Investors should look to Broadcom (AVGO) as leading AI labs adopt custom application-specific chips like the Jalapeno processor to significantly lower computing and energy costs. While NVIDIA (NVDA) maintains dominant market share with strong forward quarterly revenue guidance of $108 billion, investors should closely monitor its profit margins as custom silicon alternatives enter the market. Tesla (TSLA) offers high-margin growth potential as it scales its autonomous Cybercab / Robotaxi network, targeting 2,500 commercial vehicles in Las Vegas over the coming year. Allocate capital to the Nuclear Energy and Small Modular Reactor (SMR) theme, particularly HALEU uranium enrichment suppliers, to profit from the surging power demand of AI data centers ahead of 2030. Finally, Apple (AAPL) provides an attractive edge-computing play as its Mac Studio M5 Ultra enables privacy-sensitive enterprises like healthcare and legal firms to run frontier AI models locally instead of paying expensive cloud subscription fees.

NVIDIA Corporation (NVDA)

• NVIDIA reported $96.2 billion in quarterly revenue (up 106% year-over-year) with forward guidance of $108 billion (over $1 billion per day) and long-term 2028 growth guidance of 70%.
  • Gross margins remain exceptionally high at 80% to 85%, driven by near-total dominance in AI model training clusters.
  • The company faces key structural vulnerabilities, notably 100% reliance on TSMC for fabrication and growing questions around circular financing or balance-sheet-backed customer demand.
  • The software moat around CUDA is diminishing for AI inference workloads as competitors port models to alternative silicon, shifting NVIDIA's primary competitive advantage toward networking interconnects (Mellanox) and mega-cluster coherence.

Takeaways

• NVIDIA remains the undisputed king of AI training infrastructure with sold-out capacity, but investors should monitor gross margin sustainability as custom inference chips emerge and customer concentration risk around mega-hyperscalers increases.

SpaceX (Private)

• Elon Musk issued forward projections that SpaceX could reach $3.5 trillion in annual revenue by 2033 (up from previous estimates of $1 trillion by 2029–2030).
  • Growth drivers discussed include the deployment of physical infrastructure: terrestrial and orbital data centers (Dyson swarm concepts), satellite communications via Starlink, and potential multi-trillion-dollar revenue from the Optimus humanoid robotics platform.
  • Long-term lunar industrialization (fabbing AI chips and orbital compute infrastructure directly in low-gravity environments) could unlock massive economic value enabled by Starship's heavy launch capabilities.

Takeaways

• For investors with access to private secondary markets, SpaceX represents a broad bet across AI physical infrastructure, space logistics, orbital computing, and robotics.

Tesla, Inc. (TSLA)

• Tesla is expanding commercial operations for its autonomous Cybercab / Robotaxi network, registering 79 Model Y units in Texas and targeting 2,500 vehicles in Las Vegas over the coming year.
  • The long-term target is deploying 1 million robotaxis in commercial service, which could generate an estimated $100 billion in high-margin rideshare revenue.
  • Tesla's pure-vision autonomous approach (utilizing 8 cameras and end-to-end neural networks rather than costly LIDAR suites) provides a massive hardware manufacturing cost advantage over competing fleets like Waymo.

Takeaways

• The primary catalyst for Tesla remains the transition from automotive hardware sales to recurring, high-margin software and autonomous fleet revenue per mile through Full Self-Driving (FSD) and commercial Cybercab fleets.

Broadcom Inc. (AVGO) & Custom AI Silicon

• OpenAI announced performance figures for its custom in-house AI inference chip, Jalapeno, co-developed with Broadcom.
  • Jalapeno delivers 1.5x to 1.9x more compute per watt and up to 3.6x lower end-to-end latency compared to NVIDIA GB200/GB300 systems, running at 700 watts versus 1,400 watts.
  • On open-source models such as GPT-OSS, the chip achieved an estimated 54x throughput increase per user, highlighting how hyperscalers and frontier AI labs are designing custom Application-Specific Integrated Circuits (ASICs) to bypass standard GPU costs for inference.

Takeaways

• Custom ASIC design partners like Broadcom (AVGO) stand to benefit significantly as leading AI frontier labs diversify away from standard GPUs toward custom silicon optimized specifically for inference workloads.

Apple Inc. (AAPL)

• Apple introduced the Mac Studio featuring the M5 Ultra processor with up to 512 GB of unified memory, alongside the 2-nanometer M6 chip.
  • The high unified memory architecture allows enterprises to run large frontier-class AI models locally on desktop hardware, shifting economics from perpetual cloud API token fees to a one-time capital hardware purchase.
  • Key target enterprise markets include privacy-sensitive sectors such as healthcare (HIPAA compliance) and legal services that cannot send proprietary data to third-party public clouds.

Takeaways

• While Apple continues to face criticism regarding consumer-facing AI software integration, its unified memory hardware positions it strongly to capture the growing niche for on-premise, secure enterprise edge computing.

Alibaba Group Holding Limited (BABA)

• Alibaba launched WAN 3.0, an AI video generation model capable of turning spreadsheets, documents, and slides into 30-second videos with API costs collapsing to $0.05 to $0.20 per second.
  • This product release coincided with a $10.2 billion private follow-on offering in Hong Kong.
  • Video and world model generation now represents approximately 70% of total AI token consumption in China, signaling a distinct strategic focus on robotics, simulation, and commerce video versus pure text LLMs.

Takeaways

• Chinese tech giants like Alibaba (BABA) are aggressively driving down the unit economics of generative video and spatial world models, offering low-cost media production APIs that could disrupt traditional digital content creation pipelines.

Nuclear Energy & Small Modular Reactors (HALEU / SMR Theme)

• Advanced nuclear startup Actinide demonstrated the ability to enrich uranium into High-Assay Low-Enriched Uranium (HALEU) up to 15.38% purity (compared to traditional commercial levels of 3% to 5%).
  • Fuel availability is the primary critical-path bottleneck for next-generation Gen 4 fission and Small Modular Reactors (SMRs) scheduled for commercial deployment by 2030.
  • Surging baseload power demand from AI data centers is forcing cloud providers and utilities to secure dedicated nuclear and clean energy supply chains.

Takeaways

• The intersection of AI data center power demand and advanced nuclear infrastructure creates long-term secular tailwinds for uranium enrichment technologies, HALEU fuel suppliers, and venture-backed SMR developers heading toward public markets.