Enki AI

04/14/2026

Two signals tell me a technology is becoming unstoppable.

The first is installations.

Each new installation tells you someone chose that product over everything else available. Each bigger installation tells you they want more. Installations are not announcements.

They are decisions. Repeated decisions, across different customers and different markets, are market validation you can trace.

The second is unit economics.

When a new technology enters the mainstream, the cost per unit should be falling. Not gradually. Structurally. The solar curve is the clearest example in history. Every time cumulative capacity doubled, costs dropped by roughly 20%. That is the Swanson curve. It did not stop. It compounded.

When I see both signals together, growing installations and falling unit economics, that is when I know the technology is no longer a forecast. It is a fact.

Right now I am watching both signals in battery storage.

US operating storage capacity reached 37 GW by the end of 2025, up 32% in a single year. Lithium-ion battery costs dropped 40% in 2024 alone, following a 90% decline over the prior decade.

Installations accelerating. Unit economics compressing.

That combination does not reverse.

What technology are you tracking these two signals on right now?

04/13/2026

Maritime hydrogen peaked and froze within the same six-month window.

Q1 2026 was the most active quarter the sector had seen. Fincantieri launched the first hydrogen cruise ship. Ballard Power Systems secured an order for 32 FCwave engines.

PowerCell contracted fuel cell systems for two hydrogen bulk carriers. Japan Suiso Energy partnered to build a liquefied hydrogen carrier. Norway had committed $100 million in government funding in Q4 2024. The pipeline looked real.

Q2 2026: commercial events fell to near-zero. PR activity became negligible. Positive and negative sentiment both disappeared. The Enki activity chart shows two lines converging flat at the bottom.

A market going negative is a correction. A market going silent is something different.

Silence in commercial activity data means financing dried up and deals were suspended simultaneously. It means the companies active in Q1 are not announcing exits. They are not announcing anything. That pattern, peak followed immediately by freeze rather than gradual decline, is the signal that the underlying commercial case broke faster than the public narrative caught up.

The structural problem in maritime hydrogen was always the same one that exists across every hydrogen application. The technology on the vessel works.

The supply chain to fuel the vessel does not exist at commercial scale. Norway's hydrogen ferries operate in a closed, government-supported corridor with dedicated fueling infrastructure. That model does not replicate across bulk carriers and cruise ships operating on global trade routes without a parallel infrastructure build that has not been funded.

The Q1 2026 activity surge was real commercial ex*****on within the existing pilot envelope. The Q2 freeze is what happens when the pilot envelope runs out and the next layer of investment, the one that requires commercial-scale fueling infrastructure, does not materialise.

The signal I am watching: whether Q3 2026 shows recovery or continued silence.

Recovery requires either a government-backed infrastructure commitment covering fueling at scale, or a reframe of the application toward closed-corridor routes where dedicated infrastructure is viable.

Continued silence confirms the market has reached the boundary of what the current funding model can support.

Watch the commercial activity data, not the conference announcements.

04/09/2026

A company makes a big announcement. A pilot. A new partnership. A major project.
Then nothing.
No follow-up. No update. No next step.
In most cases that silence is the signal.
Cancellations in clean energy do not come with press releases. They come with silence. A project stops being mentioned. A partner quietly exits. A pilot never moves to deployment.
This is exactly what happened in green hydrogen in 2025.
Nearly 60 major projects cancelled. Air Products abandoned three US projects. BP cancelled two flagship hydrogen programs. Shell halted its Aukra project in Norway. Most of it happened quietly.
The announcements were loud. The exits were not.
This is why I built cancellation and no-follow-up tracking into Enki. Not because exits are easy to find. Because they are nearly impossible to find without a system built specifically to surface them.
The absence of activity is data. Most teams are only tracking what companies say. The more important signal is what they stopped saying.

04/08/2026

I spent the last hour going through the hydrogen rail data for 2026.

Here is what jumped out.

