Anthropic Walks Away From Proposed $6 Billion Acquisition of AI Startup Decart

Anthropic Walks Away From Proposed $6 Billion Acquisition of AI Startup Decart

Anthropic has decided not to pursue its proposed acquisition of artificial-intelligence startup Decart AI, walking away from a transaction that had been discussed at a valuation of around $6 billion after completing due diligence on the company.

The decision ends, at least for now, what could have become one of Anthropic's most significant strategic acquisitions and a major consolidation deal within the rapidly expanding AI infrastructure industry.

Anthropic had been evaluating Decart as it looked for technologies capable of improving the efficiency of the enormous computing infrastructure required to train and operate advanced artificial-intelligence models.

Decart develops technology designed to make AI computing hardware work more efficiently, potentially lowering the cost of:

AI model training,

inference,

and large-scale deployment.

The proposed transaction was therefore strategically significant for Anthropic, whose Claude models require substantial computing resources as adoption expands among consumers and enterprise customers.

The companies could still explore other forms of collaboration despite the acquisition talks ending.

Neither Anthropic nor Decart has publicly provided a detailed explanation for why the transaction was abandoned.

Anthropic Walks Away After Conducting Due Diligence

Anthropic had progressed far enough in its discussions with Decart to conduct:

due diligence.

Due diligence is a critical stage of an acquisition in which a potential buyer examines areas including:

technology,

finances,

intellectual property,

contracts,

customers,

employees,

legal exposure,

and business prospects.

After completing that process, Anthropic ultimately decided against proceeding with the acquisition.

The precise reason behind the decision hasn't been disclosed.

There is currently no confirmed indication that the deal collapsed because of:

valuation,

technology concerns,

financial considerations,

or findings during due diligence.

The decision should therefore be viewed as Anthropic choosing not to continue with the proposed transaction rather than evidence of a specific problem at Decart.

Proposed Deal Was Valued at Around $6 Billion

Earlier discussions had placed the potential acquisition value at approximately:

$6 billion.

That would have represented a substantial premium to Decart's valuation in its most recent funding round.

Decart raised approximately:

$300 million

in May 2026.

That financing valued the company at close to:

$4 billion.

A $6 billion acquisition would therefore have valued Decart roughly 50% above the level established only months earlier.

The company's investors include prominent technology and venture-capital names, including:

Nvidia,

Sequoia Capital,

Benchmark,

Radical Ventures,

Adobe Ventures,

Valor Equity Partners,

Atreides Management,

and Zeev Ventures.

The quality of that investor base reflects the strategic interest surrounding technologies that can improve AI computing efficiency.

Why Anthropic Was Interested in Decart

The strategic rationale behind the proposed acquisition was closely connected to one of the AI industry's biggest challenges:

compute economics.

Developing increasingly capable AI systems requires enormous amounts of computing power.

Large AI laboratories need thousands of sophisticated accelerators operating across massive data-centre clusters to:

train models,

serve customer queries,

run reasoning workloads,

conduct research,

and develop future generations of AI systems.

That creates enormous infrastructure expenses.

Reducing the amount of computing capacity required for each workload can therefore generate significant economic value.

Decart's technology is designed to improve how efficiently AI workloads use existing hardware.

For a company operating at Anthropic's scale, even relatively small improvements in compute efficiency could potentially translate into substantial savings.

Decart Focuses on Making AI Chips More Efficient

Decart occupies an increasingly important part of the artificial-intelligence technology stack.

Rather than simply developing another general-purpose chatbot, the company has worked on technology intended to extract greater performance from computing hardware.

The fundamental objective is straightforward:

make every AI chip perform more useful work.

That matters because advanced AI accelerators remain among the most valuable and strategically important resources in the technology industry.

AI laboratories are spending enormous amounts of capital securing:

GPUs,

AI accelerators,

networking systems,

memory,

data-centre capacity,

and electricity.

Improving the utilisation of those resources can therefore reduce the effective cost of developing and operating AI services.

