This shift has sparked a wave of innovation aimed at solving memory wall and scale-up compute issues—the growing gaps between compute speed, memory bandwidth and data transmission. Strategies are emerging in digital-in-memory compute, which embeds logic directly into memory chips and reduces latency, power consumption and cost by minimizing data movement, and memory disaggregation, which is the separation of memory resources from compute nodes. The other issue being addressed is connectivity, where a transition from electrical signals to photonic fabrics will be necessary to address the laws of physics and the long-term viability of electrical signals to support the highest-performance needs of AI compute. As a result, AI architectures are evolving from monolithic GPU setups to modular systems that prioritize data locality, optical signaling and memory-aware designs.
“As enterprises scale AI, they consider infrastructure as a critical factor in performance, cost efficiency and long-term competitiveness,” said Mark Edelstone, Chairman of Global Semiconductor Investment Banking. “Founders and their teams can consider systems that optimize data movement, while investors can look for companies that rethink infrastructure from the ground up—not just building smarter models.”
3. AI-Driven Personalization Is Redefining Customer Relationships
AI is ushering in a new era of hyper-personalized products, and financial services is one industry that is already leveraging AI to tailor communications and offerings that support individual customer behaviors, preferences and financial needs.
Panelists at Spark cited how AI can now determine whether a customer is more likely to respond to a call, text or email, and adjust their tone, language and timing in each of these formats. One speaker noted that AI is capturing data on individual digital experiences in real time, letting companies engage customers with more precision. Another noted how AI can surface choices that customers may want to consider—whether it’s refinancing a loan, adjusting investment allocations or exploring new benefits.
AI is also helping to embed financial services in non-financial experiences, such as e-commerce, travel and healthcare. This means that customers may receive tailored financial offers or nudges—like installment options, insurance coverage or savings prompts—when they’re most relevant. The result is a shift to personalization that is about knowing the customer and the right timing.
4. Cybersecurity at the Speed of AI
AI is reshaping the cybersecurity landscape by accelerating threats and enhancing defenses. Panelists at Spark emphasized that attackers are already leveraging generative tools to automate phishing, obfuscate malware and exploit vulnerabilities: Nearly one-sixth of attacks are already AI-generated. Companies are testing matching threat velocity with real-time, autonomous response capabilities.
Spark panelists highlighted how agentic AI is beginning to transform security operations centers (SOCs), more to close a skills gap than to replace humans. Companies are piloting agents that replicate specialized roles, such as threat hunters, detection engineers and credential managers. Some speakers cited deploying agents to simulate SOC workflows, enabling faster triage, investigation and remediation. While agents can operate at or above human speed, panelists stressed that human-in-the-loop oversight remains critical for orchestration, accuracy and accountability.
Beyond automation, the integration of cybersecurity platforms was a key theme. Security leaders noted that many platforms are still fragmented and act more as portfolios of disconnected tools rather than unified systems. As AI agents proliferate, seamless integration across vendors is becoming a top priority for founders and enterprises. Companies want modular cybersecurity architectures that support fast plug-and-play deployment, especially as AI investments facilitate the use of multiple vendors, partnerships between companies and acquisitions.