In Q1 2026, PR activity and commercial events are nearly at parity.

Value: 10 vs 9.

That number matters more than any market forecast.

For most of 2024 and 2025, the gap between what was announced and what was actually deployed was the story. Germany's Alstom fleet delayed three years. A hydrogen tram in China shut down because the economics didn't work. A full fleet reverting to diesel because of a fuel cell shortage.

The technology was real. The ecosystem wasn't ready.

Q1 2026 looks different.

India deployed its first hydrogen train. Ridgewood Infrastructure acquired Sierra Railroad — a financial investor buying into a hydrogen locomotive asset. Concord Control Systems received a Rs. 47 crore commercial order from NTPC. Ballard began launching hydrogen trains for Berlin's Heidekrautbahn.

These are not pilot announcements. These are commercial events.

The signal gap is closing. That is the inflection point worth tracking.

The question now isn't whether hydrogen rail works. Germany proved it works in 2022. The question is whether the supply chain, infrastructure, and economics can hold at scale.

Q1 2026 says: maybe yes.

Full analysis on the Enki platform.

[INSERT ARTICLE URL]

04/07/2026

The global LNG market has been pricing a supply glut for two years.

New capacity from Qatar's North Field expansion. US modular projects sanctioning at record pace. The consensus view heading into 2026 was oversupply, price competition, and margin pressure for producers.

That consensus was built on one assumption nobody was pricing seriously: that 20% of global LNG supply concentrated in a single industrial complex at Ras Laffan operates without interruption.

Geographic concentration is the oldest risk in commodity markets. It is also the most consistently underpriced one because it rarely materialises and the cost of hedging it feels unnecessary until it does.

Here is what the structure of the risk actually looks like.

Ras Laffan is not just an LNG terminal. It is an integrated industrial city processing natural gas from the North Field into LNG, blue ammonia, helium, and petrochemicals simultaneously. A disruption does not affect one product stream. It cascades across all of them. Buyers holding 27-year Qatari SPAs signed on cost and volume assumptions have no geographic alternative embedded in those contracts.

Shell holds 6.8 MTPA of contracted Qatari volume. A production halt does not stop Shell's downstream obligations. It forces Shell to declare force majeure on its own customers, propagating the supply shock down the chain. The concentration risk at the source becomes systemic risk at the trading level within days.

European storage entered 2026 at 30% utilisation against a 54% seasonal average. The buffer that absorbs supply disruption was already below historical norms before any event tested it.

The oversupply consensus was directionally correct on new capacity additions. It was structurally incomplete on what happens when the concentration risk that everyone knows about finally gets tested.

The signal I watch is not price. It is contract restructuring.

When long-term SPA holders begin inserting geographic diversification clauses, secondary source obligations, or explicit force majeure carve-outs into new or renegotiated contracts, the market has formally repriced the concentration risk it spent two years ignoring.

Watch the contract terms, not the headlines.

04/06/2026

Qatar supplies 20% of global LNG from a single industrial complex.

The 27-year contracts signed with European and Asian buyers are priced against that concentration. Most have never formally stress-tested simultaneous failure across the Ras Laffan complex.

The commercial evidence of repricing is already visible.

European gas storage entered 2026 at 30% utilisation against a 54% seasonal average. Venture Global's Calcasieu Pass force majeure dispute proved the legal mechanism exists and will be used commercially. Against that backdrop, destination-flexible US LNG contracts from Gulf Coast terminals are being signed at a premium to Qatari SPA pricing by buyers who have updated their supply security models.

CP2 Phase 2 financing was oversubscribed at $19 billion against an $8.6 billion ask. That ratio is capital markets pricing supply security, not just project returns.

**Three signals Enki is tracking:**

**Strike prices in new US LNG SPAs in 2026.** A material premium above equivalent Qatari contract pricing confirms geographic concentration risk is now formally embedded in commercial contract structures.

**European utility diversification away from Qatari volumes.** Voluntary Qatari contract reduction before expiry is the clearest signal buyers are acting on concentration risk before a disruption forces the decision.

