June 4 (Reuters) - Canadian AI startup Cohere has raised

$450 million in funding from returning investors such as Nvidia ( NVDA )

and Salesforce Ventures, as well as new

investors including Cisco ( CSCO ) and Canadian pension fund PSP

Investments, according to a source familiar the matter.

This concludes the first tranche of Cohere's monthslong

fundraising efforts, while the company is still in talks to

raise more in the same round at $5 billion valuation, added the

source, speaking on condition of anonymity.

The funding marks a jump in valuation from Cohere's last

private raise, when it was valued at $2.2 billion from investors

including Inovia Capital last June.

The generative AI company, which makes money by selling its

models and applications to enterprises with a emphasis on data

privacy, generated $35 million in annualized revenue by the end

of March, up from $13 million last year, the source added.

Cohere declined to comment. Nvidia ( NVDA ) and Salesforce ( CRM ) did not

immediately reply to requests for comment.

Cohere was set out to raise between $500 million to $1

billion, Reuters previously reported. It competes with OpenAI,

Anthropic and Mistral, which have also raised billions of

dollars of strategic investors such as Microsoft, Google and

Amazon.

Foundation model AI companies have been racing to raise

capital to fund the expensive development of AI models that

require huge amounts of computing power and top industry talent.

One of the most high-profile Canadian startups, Cohere is

likely to benefit from the Canadian government's plan to invest

C$2.4 billion ($1.77 billion) to fund compute and AI research

for homegrown AI companies.

Founded in 2019, Toronto-headquartered Cohere builds large

language models - software systems that are trained on large

amounts of data and can generate text. Unlike OpenAI's tie-up

with Microsoft, it has steered clear of exclusive deals with

cloud providers, despite being backed by Oracle.

While AI startups largely remain money-losing given the

training cost, every company is trying to sell its technology to

the biggest companies that can pay for it for boosted

productivity. Showing revenue growth is also critical in

deciding if they could raise more capital.

The funding craze of AI startups is already slowing down,

especially on the early stage side. Venture capital deal value

for pre-seed and seed-stage AI deals dropped by 76% in the first

quarter this year to $122.9 million, from its peak of $517.7

million in Q3 2023, according to PitchBook data.