When AI factories start "self-contained energy storage": The logic and risks behind Fluence's 43.8% surge after being NVIDIA's sole designated partner

On June 1, 2026, the stock price of the global leading grid-scale energy storage company Fluence Energy (FLNC.US) jumped by 43.8% overnight, closing at $27.15, reaching a new high since February. The only reason behind this rally is clear: Siemens officially released a reference electrical and power architecture design for AI factories based on NVIDIA DSX Vera Rubin NVL72 platform, with Fluence’s SmartStack battery energy storage system incorporated into it, becoming the sole explicitly designated battery storage partner in this reference design.

This marks the first time energy storage has been “embedded” into the top-level blueprint of AI data centers. It signifies that energy storage is no longer just a supporting component for renewable energy projects but is becoming a “standard part” of trillion-dollar AI computing infrastructure.

The macro context of the AI data center energy consumption crisis: Why energy storage has become an “essentials”

To understand the rationality of Fluence’s recent surge, it’s first necessary to clarify the structural root causes of the AI data center energy consumption crisis.

According to data from the International Energy Agency (IEA), global data center electricity consumption in 2024 will reach 415 TWh, accounting for 1.5% of total global electricity use. It is expected to double to 945 TWh by 2030—comparable to Japan’s annual electricity consumption. More structurally significant is the divergence in growth rates: from 2024 to 2030, nearly 50% of the increase in U.S. data center electricity demand will account for the country’s total demand growth; by 2030, AI data processing electricity consumption in the U.S. will surpass the combined electricity use of traditional high-energy industries such as aluminum, steel, cement, and chemicals.

Morgan Stanley estimates that from 2025 to 2028, the cumulative power shortfall in U.S. data centers will reach 47 GW. In other words, insufficient power supply has become the core bottleneck constraining AI computing expansion.

The role of battery energy storage in this landscape is undergoing a qualitative change. Previously, the value of storage in data center scenarios mainly manifested as backup power (replacing or supplementing UPS). But the exponential increase in power density of AI cabinets has changed this logic—over six years, from NVIDIA A100 to Vera Rubin, the power consumption per GPU has surged from 400W to over 1,400W, and the power per rack from 30kW to nearly 180kW. This growth brings two major issues: first, traditional UPS six-level conversion architectures (with an end-to-end efficiency of about 89%) cannot support such high-density DC loads; second, GPU instantaneous power fluctuations can reach 1.5 to 2 times the average power, which the grid cannot smooth through traditional means within milliseconds.

Therefore, the positioning of energy storage systems in AI data centers has evolved from “backup” to “active support”—providing grid-level functions such as black start, voltage and frequency ride-through, and also solving transient power fluctuations through AI load smoothing. This is the fundamental logic behind Fluence’s SmartStack being incorporated into the reference design.

Fluence Energy (FLNC): The true value of NVIDIA + Siemens’ “sole designated” partner

Breakdown of cooperation details

The reference design for NVIDIA’s Vera Rubin AI factory, officially released by Siemens, covers a total facility capacity of 136 MW, with IT load at 100 MW, and the power chain spanning from 34.5kV municipal high-voltage connection to server racks. Fluence’s SmartStack energy storage system is configured at 120 MW / 240 MWh, providing key functions such as voltage and frequency ride-through, black start, grid demand response, and AI load smoothing.

It’s noteworthy that this reference design is not a “contract,” but a non-binding technical blueprint jointly issued by Siemens and NVIDIA. However, Fluence is the “only designated BESS partner” in this blueprint, and among NVIDIA’s seven infrastructure OEM design schemes, only Fluence is explicitly named.

The structural reason for this “uniqueness” is straightforward: Fluence itself is a joint venture between Siemens and AES Corporation, an American power company. Being included in the parent company’s official reference design is essentially a resource tilt. In other words, this is not a “technology bidding” victory but an “ecosystem inheritance.”

Why did the market give a 43.8% increase?

Beyond the endorsement effect of being the “only designated partner,” the market also noticed a key parameter: the reference design specifies a battery endurance of 2 to 3 hours, significantly higher than the previous market expectation of about 1 hour. This implies a larger demand for BESS per project and higher order values. Moreover, Fluence has already secured master service agreements with two ultra-large data center operators, with backlog orders totaling a record high of $10.1 billion, providing substantial predictability for revenue expectations.

