3 E Network Secures 9 MVA Power Deal for Finland AI Hub [2026]

A small Nasdaq-listed company most AI investors have never heard of just locked in the power supply for a new AI compute center in southeastern Finland, and the filing behind it tells a bigger story than the deal itself. On October 13, 2026, 3 E Network Technology Group Limited (NASDAQ:MASK) confirmed that its Finnish subsidiary, Aurora Core Technology Oy, signed a power distribution network connection agreement with Jarvi-Suomen Energia Oy, the local distribution system operator serving the Mikkeli region. The agreement locks in 9,000 kVA, roughly 9 MVA, of ordered capacity on the 20 kV medium-voltage grid, enough to begin feeding what the company describes as its planned Mikkeli AI compute center.

The news moved fast through financial media on October 19, 2026, picked up by Quiver Quantitative, the Akron Beacon Journal’s wire feed, TradingView, Konsulteer, guampdn.com and Telecompaper, all citing the same underlying announcement. Investing.com and StockTitan carried the fuller version of the release, and Hong Kong based futunn.com ran a Chinese-language flash alongside it. For a company that trades far outside the Nvidia-and-hyperscaler spotlight, that is a wide and fast pickup, and it says something about how hungry markets are right now for any concrete signal that AI infrastructure buildout is still accelerating even as compute costs climb.

·

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

What 3 E Network Actually Signed

Strip away the headline shorthand and the underlying document is narrower than it sounds. Reports describe the deal as a power distribution network connection agreement, not a long-term supply contract or a tariff lock-in. In plain terms, Aurora Core Technology Oy has secured the right to draw up to 9 MVA from the regional grid operated by Jarvi-Suomen Energia Oy, at a 20 kV connection point, once the physical infrastructure is built out. That is the plumbing step that has to happen before a data center can legally pull power at scale, and it typically follows months of site surveys, grid capacity studies and permitting with the local utility.

Mikkeli sits in Finland’s Lakeland region, a few hours northeast of Helsinki, an area better known for timber, lake tourism and small manufacturing than for hyperscale computing. That is itself part of the pattern this year: AI compute buildout is moving into secondary grid regions where power headroom still exists, rather than clustering exclusively around established hubs. Finland has leaned into this positioning for years, marketing its cool climate, stable grid and renewable generation mix as natural advantages for energy-intensive data centers, and 3 E Network’s filing suggests that pitch is still landing with smaller, faster-moving operators even as the giants chase gigawatt-scale sites elsewhere.

Company statements say the planned facility will run on what is described as NVIDIA’s Vera Rubin architecture, the next generation GPU platform set to succeed the Blackwell and Blackwell Ultra lines that currently anchor most AI training clusters. If that plan holds, Mikkeli would be one of the earlier announced sites outside the usual hyperscaler roster to publicly tie itself to Vera Rubin hardware, though the company has not disclosed a GPU order size, a vendor contract, or a delivery schedule. Readers should treat the architecture detail as the company’s stated plan rather than a confirmed hardware order.

The Infrastructure Build Behind the Power Deal

Securing grid capacity is only the first domino. According to the company’s own disclosures, engineering work to integrate the external grid connection with the site’s planned internal infrastructure has already begun, alongside supply-chain coordination for the mechanical, electrical and plumbing systems that every modern AI data center needs. The planned scope includes liquid-cooling systems built for GPU racks that draw far more heat per rack than legacy air-cooled servers, interconnect infrastructure to move data between compute nodes at the bandwidth that large model training demands, step-down substations to convert the 20 kV feed into usable voltages, and backup power plus uninterruptible power supply systems to keep training runs from stalling during grid interruptions.

None of that is unusual for a facility of this scale, but the sequencing matters. Companies that announce GPU orders before locking grid capacity frequently end up with idle hardware sitting in warehouses waiting for power, a problem that has dogged several AI infrastructure projects across the US and Europe over the past two years as utilities struggled to keep pace with data center demand. By securing the 9 MVA connection first and only then layering in the mechanical and electrical build-out, 3 E Network appears to be sequencing the project in the order that actually avoids stranded capital, a detail that matters more to infrastructure investors than the raw megawatt number does.

StockTitan separately reported that 3 E Network had already deployed computational fluid dynamics simulation and digital twin technologies for the Mikkeli site, describing a software-defined engineering approach meant to model airflow, thermal load and rack layout before physical construction begins. That detail, disclosed in mid-September 2026 and referenced again in StockTitan’s October coverage, suggests the company is trying to de-risk the cooling design in software before pouring concrete, which is a increasingly common practice among AI data center developers trying to avoid costly retrofits once liquid-cooling loops are already in the ground.

