software than their engineering teams have capacity to deliver.

Companies routinely postpone:

new products,

system upgrades,

security improvements,

and internal automation

because engineers are already occupied with other priorities.

Autonomous coding agents could increase the amount of software an existing engineering workforce can produce.

AI Agents Could Address Software Backlogs

Most large organisations maintain substantial technical backlogs.

These can include:

minor bugs,

maintenance,

library upgrades,

security patches,

tests,

and internal tooling.

Many tasks are economically valuable but not urgent enough to justify scarce engineering time.

AI agents could make more of these tasks economically viable by reducing the amount of human attention required.

This is one of the strongest potential use cases for products such as Devin.

Cognition Acquired Windsurf in 2025

Cognition significantly expanded its product portfolio in:

July 2025

when it acquired:

Windsurf.

Windsurf had developed an agentic integrated development environment, or IDE, for programmers.

The acquisition included:

Windsurf's intellectual property,

products,

brand,

and business operations.

It gave Cognition a second major interface for AI-powered software development.

Windsurf Complements Devin

Devin and Windsurf address different developer workflows.

Windsurf is designed for situations where the engineer remains directly involved in:

writing,

reviewing,

and modifying code.

Devin is designed for work that can be:

delegated.

Cognition therefore covers two major categories of AI coding:

AI-assisted IDE development

and:

autonomous agents.

The combination allows customers to choose how much control they want to retain for each engineering task.

Windsurf Acquisition Accelerated Revenue

The Windsurf transaction materially increased Cognition's commercial scale.

Before the acquisition, Cognition said Devin's annual recurring revenue had grown from:

$1 million in September 2024

to:

$73 million by June 2025.

Acquiring Windsurf more than doubled the combined company's recurring revenue at the time.

It also brought an established enterprise sales organisation and additional customers.

Customer Overlap Was Limited

Cognition previously said there was:

less than 5% overlap

between Devin and Windsurf enterprise customers before the acquisition.

That made the transaction strategically attractive.

Instead of acquiring a business serving largely the same customer base, Cognition gained access to a substantially different group of enterprises.

This created cross-selling opportunities for both product lines.

Enterprise ARR Accelerated After Windsurf Deal

Following the acquisition, Cognition said combined enterprise annual recurring revenue increased by more than:

30% in seven weeks.

That early acceleration provided evidence that the combination of Devin and Windsurf could generate commercial synergies.

The subsequent rise toward a $900 million revenue run rate suggests enterprise adoption continued to accelerate into 2026.

Independent AI Coding Company Thesis Is Strengthening

Cognition's growth challenges an earlier assumption that standalone AI coding startups would eventually be displaced by large foundation-model providers.

Companies such as:

OpenAI,

Anthropic,

and Google

all offer coding capabilities.

Microsoft also has a major position through:

GitHub Copilot.

Yet Cognition's growth suggests there may still be substantial room for independent companies focused specifically on software-engineering workflows.

Cognition Wants to Remain Model Independent

One important element of Cognition's strategy is its position as an:

independent agent lab.

The company says it wants to be able to select and combine whichever underlying AI models are best suited to particular engineering tasks.

That means customers are not necessarily tied to one foundation-model provider.

Model independence could become strategically valuable if different models excel at different categories of engineering work.

AI Model Competition Can Benefit Application Companies

Foundation-model competition is becoming intense.

OpenAI, Anthropic, Google, Meta and others are continually releasing improved systems.

An application company that can switch among models may benefit from:

better performance,

lower costs,

and reduced dependence on a single supplier.

This could allow Cognition to focus on the engineering workflow rather than competing directly to build the largest general-purpose model.

Cognition Also Develops Its Own Models

Model independence does not mean Cognition avoids proprietary research.

The company also develops its own AI systems designed specifically for software engineering.

A specialised model can potentially optimise for:

coding speed,

tool use,

software repositories,

and engineering tasks.

Cognition can therefore combine internally developed technology with external foundation models.

Global Expansion Has Accelerated

Cognition has expanded its physical presence significantly over the past year.

The company has opened offices in:

Washington, D.C.,

Tokyo,

Singapore,

London,

São Paulo,

and Madrid.

These locations complement existing hubs in:

San Francisco,

New York,

and Austin.

The expansion reflects increasing enterprise demand across multiple regions.

Local Enterprise Presence Matters

Large enterprise software contracts often require close customer relationships.

Customers may want help with:

implementation,

security reviews,

workflow design,

procurement,

and organisational change.

Having local teams can make these engagements easier.

Cognition's international office expansion suggests the company expects enterprise sales and deployment to remain a significant part of its growth strategy.

New Funding Gives Cognition Significant Expansion Capacity

More than $2 billion in fresh funding provides Cognition with substantial resources.

The company can invest across areas including:

AI research,

computing infrastructure,

product engineering,

enterprise sales,

customer support,

and international expansion.

The capital can also provide flexibility for:

acquisitions

or:

strategic investments.

Cognition has not disclosed a detailed allocation for every dollar raised.

AI Software Requires Significant Compute

Although Cognition is an application-layer software company, autonomous agents still consume substantial computing resources.

Agents may need to:

analyse large codebases,

run repeated model calls,

execute tests,

and operate for long periods.

As customers assign larger tasks to agents, compute requirements can rise significantly.

Part of Cognition's capital needs therefore comes from supporting high-volume inference and product development.

Usage-Based Economics Could Become Important

AI software differs from traditional SaaS because serving each additional user can carry meaningful compute cost.

A conventional SaaS application may have relatively low incremental costs once infrastructure is built.

AI agents may consume significant resources every time they perform complex tasks.

This means Cognition must balance:

usage growth,

pricing,

and inference costs.