Financial data: the divergence between revenue growth and ongoing losses

In Q2 of fiscal 2026, Fluence achieved revenue of $464.89 million, up 7.71% year-over-year, but below the market consensus of $614.93 million; net loss was $20.93M. Trailing twelve months (TTM) revenue is approximately $2.58 billion, but TTM net profit remains negative, with a P/E ratio of -59.66. The debt-to-equity ratio is 87.73%, and EBITDA is still negative.

In other words, Fluence is a rapidly growing but still unprofitable company. The 43.8% single-day surge is not based on recent profitability improvements but on an upward revision of future order conversion expectations.

Is the market overreacting?

This is the most critical judgmental issue in this article. It requires reasoning from the following angles:

Factors supporting “rationality”: Being included in NVIDIA’s reference design is equivalent to obtaining a “joint certification” from NVIDIA and Siemens. This reference design will become the preferred template for large-scale data center construction. If each 136 MW AI data center is configured with 120 MW / 240 MWh storage according to this standard, the value of a single project’s BESS could range from tens of millions to hundreds of millions of dollars based on current storage costs. Considering the potential scale of global AI data center construction, this “ecosystem lock-in” has high long-term value.

Factors pointing to “overvaluation”: The reference design is currently only a non-binding blueprint, not yet translated into actual orders. Fluence’s revenue growth is accelerating, but gross margin is only about 11.71%. Whether scale effects can translate into positive profits remains to be seen. As of early June, analysts’ average 12-month target price was only $18.59, about 25% below the surge-high stock price. The post-rally price has already exceeded some analysts’ target ranges—Canaccord Genuity’s buy rating target is $28, and the June 1 close at $27.15 is close to that level.

Fluence’s 43.8% increase in a single day has a logical basis at the “concept reaction” level, but there is some “pre-pricing” involved. The narrative that energy storage is integrated into the top-level design of AI data centers has a long tail effect, but the short-term path from 1 to N remains uncertain. From an investment perspective, Fluence’s core risk lies not in demand (this logic is generally sound), but in: when will profitability turn positive; and to what extent can the “only designated” status translate into an actual exclusive market share advantage.

Horizontal comparison: three different dimensions of “AI power stocks”

Bloom Energy (BE): AI-exclusive distributed fuel cell solutions

Fluence’s story is about “storage integrated into the grid.” Bloom Energy’s story is entirely different: its solid oxide fuel cells (SOFC) are standalone off-grid power generators, deployable directly within data center campuses, with delivery within 90 days.

In Q1 2026, Bloom Energy achieved revenue of $751 million, up 130.37% year-over-year, far exceeding market expectations (expected $540 million); adjusted EPS was $0.44, up over 400% YoY. The full-year revenue guidance was raised to $3.4–3.8 billion, an approximately 80% increase. The explosive growth in AI data center orders is the core driver—about one-third of product orders come from AI data center clients, with long-term service orders reaching $9.6 billion.

However, Bloom Energy faces a very different risk from Fluence: extreme valuation distortion. The company’s P/E (TTM) is about 11,970 times, and P/S is about 34 times. Gurufocus’s fair value estimate shows that the current stock price (~$285) is about -2,760% above fair value (~$10.27). The issue isn’t the fundamentals—demand for distributed power in AI data centers is real and strong—but the current valuation already embeds extremely high growth assumptions, leaving little room for any quarter below expectations.

NextEra Energy (NEE): A large-scale renewable + storage all-in-one platform

Compared to the previous two, NextEra Energy is a fully mature and profitable utility giant. The company operates two major segments: regulated utility Florida Power & Light (FPL), and NextEra Energy Resources, a clean energy development arm.

In Q1 2026, NEE reported revenue of $6.7 billion, up 21%; net profit was $2.18 billion, up 162%; and new renewable and storage orders added 4 GW (including 1.3 GW of battery storage), with total backlog orders rising to about 28–33 GW. The company maintained its 2026 adjusted EPS guidance of $3.92–$4.02.

NEE’s advantage lies in risk diversification and dividend returns (current dividend yield about 2.6%), but its yield elasticity is relatively limited. Driven by AI power demand, the market’s valuation premium for NEE is expanding (current P/E forecast at about 23.7x, above the past two-year average of 21.4x), but the valuation space for regulated utility assets is inherently lower than for pure growth AI power infrastructure stocks.

Vistra Energy (VST): The benefit logic of traditional power producers’ AI power demand

Vistra is one of the largest competitive power producers in the U.S., operating a mix of natural gas, nuclear, coal, solar, and battery storage assets. Its benefit logic is straightforward: AI data centers drive U.S. wholesale power prices higher, and as a net power seller, Vistra benefits directly from this spread expansion.

