Crypto miner Hive CEO anticipates $100M annual revenue from GPU computing
Hive Digital Applied sciences has ambitions to exponentially develop revenues derived from its graphics processing unit (GPU) card capabilities to $100 million yearly.
The agency is in talks with firms and marketplaces because it strikes to capitalize on — and strengthen — the capabilities of its machines previously used to mine ether (ETH), Hive CEO Aydin Kilic instructed Blockworks.
Hive’s identify change — from Hive Blockchain Applied sciences to Hive Digital Applied sciences — took impact Wednesday. The thought: to mirror a newfound give attention to the expansion potential of synthetic intelligence (AI).
The Vancouver-based agency, beforehand one of many largest publicly traded miners of ether, mined 7,675 ETH throughout 2022’s second quarter — simply earlier than Ethereum’s swap to an underlying proof-of-stake consensus mannequin.
The corporate has a fleet of 38,000 Nvidia graphics processing unit (GPU) playing cards, in response to a Wednesday submitting, presenting the items as a brand new income supply.
“Now that Ethereum has gone away, we’ve a brand new lever to experiment — or play — with,” Kilic stated.
The corporate in February made plans to launch its Hive Efficiency Cloud enterprise within the second quarter, including its proof-of-concept to make the most of its fleet of GPUs, which produced annual income on a run-rate foundation over $1 million.
Finally, Kilic stated, it might herald $100 million yearly for Hive.
“We’ve now confirmed this idea…and so we’re seeking to scale this enterprise,” he stated. “I’m hoping we might hit one thing like $15 million a 12 months income this 12 months alone, but it surely may very well be greater faster.”
The street to scaling
Hive’s GPUs help Tensor Core and Ray Tracing core structure, enabling the execution of AI and high-performance computing duties.
“The following factor we have to do is populate them in additional strong servers which have higher [central processing unit] clock speeds, extra threads, extra storage,” Kilic stated.
The corporate’s buy of $6 million of SuperMicro servers, holding 10 GPUs and that includes 7.6 terabytes (TB) of storage, are set to reach within the subsequent month.
They are going to initially run out of Hive’s datacenter in Sweden — with discussions underway round a North American enlargement — and so they’ve been positioned as providing extra computing energy and storage for Hive.
Hive hopes the servers will assist it attain $15 million in annualized income this 12 months.
“That’s like a four-month [return on investment], which is manner higher than crypto mining proper now,” he stated.“We’re getting clusters and clusters of those servers. After they’re clustered collectively, you may have huge computing assets.”
One solution to understand income from the Nvidia GPUs is to listing them for lease on a market.
The opposite route — through Hive Cloud — is working with firms that may want weeks or months of processing energy for rendering or different duties.
Hive is presently speaking with 4 marketplaces and 4 firms, Kilic stated, declining to specify them.
Use circumstances to help
Hive has cited demand for ChatGPT, medical analysis, machine studying and rendering as causes for its bullishness on the expansion of its computing enterprise.
ChatGPT — a pure language processing instrument launched by OpenAI in November — reached an estimated 100 million customers months after its launch.
One other rising know-how is zero-knowledge (ZK) proof, a cryptographic protocol permitting a celebration to show the validity of a press release with out disclosing any extra info.
A major impediment in utilizing ZK proof for on-line purposes is the efficiency overhead of its proof technology, in response to the Affiliation for Computing Equipment.
Kilic stated this space is just not but one thing that impacts day-to-day crypto customers, however added that Hive has begun researching it.
“ZK proofs, I believe, can be an integral side of the following technology of scaling layer-1s, however you want the confluence of a number of areas,” the CEO stated. “You want the computing energy, you want the algorithms that drive these computing powers and then you definately want the appliance layer the place there are initiatives the place there are many customers.”