**Qatar North Field expansion financing conditions.** A widening spread versus equivalent US Gulf Coast project financing costs is the market pricing concentration risk into the capital structure of the concentration itself.

The structural deficit risk does not require a disruption to be commercially relevant. It is already being priced.

Enki tracks the commercial signals that confirm when a structural risk has moved from theoretical to priced.

Follow for weekly signal analysis or book a demo.

04/02/2026

04/01/2026

Everyone agrees green hydrogen is the future.

That consensus is the problem.

I saw this pattern in shale in 2014. Everyone agreed. Investment flooded in. Drilling costs spiked. The market assumed infinite demand. Someone would always buy.

What happened? Cost of accrues went up. The flood of money crashed the market. It delayed adoption by a decade.

Green hydrogen is at the same inflection point today.

Global electrolyzer investment grew 80% in 2025. China has 10 GW under construction. Everyone agrees this is the decade of green hydrogen.

When everyone agrees, watch the unit cost. That is the signal.

Not the headline. Not the investment number. The unit cost of production.

That is what I track.

03/31/2026

TotalEnergies just handed back two US offshore wind leases it paid for in 2022.

The Department of the Interior agreed to return the lease fees. TotalEnergies agreed to reinvest that same capital into US gas and power projects.

That transaction structure is the signal. Not the exit itself.

A major integrated energy company did not simply walk away from offshore wind in the US. It negotiated a mechanism to recover sunk costs and redeploy them into LNG.

The Carolina Long Bay and New York Bight leases, both awarded in 2022, are gone. The capital is now heading toward Rio Grande LNG at 29 million tonnes per annum capacity and a letter of intent for 2 MMtpa offtake from Alaska LNG, subject to final investment decision.

TotalEnergies stated that US offshore wind development costs came in higher than comparable European projects, raising concerns about long-term power affordability. That cost differential is not a TotalEnergies-specific problem. It is a structural feature of the US offshore wind market that every developer with active leases is currently pricing into their own project models.

BP wrote down $1 billion on US offshore wind in 2025. Equinor paused development on multiple projects. Over $35 billion in US offshore wind capacity was canceled or restructured in the two years prior.

TotalEnergies is not an outlier. It is the most recent data point in a consistent pattern of capital reallocation away from US offshore wind toward assets with clearer near-term return profiles.

The direction of that reallocation is the part worth tracking carefully.

Every dollar exiting US offshore wind is not sitting idle. It is moving into LNG export infrastructure, upstream gas production, and power generation assets tied to demand growth from data centers and industrial reshoring.

TotalEnergies' settlement structure made that reallocation explicit and contractually defined.
The signal I am watching: Rio Grande LNG final investment decision timeline.

TotalEnergies is now a committed capital partner in a 29 MMtpa LNG export facility. The FID on that project in 2026 is the confirmation that the capital reallocated from offshore wind has found a bankable home.

That is when the reallocation becomes irreversible at scale.

The question for every offshore wind developer still holding US leases is not whether TotalEnergies made the right call. It is whether their own project economics tell a different story.

03/30/2026

Bloom Energy started as a niche application in data centers.

Repeated use cases forming quietly.

Now the stock is 20x from where it started.

I track niche applications because that is where the commercial signal appears before any analyst report confirms it.

Small modular reactors are at that stage right now.

Military bases. Remote industrial sites. Dedicated data centers.

Niche applications, repeated across different end markets.

Meta just committed to eight SMR plants. NVIDIA's investment arm is in. The NRC is expected to issue the first commercial construction permits this year.

The pattern is the same. The question is whether your team is tracking the signal now or waiting for the headline.

What niche application are you watching right now?

03/27/2026

€17.61 billion order backlog at Siemens Energy in Q1 2026.

Thirty percent growth year over year.

