Broadcom Inc. is holding a conference call to discuss their financial results for the third quarter of fiscal year 2024. The call is being led by Ji Yoo, Head of Investor Relations, and will include remarks from President and CEO Hock Tan, CFO Kirsten Spears, and President of Semiconductor Solutions Group Charlie Kawwas. The company's financial performance for the quarter can be found on their website, and the call is being webcasted live. Non-GAAP financial measures will be discussed, and a reconciliation between GAAP and non-GAAP measures can be found in the press release. In the third quarter of fiscal year 2024, net revenue and operating profit were up significantly compared to the previous year.

In the third quarter, the company saw strong results due to the growth of AI revenue, increased VMware bookings, and stabilized non-AI semiconductor revenue. The company has shifted from providing annual guidance to quarterly guidance as they integrate VMware. The infrastructure software segment saw a 200% increase in revenue, driven by VMware's contribution. The success of VMware's strategy was reflected in the booking of more than 15 million CPU costs of VCF and an annualized booking value of $2.5 billion. The company has also been successful in reducing costs in VMware and is on track to achieve their adjusted EBITDA goal by fiscal year 2025.

In Q3, networking revenue grew 43% year-on-year due to strong demand from hyperscalers for AI networking and custom AI accelerators. Custom AI accelerators grew 3.5 times year-on-year, while Ethernet switching, optical interconnects, and PCI Express switches also saw significant growth. Non-AI networking products also showed signs of recovery, with revenue up 17% sequentially. In Q4, total networking revenue is expected to grow over 40% year-on-year. Server storage revenue was up 5% sequentially and wireless revenue grew 1% year-on-year, with expectations for strong growth in Q4 due to the launch of next-generation devices.

In the third quarter, revenue for broadband declined by 49% and is expected to continue to decline in the fourth quarter, but to begin recovering in 2025. Industrial resales also declined in the third quarter but are expected to recover in the fourth quarter. Overall, non-AI markets have reached bottom and a recovery is expected in the fourth quarter, while demand for AI remains strong. The company is forecasting a 9% increase in semiconductor revenue and a 51% increase in consolidated revenue in the fourth quarter. The outlook for fiscal 2024 has been raised to $51.5 billion in revenue and 61.5% in adjusted EBITDA.

In the third quarter, R&D and operating expenses increased due to the consolidation of VMware, resulting in a 44% increase in operating income and a 63% adjusted EBITDA. The semiconductor segment had a 5% increase in revenue and a 270 basis point decrease in gross margins, while the infrastructure software segment saw a 200% increase in revenue and a 90% gross margin. Free cash flow was $4.8 billion, representing 37% of revenue, and excluding certain costs, it was $5.3 billion, up 14% from the previous year. The decline in free cash flow as a percentage of revenue was due to higher interest expenses and taxes.

In the third quarter, the company spent $172 million on capital expenditures and maintained a disciplined approach to managing inventory. They ended the quarter with $10 billion in cash and $72.3 billion in gross principal debt. During the quarter, they replaced $5 billion of floating rate notes with new fixed senior notes and used proceeds from a sale to reduce floating rate debt by an additional $4.2 billion. The company expects to repay $1.9 billion in fixed rate senior notes in the fourth quarter. In the third quarter, they paid $2.5 billion in cash dividends and eliminated 8.4 million shares through the payment of withholding taxes. For the fourth quarter, the company expects consolidated revenue of $14 billion and adjusted EBITDA of approximately 64%. They also anticipate a 100 basis point decrease in gross margins due to a higher revenue mix of semiconductors and product mix within semiconductors.

In paragraph 7, the speaker discusses the impact of higher taxes, restructuring, and integration costs on GAAP net income and cash flows in Q4. They also mention resuming guidance for fiscal 2025, which has been a transition and integration year following the VMware acquisition. The call is then opened for questions. In response to a question, the speaker clarifies the mix of compute and networking in AI revenue for Q3 and expects a similar trend for Q4. They also express confidence in strong growth for AI revenue in fiscal 2025. The next question addresses a shift in revenue from cloud service providers to enterprise in the AI industry.

The speaker is responding to a question about the potential impact of the shift towards cloud customers on their revenue outlook. They clarify that they do not focus on the enterprise AI market and instead primarily work with hyperscalers, cloud providers, and digital natives. They attribute their success in the VMware business to converting customers to subscription licenses and expect this trend to continue in the fourth quarter and into 2025.

