January 30, 2025

Opening Graph

The AI energy boom is looking a little shaky. DeepSeek, a Chinese AI startup, has thrown a wrench into U.S. electricity demand forecasts with models that require a fraction of the power once assumed necessary. Meanwhile, PJM has agreed to cap capacity prices after last year’s runaway auction costs, and global lenders have pledged billions to electrify Africa. The energy landscape is shifting rapidly—adapt or be left behind.

Chinese AI and the Energy Outlook

The market just got a hard lesson in volatility. Shares of independent power producers and advanced nuclear firms plummeted Monday after Chinese AI startup DeepSeek released high-performing models that appear to require far less energy than their U.S. counterparts. The news sent Vistra and Talen tumbling over 20%, with Constellation Energy suffering a similar fate.

DeepSeek’s models reportedly cost 45 times less to train than OpenAI and Anthropic’s latest offerings. That efficiency threatens the foundation of recent U.S. electricity demand forecasts, which had been largely propped up by expectations of massive AI-driven power consumption. Investment bank Jefferies warned that AI accounts for about 75% of projected U.S. load growth through 2035—an assumption that may now be wildly overstated.

The fallout hit advanced nuclear firms hard. Oklo, which recently secured a 20-year deal with data center developer Switch for up to 12 GW of power, saw its stock price drop by over 20%. NuScale, which had been riding high on a 580% gain since January 2024, suffered the same fate. Meanwhile, Amazon and Google’s nuclear partnerships with X-energy and Kairos Power remain intact, but whether their projected energy needs hold up is now an open question.

Not everyone is convinced this is the beginning of the end for AI energy demand. J.P. Morgan analysts pointed to Jevons Paradox, where increased efficiency leads to even greater consumption. If AI becomes cheaper to operate, broader adoption could push demand even higher. At the same time, Trump’s recently announced AI joint venture, Stargate, promises to inject $500 billion into data center expansion, potentially counteracting DeepSeek’s efficiency gains.

The bottom line: DeepSeek may have shaken confidence in AI-driven power demand, but don’t bet against the energy sector just yet. The race between efficiency gains and exponential AI adoption is far from over.

PJM’s Price Cap Battle

Four governors and a coalition of utility regulators have called on FERC to lower the price cap for PJM’s capacity auctions, arguing that record-high prices from the July auction are unjustifiable. With capacity costs jumping from $2.2 billion to $14.7 billion for the upcoming delivery year, the stakes are immense for both consumers and grid operators. Pennsylvania Governor Josh Shapiro’s proposal suggests capping prices at 1.5 times the Net Cost of New Entry, or Net CONE, as a temporary measure to stabilize costs while reforms are developed.

Opposition from power producers, including Constellation Energy, highlights the tension between consumer relief and long-term reliability. Generators argue that higher prices signal the need for new power supplies, and capping them could deter critical investment. PJM itself is caught in the middle, acknowledging the challenges of managing resource adequacy while protecting consumers. The debate underscores a broader challenge in balancing affordability, reliability, and the push for cleaner energy sources.

PJM Price Cap Deal

After months of political and regulatory pressure, PJM Interconnection has agreed to lower its capacity auction price cap in a deal with Pennsylvania Governor Josh Shapiro. The new cap will be set at $325/MW-day, with a floor at $175/MW-day for the 2026/27 and 2027/28 capacity auctions.

Shapiro’s office claims the deal will save PJM ratepayers $21 billion over two years. The previous auction saw capacity prices skyrocket from $28.92/MW-day to $269.92/MW-day in most zones, with some areas hitting price caps of over $450/MW-day. That surge sparked a backlash from governors in Delaware, Illinois, Maryland, and New Jersey, as well as multiple state utility commissions and consumer advocates.