Why 9 MVA Is a Meaningful but Modest Number

Nine megavolt-amperes is enough to matter locally and small enough to clear a regional grid operator’s books without years of transmission upgrades. For context on scale, a single rack of next-generation GPU servers running dense AI training workloads can draw well into the tens of kilowatts, meaning a 9 MVA connection could plausibly support a meaningful cluster of racks, though the company has not published a target rack count or an expected GPU unit total for Mikkeli. That is a very different order of magnitude from the gigawatt-scale campuses hyperscalers like Microsoft, Google and Meta have announced in the US, where some single sites now carry grid interconnection requests measured in the hundreds of megawatts or more.

The modesty of the number is arguably the point. Utilities across Europe and North America have grown increasingly cautious about large data center interconnection requests, worried about grid stability and the risk of stranded transmission investment if a project stalls. A 9 MVA ask is small enough that Jarvi-Suomen Energia Oy could plausibly accommodate it on existing medium-voltage infrastructure without the multi-year queue times that have delayed gigawatt-class projects elsewhere. That gives 3 E Network a faster path to actually energizing a site, even if the resulting facility will be a fraction of the size of a flagship hyperscaler campus.

It is worth comparing this move against other AI infrastructure power stories that have surfaced in Finland and the broader Nordic region this year. Finland’s cold climate, hydro and nuclear-heavy grid mix, and political stability have made it a recurring destination for data center siting discussions, though most of the headline-grabbing announcements in the region have come from larger players securing hundreds of megawatts at a time. 3 E Network’s 9 MVA deal is a reminder that smaller, Nasdaq-listed challengers are also trying to grab a foothold in the same geography, betting that even modest-scale regional sites can find customers in an AI compute market where demand has consistently outstripped supply.

3 E Network Technology Group: Who Is Behind the Filing

3 E Network Technology Group Limited trades on Nasdaq under the ticker MASK, and its Finland push runs through the newly visible subsidiary Aurora Core Technology Oy. MarketBeat’s news tracker flagged the Mikkeli announcement as among the company’s most-followed stories this week, a sign that investor attention on the stock has picked up alongside the Finland plan. StockTitan’s coverage has also tracked separate company disclosures, including a reported CEO-linked private placement, that point to a leadership team actively building out capital alongside the infrastructure plan, though neither the size of that placement nor any specific financing terms for the Mikkeli project itself have been disclosed in the cited material.

What is notable is how little of this company’s broader business model has previously centered on AI infrastructure. The Mikkeli project, the CFD and digital twin engineering work, and now the grid connection agreement collectively read as a company repositioning itself toward AI compute hosting, a pivot that mirrors a wider trend of smaller, previously niche technology and industrial firms rebranding around AI data center development as a growth story. Investors chasing that story should weigh the fact that no price, no GPU purchase order, and no binding revenue contract for the Mikkeli site has been made public alongside the power agreement.

Market Reaction and What the Stock Pickup Signals

The breadth of the pickup, from Quiver Quantitative’s market-moving-news tracker to a Pluang retail-investor newsletter in Southeast Asia to a Chinese-language flash from futunn.com, shows how globally distributed the appetite for AI infrastructure headlines has become. A single regional grid connection agreement for a company most retail investors could not have named a month ago generated coverage spanning at least four language markets within days. That is less a reflection of the deal’s size and more a reflection of how starved markets currently are for any tangible proof point that the AI buildout cycle, which has already reshaped chip demand, power markets and data center real estate, is continuing to expand into new geographies and new categories of operator.

That dynamic cuts both ways. It means smaller companies with real, if modest, infrastructure milestones can generate outsized media attention relative to their size, which is good for near-term visibility but also raises the bar for execution. A grid connection agreement is a necessary step, not a finished data center, and the gap between signing a power deal and actually energizing a site at full capacity in 2027, as the company has stated it expects to do, is where projects of this kind most often slip.

How Mikkeli Compares to Other AI Data Center Announcements

Placing the Mikkeli deal next to other recent AI infrastructure and chip-supply announcements helps calibrate expectations. Micron has been publicly walking investors through its HBM4 memory roadmap, arguing that next-generation AI training clusters will need roughly 2.6 times the memory bandwidth of current HBM3E designs to keep pace with larger models, a roadmap detailed in TrendInTech’s coverage of Micron’s HBM4 memory roadmap. Samsung, meanwhile, has reportedly tapped TSMC’s 3nm production lines to help manufacture base dies for Nvidia’s next HBM generation, a supply-chain maneuver covered in this site’s report on Samsung and TSMC’s HBM base die deal. Both stories point to the same underlying constraint driving projects like Mikkeli: AI compute growth is currently bottlenecked less by chip design than by the physical inputs, memory, power and cooling, needed to actually run the chips at scale.