In Q1 2026, Vistra achieved revenue of $5.64 billion, up over 43%; net profit was $1.03 billion; adjusted EBITDA was $1.49B. The company completed a $6.13 billion share buyback plan and was upgraded to investment grade by two major rating agencies. Vistra expects its core profit from continuing operations in 2026 to be between $6.8 and $7.6 billion, a significant increase over 2025 forecasts.

The risks for Vistra include high leverage and profit volatility due to wholesale power price fluctuations. But based solely on Q1 2026 data, its performance delivery capacity is the most solid among the four stocks.

Cross-sectional comparison matrix

| Dimension | Fluence (FLNC) | Bloom Energy (BE) | NextEra (NEE) | Vistra (VST) | | --- | --- | --- | --- | --- | | AI data center exposure | Storage integrated into NVIDIA reference design | Distributed SOFC, 90-day deployment | Large-scale renewable + storage orders | Wholesale power price-driven benefits | | Recent revenue growth | +7.7% (Q2 2026) | +130% (Q1 2026) | +21% (Q1 2026) | +43% (Q1 2026) | | Profitability | Loss, TTM net profit negative | P/E ~12kx | P/E ~24.7x, profitable | Profitable, high cash flow | | Core risks | Not yet profitable + order conversion uncertainty | Valuation bubble | Limited yield elasticity | Power price volatility + high leverage | | Core catalysts | NVIDIA reference design adoption | AI data center distributed power orders | Conversion of clean energy backlog | Continuous rise in wholesale power prices |

Overall, these four stocks represent four different risk-reward profiles within the AI power sector:

• Fluence: The most “conceptually sexy,” with the clearest ecosystem lock-in, but the longest cycle for profit validation.
• Bloom Energy: The most direct beneficiary of AI data center distributed power demand, but with the highest valuation risk already priced in.
• NextEra: The most stable long-term holding, with limited return elasticity.
• Vistra: The clearest transmission of AI demand, with the highest short-term performance validation, but high leverage and volatility require close monitoring.

Latest developments: Gate officially launches US stock trading, breaking down barriers between crypto assets and traditional markets

For investment opportunities in the energy storage sector to materialize, investors need convenient trading channels. In June 2026, Gate officially launched US stock trading, achieving a key breakthrough by enabling direct US stock trading with USDT.

Gate Stocks’ core features include: users can trade over 10,000 major US stocks and ETFs through a unified Gate account, without switching to traditional brokers or opening additional accounts. Orders are executed directly on major US exchanges. Dividends are automatically credited in USDT, with no funding rates or overnight fees.

Particularly noteworthy is that Gate has integrated stock trading into its VIP tier system—users with holdings of at least $2,000 can upgrade to VIP and enjoy a minimum fee rate of 0.023%. Trading hours have expanded from 6.5×5 to 16×5, covering pre-market, regular, and after-hours trading, allowing investors to respond promptly to major news (such as Fluence’s announcement during pre-market). It also supports fractional trading starting from 0.01 shares, greatly lowering the barrier to investing in high-priced tech stocks.

For the four stocks discussed in this article (FLNC, BE, NEE, VST), investors can participate directly via USDT on Gate Stocks, with a unified account system enabling cross-market management of crypto and traditional securities assets.

Conclusion

The AI data center energy consumption crisis is reshaping the valuation paradigm of storage, power, and new energy infrastructure industries. The joint release of NVIDIA and Siemens’ reference design, integrating energy storage into the core architecture of AI factories, marks the transition of this sector from “concept” to “standardized configuration.” Fluence Energy’s 43.8% single-day surge has made it the most market-focused stock in this paradigm shift, but its ongoing losses and the short-term uncertainty of the non-binding blueprint imply some degree of early pricing.

From a horizontal comparison, Bloom Energy benefits from the most direct realization of AI data center distributed power demand but faces the highest valuation risk; NextEra Energy offers a stable long-term option with a backlog of storage orders; Vistra Energy’s AI demand transmission path is the clearest, with the highest validation in Q1 but also high leverage and volatility. These four stocks together form a complete spectrum of “high expectation” to “high realization” in the AI power track.

With Gate’s US stock trading now available, investors can easily participate across markets. However, regardless of how long-term the energy storage narrative is, short-term stock price fluctuations remain influenced by order conversion pace, interest rate environment, and market sentiment. Any sharp price movements triggered by a single event should be understood within the constraints of fundamentals and valuation.