Is Iren comparable to Nebius?
Are NBIS 0.00%↑ and IREN 0.00%↑ comparable?
Iren's zone of genius is building data center infrastructure for energy intensive activities (Bitcoin mining, HPC) that is cost efficient. That's definitey way more than a Bitcoin miner!
Nebius provide a full stack solution for AI, usually renting external data centers (they own one in Finland), while Iren owns data centers. So, they do operate in the same industry, but Nebius could potentially be Iren's customer.
Think of it this way: the main competitors for Nebius are both GPU Clouds and Hyperscalers, while for Iren would be other data center operators and bitcoin miners. So, they are comparable to an extent, but they operate in different parts of the value chain.
Also, NBIS has other business lines: Toloka (partner for AI training labeling), Tripleten (edtech), Avride (autonomous self-driving cars), and a 28% stake in Clickhouse.
Is IREN 0.00%↑ only a Bitcoin miner?
Well, yes and no. Iren has bootstrapped operations through Bitcoin mining and has become the most efficient Bitcoin miner in the industry in terms of variable cost ($40k) per Bitcoin. Also, the bulk of revenues (>90%) comes from Bitcoin mining.
However, Iren does not hold any Bitcoin in their balance. This is intentional. In words of Dan Roberts, Iren Co-CEO, holding Bitcoin would effectively dilute shareholders as they need capital for growth. This is reflective of Iren's different approach in comparison to the mining industry. Also, Iren offers AI Cloud Services (<3% of current revenue) and decided to develop a new 75 MW liquid-cooled data center at Childress for AI/HPC designed to support NVIDIA Blackwell. This is a move one step closer to AI/HPC services.
The main critic I've read regarding this is that Iren does not have secured revenues from HPC/AI. I believe that critic has valid merits considering low revenues from current AI Cloud Services, though, sometimes companies only they need is time. Market rewards patience.
One thing I would add is that Dan Roberts could do better at shaping the narrative. I've seen Arkady Volozh more active (hence driving eyeballs) in the last 3-6 months than Dan in the last 3-5 years. This is especially important considering CAPEX needs (for both companies). e.g. IREN 0.00%↑ said they expected to raise $1 billion ATM to continue expansion and this was a self-inflicted wound, driving stock price down. I love Dan Roberts’ current approach focusing on "revenues-first", but having a strong stock will help on funding the company's growth.
Is NBIS 0.00%↑ financially healthier than IREN 0.00%↑?
Cash perspective: NBIS 0.00%↑ has $2 billion in cash while IREN 0.00%↑ has no cash (invested cash in balance in Jan/25 in miners to expand current capacity) and convertible debt.
Operational perspective: NBIS 0.00%↑ is burning cash (breakeven late 2025) and IREN 0.00%↑ generates $250m yearly EBITDA (>$600m late 2025). Iren is already a profitable business, while Nebius is still in an early phase.
Recap
NBIS 0.00%↑ is still in an early phase and is a cash burning business. NBIS 0.00%↑ has a $2 billion war chest and provides a full stack solution for AI, plus additional businesses where the most attractive is Avride. I see complementary businesses as optionality.
While NBIS 0.00%↑ has provided guidance of $1bn ARR by 2025, right now investing in NBIS is still investing in the team + thesis. As they execute during 2025, we will be able to evaluate more. I gravitate more to proven businesses vs. guidances, but following NBIS 0.00%↑ closely.
IREN 0.00%↑ is on a growth phase and produces $250m of yearly EBITDA ($640m guidance for 2025) with a poven track record on data center infrastructure and especially Bitcoin mining, and moderate experience in AI cloud service.
While IREN 0.00%↑ is intentionally building a 75MW liquid-cooled data center designed specifically for AI/HPC, we still do not have a proven track record of this team being able to scale this. I believe the company is currently undervalued and see AI/HPC play as a positive optionality.