Israel stocks higher at close of trade; TA 35 up 0.62%

Investing.com – Greece stocks were lower after the close on Friday, as losses in the Telecoms, Household and Basic Resources sectors led shares lower.

At the close in Athens, the Athens General Composite declined 0.85%.

The best performers of the session on the Athens General Composite were Fourlis Hld (AT:FRLr), which rose 3.76% or 0.16 points to trade at 4.41 at the close. Meanwhile, Alumil (AT:ALMr) added 3.68% or 0.20 points to end at 5.64 and Ilyda SA Class R (AT:ILYr) was up 3.01% or 0.15 points to 5.14 in late trade.

The worst performers of the session were Elvalhalcor Hellenic Copper and Aluminium Industry SA (AT:ELHA), which fell 3.74% or 0.18 points to trade at 4.63 at the close. Titan Cement International SA (AT:TITCr) declined 3.57% or 2.10 points to end at 56.70 and Lavipharm SA (AT:LPHr) was down 2.76% or 0.03 points to 1.20.

Falling stocks outnumbered advancing ones on the Athens Stock Exchange by 79 to 39 and 17 ended unchanged.

Gold Futures for April delivery was down 4.93% or 264.16 to $5,090.64 a troy ounce. Elsewhere in commodities trading, Crude oil for delivery in March unchanged 0.00% or 0.00 to hit $65.42 a barrel, while the April Brent oil contract rose 0.03% or 0.02 to trade at $69.61 a barrel.

EUR/USD was down 0.66% to 1.19, while EUR/GBP unchanged 0.03% to 0.87.

The US Dollar Index Futures was up 0.55% at 96.67.

Investing.com -- Large AI buyers are rethinking infrastructure decisions as models shift from experimentation to mass-market production, a transition that investors say is redefining the economics that govern the sector. 

In comments to Investing.com, Natalie Hwang, founding managing partner of Apeira Capital, said the industry is entering a period where “the economics of running intelligence at scale,” not model size, will determine competitive advantage.

Hwang believes inference is becoming the dominant cost driver, forcing companies to reassess how they scale. 

“Training costs tend to be episodic and front-loaded, whereas inference introduces a continuous operating expense that scales directly with usage,” Hwang commented. “For companies targeting mass-market deployment, this shifts the focus from headline model capability to unit economics — cost per inference, power efficiency, and system utilization.”

She added that companies dependent on inefficient deployment architectures may see operating costs rise faster than revenue, with teams that design around inference economics early, optimizing for power, latency, and total cost of ownership, seen as better positioned to scale sustainably rather than simply scale quickly.

Furthermore, power availability is a key determinant of where AI infrastructure is built. 

Hwang feels that location is increasingly important “not for proximity to users alone, but for proximity to power,” warning that regions with constrained grids, slow permitting, or high energy costs could face structural challenges.

“Over time, we expect infrastructure decisions to be shaped as much by energy availability and deployment feasibility as by network connectivity or real estate considerations,” Hwang remarked.

The investor said the companies best positioned for this transition are those treating AI as an operating system and designing around deployment constraints early, including power, cost, latency, and reliability. 

“What increasingly matters is whether the system can operate efficiently and predictably under real-world conditions,” Hwang said.

This period is described as the industrialization of AI, where physical limits and unit economics will dictate capital allocation and defensibility. 

“This shift is reorganizing capital allocation and competitive advantage,” she said, noting that many of these forces are not yet reflected in market expectations.

Investing.com -- Advanced Micro Devices (NASDAQ:AMD) stock declined Friday, falling as much as 6% before paring losses to around 3%, following reports that the company’s next-generation MI450-series AI accelerators might face delays.

The downward pressure came after SemiAnalysis reportedly claimed AMD’s MI450-series chips were behind schedule. However, Wells Fargo analyst Aaron Rakers quickly countered these assertions, suggesting the claims were unfounded based on his team’s research.

According to Rakers, AMD’s development on TSMC’s N2 (2nm) process technology remains on track, with the company having already completed tape-out and shipped Venice EPYC processor samples that use the same process node as the upcoming MI450-series. The N2 process is crucial for AMD’s next-generation AI accelerators.

Wells Fargo expects AMD to reaffirm its confidence in the MI450-series launch timeline during the company’s upcoming earnings report next week. The chipmaker has previously indicated that volume production would commence in the second half of 2026, with significant shipments weighted toward the fourth quarter of that year to customers including OpenAI and Oracle.

Despite Friday’s stock decline, Rakers maintained a bullish stance on AMD, suggesting investors view the weakness as a buying opportunity. The analyst anticipates AMD will report server revenue exceeding $38 billion and remains focused on the company’s path toward achieving more than $35 billion in data center GPU revenue by 2027.

Investing.com -- Citizens downgraded SAP to Market Perform from Market Outperform after the company posted weaker-than-expected top-line results for the fourth quarter and signaled slower cloud momentum into 2026.

SAP reported non-IFRS EPS of €1.62 and operating profit of €2.83 billion, both ahead of consensus, but revenue missed forecasts. Total revenue came in at €9.68 billion versus expectations of €9.75 billion, while cloud revenue of €5.61 billion also fell short. More importantly, current cloud backlog growth slowed to 25% at constant currency, below expectations and the slowest pace in nine quarters.

Citizens said SAP cited a more pronounced slowdown than anticipated, driven by a deal mix skewed toward larger transformation projects with longer ramp periods and flexible structures. The firm also pointed to rising geopolitical tensions, which are pushing customers to consider sovereign SaaS options. The shares fell about 15% following the results and are down roughly 18% year to date.

Guidance for 2026 also came in largely below expectations on an as-reported basis due to foreign exchange. SAP guided to cloud revenue of about €25.3 billion to €25.7 billion, below consensus of €26.0 billion, and cloud and software revenue of €35.6 billion to €36.1 billion versus €36.3 billion expected. Non-IFRS operating profit guidance of €11.5 billion to €11.9 billion was roughly in line, while free cash flow of €10 billion was slightly ahead of consensus. SAP also said constant-currency cloud backlog growth will slightly decelerate in 2026 from 25% in 2025.

While Citizens said SAP remains a high-quality business with strong leadership and long-term potential in AI and the Business Data Cloud, it struck a more cautious tone heading into 2026. The firm said SAP appears to be past the peak of the S/4HANA upgrade cycle as a growth driver, while the Business AI cycle is still early and may require changes to products, implementation, pricing and support to scale meaningfully.

Citizens cut its 2026 non-IFRS EPS estimate to €6.84 from €6.87 and lowered its 2027 forecast to €7.41 from €8.00, both below consensus. It also introduced a 2028 EPS estimate of €8.08. The broker said SAP looks fairly valued at about 4.5x 2027 EV/revenue and 17x 2027 EV/free cash flow, broadly in line with peers.