The speaker discusses the current state of the company's directional trend and guidance for Q4 '24. They mention seeing accelerated bookings and growth. The next question comes from Stacy Rasgon, who asks about the semiconductor and software segments. The speaker explains that the semiconductor segment has gone through a down cycle but is starting to see signs of improvement. The software segment, excluding VMware, is back to its pre-Brocade levels and is expected to continue growing at a similar rate.

In the non-AI semiconductor market, there has been a 20% increase in bookings in Q3, indicating a recovery from the previous down cycle. While the timing of recovery may vary across end markets, enterprise and broadband sectors are showing signs of improvement. The company believes that the current uptick will continue into 2025 and could potentially lead to an up cycle surpassing previous ones, driven by the need for hardware upgrades to support AI workloads. The company expects the lowest point of the uptick to be in 2024.

Brocade's software revenue has reached a stable level, but the company is focused on the growth of Vmware in the next year and a half. In the third quarter, the company's AI revenue was in line with expectations, and the fourth quarter forecast has led to an increase in full-year AI revenue guidance. The growth in AI revenue next year is expected to come from hyperscalers, cloud, and digital natives, and is based on backlog. The company is not providing guidance beyond the backlog, so it is unknown if there will be any new customers.

The operator introduces the next question from Karl Ackerman regarding the relocation of IP back to the US and a potential $4.5 billion tax liability. Kirsten Spears clarifies that the relocation was not related to any asset sales and the charge was offset by a deferred tax liability. The operator then introduces a question from Timothy Arcuri about the growth rate of AI revenue compared to GPU compute. Hock Tan asks for clarification and the question is repeated.

The question is whether the growth rate of the company's AI revenue will be different from the growth rate of GPU compute next year. The CEO states that it is difficult to answer because they do not play in the enterprise market and it is better to ask the companies that do. However, they are focused on serving large hyperscalers and believe that these companies will eventually create their own custom accelerators for AI workloads. This is a different trajectory from the merchant GPU market and there is a transition happening.

The speaker discusses the profitability of VMware, stating that the company's operating margin is expected to grow as revenue increases and expenses continue to decrease. They also mention that the software gross margin has improved since the acquisition of VMware and is expected to continue increasing as revenue reaches $4 billion and beyond.

Hock Tan, CEO of Broadcom, discusses the company's software gross margin and its relevance to their business. He also mentions the contribution of custom AI revenue and the company's recent acquisition of Seagate's hard disk drive SoC assets. He emphasizes the importance of meaningful shipments to customers in the emerging AI market.

Seagate believes that hard disk drive storage will continue to be a sustainable and important storage medium for hyperscalers in the long term. They have collaborated with another company to work on developing technology that will increase the capacity of hard disk drives to 50 terabytes within the next five years. Seagate also plans to participate in NVIDIA's upcoming Blackwell product cycle, potentially in the optical side.

The speaker is not directly involved in NVIDIA's product roadmap, but is open to sharing their technology for use in NVIDIA's Blackwell product. They are not currently focused on M&A, as they are focused on transforming the business model of VMware.

Hock Tan, CEO of VMware, is focused on accelerating the deployment of private cloud in large enterprises. He believes it may take another year or two to complete this transformation. In the second quarter, there was an increase in R&D investments by AI customers, leading to follow-on wins for their next-generation XPU ASIC programs. There has also been an increase in demand for XPUs and networking, which may be greater than the current supply. This trend is expected to continue in 2025.

The speaker discusses the trend of AI compute moving towards ASICs and XPUs instead of general purpose GPUs. They mention that there is a high demand for these specialized chips and that the company is able to meet this demand. The speaker also addresses the question of whether more companies will move towards custom ASICs for specific workloads, and acknowledges that in the past they believed general purpose merchant silicon would prevail, but now it seems that specialized chips are becoming more popular.

The speaker has changed their view on AI accelerators and now believes that there are two markets for them: one for enterprises and one for cloud hyperscalers. The hyperscalers have the resources to create their own custom accelerators and are doing so to gain control over their platforms and enable training of large language models. This process takes time, but the hyperscalers are catching up to the biggest player in the market, who has been doing this for seven years.

Broadcom, a technology company, is predicting that hyperscalers and platform companies will create their own large language models and train them on custom silicon, or XPUs. This will lead to an accelerating curve in the market for merchant silicon, with demand from hyperscalers rivaling that in the enterprise. Broadcom will be presenting at upcoming conferences and plans to report its earnings for the fourth quarter and fiscal year 2024 in December.

This summary was generated with AI and may contain some inaccuracies.