For Anthropic, acquiring Decart could have provided direct ownership of technology capable of improving the economics of its infrastructure.

AI Inference Costs Are Becoming Strategically Important

Much of the early AI investment cycle focused on:

training.

Training advanced foundation models requires huge clusters of accelerators operating for extended periods.

But as AI services gain millions of users, another expense becomes increasingly important:

inference.

Inference occurs whenever a trained model generates a response or performs a task for a user.

Every Claude interaction therefore consumes computing resources.

As Anthropic expands into:

enterprise AI,

software development,

research,

agents,

business automation,

and consumer applications,

the total number of inference workloads can increase dramatically.

The economics of serving those requests becomes crucial.

A company that can generate more AI output from the same computing infrastructure can potentially improve:

gross margins,

capacity utilisation,

customer economics,

and scalability.

That explains why infrastructure-optimisation technology has become strategically valuable.

Decart Has Also Developed Advanced AI Models

Decart's business extends beyond infrastructure optimisation.

The company has developed AI systems including:

Lucy

and:

Oasis.

Lucy is a live-video generation and editing technology capable of manipulating visual content in real time.

Such technology could eventually support applications across:

e-commerce,

virtual clothing,

entertainment,

live streaming,

advertising,

and interactive digital experiences.

Oasis represents another area of Decart's work around interactive AI-generated environments and simulation.

These technologies have helped give Decart a profile extending beyond infrastructure software into the emerging field of:

world models.

World models attempt to develop AI systems capable of understanding and generating dynamic environments rather than working primarily with text.

Anthropic's Main Interest Appeared to Be Infrastructure Efficiency

Despite Decart's work on world models, the strategic logic behind Anthropic's proposed acquisition appeared to be primarily:

computational efficiency.

Anthropic operates one of the world's largest families of commercial AI models through Claude.

As the company's customer base expands, infrastructure requirements can rise extremely quickly.

The company therefore has strong incentives to improve the efficiency of:

training,

inference,

hardware utilisation,

and model serving.

Buying Decart could have allowed Anthropic to integrate optimisation technology deeply into its own AI infrastructure.

Instead, Anthropic may now pursue those capabilities through:

internal development,

commercial partnerships,

licensing,

or collaboration.

Anthropic and Decart Could Still Collaborate

The collapse of the acquisition doesn't necessarily mean the relationship between the two companies has ended.

Anthropic and Decart could still pursue:

other forms of collaboration.

That could potentially allow Anthropic to gain access to Decart's technology without committing billions of dollars to purchasing the entire company.

A commercial partnership could offer several advantages.

Anthropic could potentially deploy Decart's optimisation tools while avoiding:

acquisition integration,

large upfront consideration,

ownership complexity,

and additional organisational expansion.

Decart, meanwhile, could remain independent while gaining one of the world's largest AI laboratories as a potential technology partner.

No specific collaboration has yet been publicly confirmed.

$6 Billion Deal Would Have Been Unusually Large for Anthropic

Anthropic has historically been relatively selective about major acquisitions.

A $6 billion purchase would therefore have represented a significant shift in its M&A strategy.

The proposed transaction would also have highlighted how rapidly valuations are increasing for companies developing strategically important AI infrastructure.

The AI industry's largest companies increasingly compete not only for:

models

and:

customers,

but also for the technologies required to operate those models economically.

That includes:

semiconductors,

data centres,

networking,

memory,

energy,

model optimisation,

and inference infrastructure.

As the industry matures, infrastructure efficiency may become just as important as raw model intelligence.

Anthropic Is Spending Heavily on Computing Capacity

Anthropic's interest in Decart comes as the company continues to spend heavily on the computing infrastructure required to develop and operate Claude.

Frontier AI development is extraordinarily capital intensive.

Companies competing at the leading edge must continually invest in:

new accelerator clusters,

data-centre capacity,

cloud infrastructure,

research,

model training,

and inference systems.