On the power side specifically, Mikkeli sits in a crowded field of 2026 grid-related AI infrastructure stories. Microsoft’s decision to pull out of a UK data center deal with Nscale, detailed in TrendInTech’s report on the Nscale withdrawal, shows that even the largest hyperscalers are reassessing data center commitments as costs and timelines shift. At the grid-operator level, FERC’s approval of PJM’s new data center registry requiring utilities to track facilities above 50 megawatts, covered in this outlet’s piece on the FERC-PJM registry rule, reflects regulators scrambling to get visibility into exactly how much new load AI data centers are putting on regional grids, a concern that applies as much to Jarvi-Suomen Energia Oy’s 20 kV network in Mikkeli as it does to PJM’s mid-Atlantic territory.

Nvidia, AMD, Google and Meta have separately moved to standardize how AI data centers interact with the grid itself, agreeing on protocols including a two-second response standard for managing sudden power swings caused by GPU clusters ramping training jobs up or down, a story detailed in TrendInTech’s coverage of the new grid interaction standard. That kind of standardization effort exists precisely because AI training workloads create power demand spikes that traditional grid infrastructure, including the kind of medium-voltage distribution network Aurora Core Technology Oy just connected to, was never designed to absorb smoothly.

Historical Context: Finland’s Data Center Pitch

Finland’s data center strategy did not start with AI. The country spent the 2010s courting hyperscalers like Google and Microsoft with the same pitch now attached to Mikkeli: cool ambient temperatures that cut cooling costs, a grid with a high share of nuclear and renewable generation, and political and regulatory stability that larger operators value for multi-decade infrastructure commitments. What has changed since 2023 is the nature of the demand. Earlier data center waves in Finland were largely built around general cloud and enterprise workloads with relatively stable, predictable power draw. AI training workloads behave differently, with power consumption that can spike sharply during active training runs and fall during idle periods, a pattern that is part of why standardization efforts around grid interaction have become a live industry issue rather than a theoretical one.

The Mikkeli project, if it proceeds as planned, would extend Finland’s data center footprint beyond its traditional hubs and into a smaller regional center, a pattern that is becoming more common globally as prime real estate and grid capacity near major hubs get consumed by the largest operators. Smaller markets and secondary cities increasingly represent the remaining available capacity for new AI infrastructure projects, which is consistent with why a company like 3 E Network would choose Mikkeli over a more contested location closer to Helsinki.

Data Snapshot: The Mikkeli Agreement at a Glance

Detail	Figure or Status
Parent company	3 E Network Technology Group Limited (NASDAQ:MASK)
Finnish subsidiary	Aurora Core Technology Oy
Grid counterparty	Jarvi-Suomen Energia Oy (local distribution system operator)
Agreement signed	October 13, 2026
Grid voltage	20 kV medium-voltage connection
Ordered capacity	9,000 kVA (9 MVA)
Planned GPU architecture	NVIDIA Vera Rubin (company-stated plan, unconfirmed order)
Target energization	2027 (planned full capacity)
Site location	Mikkeli, Finland (Lakeland region)

Data Snapshot: Related AI Power and Infrastructure Moves in 2026

Story	Key Figure	Relevance to Mikkeli
Micron HBM4 roadmap	2.6x bandwidth need vs current HBM	Memory bottleneck for any Vera Rubin-class cluster
Samsung/TSMC HBM base die deal	3nm production lines tapped	Chip supply chain feeding next-gen GPU clusters
Microsoft/Nscale UK withdrawal	Deal cancelled	Shows hyperscaler caution on data center commitments
FERC/PJM data center registry	50MW+ facilities tracked	Regulatory push for grid-load visibility
Nvidia/AMD/Google/Meta grid standard	2-second response protocol	Addresses power swings from GPU training spikes
3 E Network Mikkeli deal	9 MVA ordered capacity	Regional-scale entrant in same AI power cycle

Risks and Open Questions

Several details that would typically accompany a project of this kind remain undisclosed. No price has been reported for the power connection itself or for the broader Mikkeli build-out. No named company executive or Jarvi-Suomen Energia Oy representative has been quoted directly in the reporting available so far. There is no confirmed GPU purchase order tying the project to Nvidia’s Vera Rubin line beyond the company’s stated plan, and nothing has been said publicly about expected customer contracts, lease agreements, or anchor tenants for the compute capacity once it comes online.