Driven overwhelmingly by data center demand they cannot fulfill fast enough.
That backlog figure is the most precise available measure of the gap between AI infrastructure deployment ambition and the physical hardware required to execute it. Power transformers and switchgear are the constraint. Not compute. Not real estate. Not permitting. The electrical equipment that energizes the facility.

The capital response has been direct. Siemens committed over $1.5 billion to US manufacturing expansion in eighteen months. A $190 million Fort Worth facility opened in March 2025 doubling low and medium-voltage equipment output.

A $1 billion Siemens Energy commitment in February 2026 targets transformer and grid equipment production including a new Mississippi plant. The investment thesis is straightforward: control your own supply chain or watch competitors who do take your order book.

The partnership structure confirms the same logic. Siemens and Eaton built a modular off-grid power system collaboration in June 2025 specifically to bypass utility interconnection queues. Siemens, NVIDIA, and nVent released a 100MW hyperscale reference architecture in December 2025. Siemens and Delta announced prefabricated modular solutions targeting 50% faster deployment in November 2025. Each partnership solves a specific deployment bottleneck. None of them are exploratory. All of them are ex*****on-focused.

Three signals Enki is tracking:
Fort Worth and Mississippi facility production throughput against capacity targets. Siemens has committed the capital and built the footprint. Actual manufacturing output relative to installed capacity, visible in quarterly earnings disclosures, will show whether the supply constraint is resolving at the pace the hyperscaler order book requires. Watch H2 2026 earnings for output metrics specifically against the new facility ramp schedules.

Watch for procurement announcements from hyperscalers and colocation operators with confirmed 2026 and 2027 delivery requirements.
The AI infrastructure build-out is not supply-constrained at the demand layer. It is supply-constrained at the factory floor. The companies that resolve that constraint first capture the order book that everyone else is waiting to fill.

Enki tracks the gap between infrastructure commitment and delivered hardware capacity.
Follow for weekly signal analysis or book a demo.

03/26/2026

Two deals. $7.65 billion in committed capital. Eighteen months.

Bloom Energy's $5 billion Brookfield partnership in October 2025 and its $2.65 billion AEP agreement in January 2026 are not isolated transactions. They are the same procurement logic expressed twice at increasing scale.
Grid interconnection queues in the US are running three to five years.

Data center operators with confirmed hyperscaler demand cannot defer deployment waiting for grid capacity that does not exist on their timeline. Solid oxide fuel cells deploy in 90 days, operate at 60% electrical efficiency, and bypass the interconnection queue entirely. The AEP deal is a regulated utility formalizing that logic at 1GW scale for AI data center load it cannot otherwise serve on a commercially viable timeline.

The technology readiness question has been answered. The ex*****on question is now the signal worth tracking.
Three signals Enki is tracking:
AEP deployment milestone confirmation in H2 2026. The January 2026 agreement is an offtake commitment. The first verified physical deployment milestone under that contract is the signal that converts a record order book into operational evidence. Watch for AEP project update disclosures in utility filings and earnings calls in Q3 and Q4 2026.

Bloom Energy manufacturing output against 2GW capacity target. The company announced a manufacturing expansion to 2GW annual capacity in August 2025. Actual production throughput data relative to that target, visible in quarterly earnings disclosures, will show whether the supply chain can execute at the pace the order book demands. A gap between capacity and output is the early warning signal for deployment delays across the AEP and Brookfield pipelines.

Second utility-scale offtake agreement beyond AEP. The AEP deal established a pricing and structure template for utility-scale SOFC procurement for data center load. A second comparable agreement from a different utility in 2026 confirms the model is replicable, not a single bespoke arrangement. Watch procurement announcements from utilities with confirmed data center load growth in PJM, ERCOT, and MISO territories.

The grid bypass market for AI data center power is past the proof-of-concept stage. The commercial signals now sit in ex*****on, supply chain throughput, and replication rate.

Enki tracks the distance between the signed agreement and the delivered megawatt.

Follow for weekly signal analysis or book a demo.