The challenge becomes even greater as models grow more sophisticated.

Advanced reasoning systems may consume substantially more computing resources than simpler AI applications because they can perform longer sequences of internal computation before delivering responses.

The industry therefore faces a fundamental economic question:

Can AI companies reduce the cost of intelligence quickly enough as usage expands?

Decart's technology addresses precisely that challenge.

Decision Comes as Anthropic Prepares for a Public Listing

The abandoned acquisition also comes during a particularly important period for Anthropic.

The company is preparing for a potential:

initial public offering.

A public listing would place significantly greater attention on Anthropic's:

revenue growth,

capital expenditure,

gross margins,

infrastructure commitments,

cash requirements,

and long-term profitability.

That could make capital-allocation discipline increasingly important.

Large acquisitions made immediately before a major public offering can attract substantial scrutiny from prospective shareholders.

Investors would want to understand whether a multibillion-dollar acquisition could produce enough:

technology advantage,

cost savings,

revenue opportunities,

or strategic protection

to justify its price.

Walking away from Decart demonstrates that even strategically attractive technology may not necessarily result in an acquisition.

Compute Economics Could Become a Major AI Competitive Battleground

The proposed Decart transaction reveals a broader transformation taking place across artificial intelligence.

The first stage of the generative-AI competition was largely defined by:

model capability.

Companies competed to demonstrate:

better reasoning,

larger context windows,

stronger coding performance,

multimodal capabilities,

and improved benchmark scores.

The next stage increasingly includes another question:

How cheaply can those capabilities be delivered?

AI companies serving hundreds of millions of requests can't ignore infrastructure economics.

A technically superior model may still face commercial difficulties if operating it is significantly more expensive than competing systems.

Efficiency therefore becomes a strategic advantage.

Companies capable of reducing the cost per AI task can potentially:

lower prices,

increase margins,

serve more users,

and deploy more computationally intensive models.

AI Infrastructure Startups Are Becoming Valuable Acquisition Targets

Decart's rapid rise in valuation illustrates how strategically important infrastructure startups have become.

The AI ecosystem increasingly includes specialised companies focused on:

model optimisation,

inference engines,

GPU utilisation,

distributed computing,

AI networking,

data-centre software,

memory efficiency,

and workload orchestration.

These companies may not always attract the same consumer recognition as major chatbot platforms.

But their technology can have enormous economic value.

If an optimisation platform reduces infrastructure requirements across a massive AI deployment, the resulting savings can potentially reach billions of dollars.

That makes infrastructure startups logical acquisition or partnership targets for frontier AI laboratories.

Anthropic Can Still Pursue Efficiency Without Owning Decart

Walking away from the acquisition doesn't eliminate Anthropic's need to improve compute efficiency.

The company has several alternative routes.

It can develop optimisation technologies internally.

It can partner with infrastructure startups.

It can negotiate deeper arrangements with semiconductor manufacturers and cloud providers.

It can improve model architectures.

And it can use software techniques to reduce the computing resources required for individual workloads.

The decision therefore changes:

how Anthropic may pursue efficiency,

rather than eliminating the strategic objective itself.

Conclusion

Anthropic's decision to abandon its proposed $6 billion acquisition of Decart AI brings an end to what could have become one of the company's largest strategic transactions.

Anthropic conducted due diligence before deciding not to proceed, while the specific reason behind the decision remains undisclosed.

The deal nevertheless reveals an important shift in the artificial-intelligence industry.

As AI models become more powerful and usage expands, the competitive battle is increasingly moving beyond model intelligence toward:

compute efficiency, infrastructure economics and the cost of serving AI at global scale.

Decart's technology was attractive because it potentially addressed that challenge by helping AI hardware perform more efficiently.

Although the acquisition has fallen through, the two companies could still explore other forms of collaboration.

For Anthropic, improving infrastructure efficiency will remain strategically important as Claude adoption grows and the company moves toward a potential public listing.