That gap between announcement and operational reality is normal at this stage of a data center project, but it is also exactly where execution risk concentrates. Grid connection agreements can and do slip, permitting can take longer than planned, and supply-chain coordination for mechanical and electrical systems, which the company says has already begun, still has to clear procurement, delivery and installation before a 2027 energization target is realistic. Investors and readers tracking MASK stock or the broader Nordic AI data center trend should treat the 2027 full-capacity date as the company’s planned target rather than a guaranteed outcome.

Predictions: Where This Goes From Here

• Expect 3 E Network to disclose additional Mikkeli construction milestones, likely tied to substation and cooling-system procurement, over the next two to three quarters as the project moves from engineering to physical build-out.
• Watch for a formal GPU supply announcement or vendor partnership confirming the Vera Rubin architecture claim; until that happens, the hardware plan remains a stated intention rather than a signed order.
• Other small and mid-cap companies with Nordic or Baltic subsidiaries are likely to pursue similarly sized, sub-10 MVA grid connection agreements in 2026 and 2027, following the same faster-approval logic that let Mikkeli clear its local utility relatively quickly.
• Regulatory scrutiny of data center grid connections, following the FERC-PJM registry precedent in the US, is likely to extend to European distribution system operators, potentially adding reporting requirements to deals like the Jarvi-Suomen Energia Oy agreement.
• If 3 E Network does not provide a capital financing update or construction progress report within the next two quarters, expect investor attention, and the wire pickups that followed this announcement, to fade quickly, since the stock’s current visibility is tied almost entirely to this single infrastructure story.

Why This Story Matters Beyond One Stock

The Mikkeli announcement is a useful barometer for how far down the market capitalization scale the AI infrastructure boom has reached. Two years ago, power purchase and grid connection agreements of this kind were almost exclusively the domain of hyperscalers and dedicated data center REITs with balance sheets large enough to self-fund gigawatt-scale campuses. A Nasdaq-listed company signing a 9 MVA agreement through a Finnish subsidiary, and getting picked up by six or more financial news outlets across multiple continents within days, suggests the AI power story has become a mainstream market-moving category even at this much smaller scale.

It also highlights a theme running through nearly every AI infrastructure story this year: the constraint has moved from chips to power. Nvidia can design Vera Rubin, Micron can promise 2.6 times the memory bandwidth, and TSMC can fabricate the dies, but none of it runs without a grid connection, a cooling loop and a substation. 3 E Network’s 9 MVA filing is a small example of that larger pattern, playing out in a Finnish lakeside town most AI headlines will never otherwise mention.

Frequently Asked Questions

What did 3 E Network Technology Group actually sign?
A power distribution network connection agreement between its Finnish subsidiary, Aurora Core Technology Oy, and local grid operator Jarvi-Suomen Energia Oy, covering 9,000 kVA (9 MVA) of capacity on a 20 kV medium-voltage connection, signed October 13, 2026.

Is 9 MVA a large amount of power for a data center?
It is modest compared to gigawatt-scale hyperscaler campuses but substantial enough to support a meaningful cluster of AI servers at a regional site. The company has not disclosed a specific rack count or server total.

When will the Mikkeli AI compute center be operational?
The company has stated it expects completion and full-capacity energization in 2027. That is a planned target, not a guaranteed date, and depends on substation construction, cooling installation and supply-chain delivery.

What GPU hardware will the Mikkeli site use?
The company has stated plans to use NVIDIA’s Vera Rubin architecture, the generation succeeding Blackwell. No GPU purchase order or vendor contract has been publicly confirmed.

What is 3 E Network Technology Group’s stock ticker?
The company trades on Nasdaq under the ticker MASK.

Where is Mikkeli, Finland?
Mikkeli is a city in Finland’s Lakeland region, roughly a few hours’ drive northeast of Helsinki, an area not traditionally associated with large-scale data center development.

Has a price or project cost been disclosed for the Mikkeli center?
No. Reports on the agreement do not include a disclosed price for the power connection or the broader project build-out.

How does this compare to other AI data center power stories in 2026?
It is smaller in scale than hyperscaler megaprojects but part of the same broader pattern of AI operators racing to secure grid capacity, a pattern also visible in Microsoft’s UK Nscale withdrawal, FERC’s PJM data center registry rule, and the grid-interaction standard recently set by Nvidia, AMD, Google and Meta.

Related Coverage

• HPE's Self-Driving Network Fixes 70-80% of Issues [2026]
• Maxx Conectado Cuts Latency 5ms With Eletronet Deal [2026]
• EU Hits KKR With Charges Over $24.7B Telecom Italia Deal
• Huosheng Lands 200-Ton Engine Deal in Ningbo [2026]
• NY-15 Rep's Bill Could End Game Licenses for Good [2026]