Efficient compute economics will be important as the business scales.

Revenue Growth Is Exceptional but Valuation Is Also High

A $48 billion valuation against nearly $900 million of run-rate revenue implies a valuation of more than:

50 times annualised revenue.

That is an unusually high multiple even within enterprise software.

Investors are therefore pricing in substantial future growth.

To justify that valuation over time, Cognition will need to continue expanding revenue rapidly while demonstrating increasingly attractive long-term margins.

AI Funding Market Remains Highly Aggressive

The Cognition round is part of a broader surge in artificial intelligence investment.

Venture funds and institutional investors have been willing to commit billions of dollars to companies that could become important platforms in:

AI models,

infrastructure,

coding,

robotics,

and enterprise applications.

Capital has increasingly concentrated around a relatively small group of companies demonstrating exceptional growth.

Cognition has now joined that upper tier.

Coding Is One of AI’s Most Competitive Markets

Software engineering is also one of the most competitive areas of generative AI.

Developers can choose among products from:

Cognition,

OpenAI,

Anthropic,

Google,

Microsoft,

and numerous independent startups.

New tools are launched frequently.

Product quality can change rapidly because underlying AI models improve every few months.

This creates both enormous opportunity and significant competitive risk.

Enterprises May Use Multiple Coding Tools

The market may not become winner-take-all.

Different engineering teams may prefer different products depending on:

programming language,

workflow,

security requirements,

and integration.

Large companies may also deploy several AI coding systems.

Cognition therefore does not necessarily need to eliminate every competitor.

It needs to remain sufficiently differentiated to capture a large share of enterprise engineering spending.

Security Will Be Central to Enterprise Adoption

Software agents can access:

source code,

internal systems,

credentials,

and infrastructure.

That makes security a critical consideration.

Enterprises need confidence that autonomous systems will not:

leak code,

introduce vulnerabilities,

or make unauthorised changes.

Cognition's ability to provide strong controls around agents will be important to expanding adoption in regulated industries.

Human Review Remains Important

Despite advances in autonomy, AI-generated code still requires oversight.

Agents can make:

incorrect assumptions,

security mistakes,

or architectural errors.

Human engineers therefore remain important for:

review,

testing,

and high-level decision-making.

Cognition's own vision emphasises engineers acting as architects rather than disappearing entirely from the development process.

Productivity Gains Could Reshape Engineering Teams

If autonomous agents become consistently reliable, software teams could change significantly.

A smaller group of engineers might be able to manage:

more projects,

larger codebases,

and faster development cycles.

Companies could also allocate more resources to projects previously considered too expensive.

This would expand the total amount of software created rather than simply replacing existing engineering work.

Developer Roles Could Become More Strategic

The value of engineers may increasingly shift toward:

system architecture,

product understanding,

technical judgment,

and agent supervision.

Routine implementation work could become increasingly automated.

Developers capable of defining clear specifications and reviewing complex AI-generated systems may become particularly valuable.

This is similar to how earlier development tools raised abstraction levels without eliminating the need for skilled engineers.

New Round Raises Pressure on Cognition to Deliver

The $48 billion valuation also creates a high bar.

Cognition is now valued like a mature global technology company despite being founded only in 2024.

Investors will expect:

continued rapid growth,

strong customer retention,

enterprise expansion,

and durable technological differentiation.

Any significant slowdown could challenge the assumptions embedded in the valuation.

Revenue Quality Will Be Closely Watched

Run-rate revenue is an important growth metric, but investors will increasingly focus on:

contract duration,

customer retention,

gross margins,

and usage consistency.

High AI usage can generate strong revenue while also producing substantial compute costs.

Cognition's long-term economics will therefore depend on more than top-line growth alone.

Enterprise Expansion Can Improve Revenue Durability

Large enterprise contracts can provide more stable revenue than individual developer subscriptions.

Enterprises tend to integrate products deeply into:

workflows,

security systems,

and development processes.

Once deployed at scale, switching becomes more difficult.

Cognition's growing enterprise customer base may therefore help create stronger retention and longer-term revenue visibility.

Independent Agent Platform Could Become Strategic Asset

Cognition's largest strategic opportunity may be to become an independent orchestration layer for software engineering.

If developers use multiple AI models but rely on Cognition for:

task planning,

tool integration,

execution,

security,

and workflow management,

the company could occupy a powerful position above individual foundation models.

That would make its value less dependent on which company produces the best underlying model at any particular moment.

Conclusion

Cognition AI has raised more than $2 billion in a Series E funding round at a $48 billion valuation, with new investors Andreessen Horowitz and Accel leading the financing alongside existing backers Founders Fund, General Catalyst and Avenir.

The round comes only a few months after Cognition raised more than $1 billion at a $26 billion valuation in May 2026, meaning the company's private-market value has risen by roughly 85% in a remarkably short period.

The rapid repricing is supported by equally rapid commercial growth. Cognition says its run-rate revenue has increased from $492 million in May to nearly $900 million, driven by expanding enterprise adoption of its Devin autonomous coding agent and the broader product portfolio created through its acquisition of Windsurf.

For investors, Cognition represents a bet that artificial intelligence will fundamentally reshape software engineering by allowing developers to delegate larger amounts of implementation work to autonomous agents.

For Cognition, however, the new valuation also raises expectations. The company must now prove that its extraordinary revenue growth can translate into a durable enterprise-software business with strong retention, attractive margins and a defensible position against OpenAI, Anthropic, Microsoft, Google and other AI coding competitors.

If autonomous software engineering becomes a foundational layer of enterprise technology, Cognition's combination of Devin, Windsurf, model independence and rapidly expanding enterprise adoption could place it among the most strategically important companies in the emerging AI software stack.