Generators, including Constellation Energy and the Electric Power Supply Association, fiercely opposed the move, arguing that price caps could deter investment in new generation. But PJM’s board has been under pressure to balance affordability with reliability. With data center growth uncertain and high capacity prices politically untenable, the grid operator may see this deal as the best way to keep both sides from launching another all-out regulatory war.

$8 Billion Boost for African Electrification

The Mission 300 initiative, an ambitious effort to connect 300 million Africans to electricity within six years, just secured an additional $8 billion in funding from global lenders. The new commitments, led by the Islamic Development Bank and the Asia Infrastructure Investment Bank, bring total pledged funding to $56 billion—still short of the estimated $90 billion needed to complete the project.

Africa’s electrification crisis remains staggering, with 600 million people still lacking access to power. Half of the planned new connections will come from grid expansions, while the other half will rely on renewables like solar and wind mini-grids.

Debt remains a major obstacle. Many African governments lack the financial capacity to fund large-scale infrastructure projects, and existing state-run utilities are often mismanaged and deeply in debt. That’s why this initiative is banking on a mix of development finance, private investment, and regulatory overhauls to attract more capital.

World Bank President Ajay Banga called the effort “foundational to everything,” noting that energy access is critical for economic development, job creation, and political stability. The big question now: Will the funding commitments translate into real-world results, or will bureaucratic inertia and political roadblocks slow progress yet again?

Upgrade to Grid Brief Premium to get extra deep dives into energy issues all over the world.

Conversation Starters

• NYT – Trump Has Now Made His Energy Project Clear
  In his first days back in office, Trump issued a flurry of executive orders on energy. He wants the U.S. to dominate rare earth mining while also cutting demand for EVs and wind power. He’s calling for more pipelines and refineries while ignoring permitting for renewables. The contradictions are glaring, but the underlying message is clear: fossil fuels are back in the driver’s seat.

• CrudeOil.com – The Small Nuclear Reactor Revolution Is Underway
  SMRs are gaining momentum as companies like Google and Amazon invest in next-generation nuclear technology to power data centers. With their modular design, smaller footprint, and faster deployment, SMRs could be a game-changer for clean baseload power. But can they scale fast enough to meet demand?

• The Hill – Permitting Reform Is Key to Unleashing American Energy. Can Trump Get It Done?
  Trump’s energy agenda isn’t just about drilling—it’s about slashing bureaucratic red tape. His executive orders push for faster environmental reviews and streamlined permitting, but the real test will be Congress. Will bipartisan momentum on permitting reform hold, or will partisan gridlock kill any meaningful changes?

Good Bet, Bad Bet

Good Bet: Cerebras
If DeepSeek’s ultra-low-cost training models signal anything, it’s that the AI arms race is shifting from sheer computing brute force to efficiency. That’s exactly where Cerebras comes in. While Nvidia dominates the AI chip market, Cerebras has quietly been building the most energy-efficient AI hardware on the planet. Its Wafer Scale Engine (WSE-3), the largest chip ever built, eliminates the need for massive GPU clusters, reducing power consumption and training costs by orders of magnitude. Instead of burning through thousands of power-hungry GPUs, companies using Cerebras can achieve similar AI performance with far fewer machines and far less electricity.

Bad Bet: The “Wonder” of Coal in the 19th Century
In the mid-1800s, Scottish chemist James Young, dubbed "Paraffin Young," believed the future of energy lay in coal-derived paraffin oil. Young’s process for distilling paraffin from coal initially looked promising and profitable, setting off a paraffin boom across Scotland. However, the discovery of crude oil in Pennsylvania in 1859 rendered paraffin distillation from coal obsolete almost overnight. Young’s optimistic forecasts for coal-derived oil missed the mark, as crude oil proved cheaper, more efficient, and easier to produce. His misjudgment serves as a reminder that energy innovations can swiftly be outclassed by unexpected developments in resource extraction and technology.

Share Grid Brief

We rely on word of mouth to grow. If you're enjoying this, don't forget to forward Grid Brief to your friends and ask them to subscribe!