-+ 0.00%
-+ 0.00%
-+ 0.00%
Palo Alto Networks (PANW.US) Q1 Conference Call: Net increase of about 170 platform-based deployments in a single quarter! The number of AI vulnerabilities discovered is close to 7 times that of the previous year, and companies are starting to use emergency budgets
Share
Listen to the news

The Zhitong Finance App learned that cybersecurity supergiant Palo Alto Networks (PANW.US) made a major disclosure during a conference call in the first quarter of fiscal year 2027 that could explain changes in demand more than revenue growth: companies have used emergency budgets to test whether their systems can block attacks with cutting-edge AI. Management has observed that the number of major and serious vulnerabilities discovered each month is currently close to 7 times the monthly average of the previous year.

Palo Alto management said that global enterprises currently face two risks: if the vulnerability is left unattended, attackers may take advantage of it first; installing patches in a hurry may also cause production systems to be disrupted. This explains why customers are willing to purchase virtual patches first, gain time for formal fixes, and then integrate testing, protection, and response tools previously scattered across multiple vendors. There was a net increase of about 170 platform-based deployments this quarter, more than the first quarter of the previous two fiscal years combined. Customers who have already adopted the platform contribute approximately two-thirds of next-generation security's annual recurring revenue, with a net revenue retention rate of over 120%.

The AI agent frenzy, which focuses on efficient AI agent-based workflows, is changing the billing logic of security software, but there is a time lag between traffic growth and revenue recognition. Intelligent traffic on the Palo Alto Networks (i.e. Palo Alto Networks) Secure Access Service Edge (SASE) platform has reached about 9 times that of the same period last year. Since overall customers are still within the purchased traffic quota, management estimates that it will take 6 to 9 months before the excess traffic will begin to contribute incremental revenue. Prisma AIRS's business model is more straightforward: the more tokens (tokens) an enterprise uses, the more words need to be checked. The product has only been in commercial use for five quarters, has more than 1,000 customers, and annual recurring revenue is close to US$170 million. Management believes that the first thing that is needed is programming agent security. Businesses already make extensive use of programming tools, yet they are often unable to answer a basic question: when a tool creates a new agent on its own, who limits what systems it can access and what actions it can perform?

The proposed acquisition of Applied Compute with all shares of $4.5 billion is aimed at areas that are difficult to cover by traditional security rules. Management advocates that the same team train a supervisory model that understands its expected behavior while doing post-training for the enterprise's business model. In this way, in addition to intercepting known attacks, the security system can also determine whether the agent deviates from the tasks approved for execution.

The company's management expects this business to contribute approximately $1 billion in next-generation security annual recurring revenue by fiscal year 2029. Computing power will be allocated in stages according to demand, and priority will be given to using cloud partners and existing contracts in the early stages. The cumulative related capital expenditure up to FY2030 is expected to be less than 3 billion US dollars. If the transaction is completed as scheduled in the third quarter of FY2027, it is expected to reduce operating margins by up to 1 and 2 percentage points in FY2027 and 2028, respectively, and return to 29% to 30% in FY2029. This increase in full-year results guidance has not yet been factored into this acquisition.

The following is the full text of the Palo Alto Networks performance conference call (artificial intelligence tool assisted translation; Pietuo Network is the Chinese translation of Palo Alto Networks):

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: Hello, welcome to Pioneer Network's FY2027 first quarter results conference call. I'm Hamza Fordwala, senior vice president and head of investor relations. Please note that today is Wednesday, November 18, 2026 at 1:30 p.m. Pacific time, and this conference call is being recorded. I am discussing the results of the first fiscal quarter with Chairman and CEO Nikish Arora, and Chief Financial Officer Deepak Golecha. After the pre-prepared presentation, Chief Product and Technology Officer Lee Claridge will join the Q&A session. You can read the press release and additional information about this conference on the company's investor relations website at investors.paloaltonetworks.com. Please click on the quarterly results link on the website to view additional financial information and results presentations for the first quarter of fiscal year 2027.

Today's conference will present forward-looking statements and predictions about the company's business operations, financial performance, and recent acquisitions. The statements made today involve a number of risks and uncertainties, and actual results may differ from those described in the forward-looking statements. For an explanation of these risks and uncertainties, please refer to our press release and recent documents submitted to the U.S. Securities and Exchange Commission. We are under no obligation to update any forward-looking statements in today's meeting. The presentation also included non-GAAP financial indicators and key indicators relating to the company's past performance and expected future performance. Non-GAAP financial measures cannot replace financial indicators prepared in accordance with US GAAP. The most directly comparable US GAAP financial indicators, as well as reconciliation statements between the two calibers, are included in the appendix to the press release and investor presentation.

Unless otherwise specified, all results and comparisons are based on a year-on-year basis for the same period of the fiscal year. I now give the floor to Nikos.

Nikish Arora, Chairman and CEO of Pietro Networks: Thank you, Hamza. Good afternoon everyone. As you can see, we had a strong start to the first quarter of this fiscal year. All indicators exceeded company guidelines. Next Generation Security Business Annual Recurring Revenue (NGS ARR) grew 66%, exceeding guidelines, the highest in the company's history. Total revenue increased 39%, and remaining performance obligations (RPO) increased 35%, both faster than in the previous quarter. Achieving such results in the seasonally poor first fiscal quarter shows strong demand. As artificial intelligence accelerates cyber attacks, enterprises are urgently replacing disparate tools with a unified platform. This momentum gives us confidence that the full-year guidance will be raised substantially. Strong revenue performance also directly translates into profit. In the first quarter, our operating margin once again exceeded 30%. While integrating large-scale acquisitions, our operating profitability reached a new high.

What's most exciting is that demand is generally strong across all platforms, and revenue growth is accelerating without exception. First, let's talk about the largest network and AI security business. This business grew 20%, and secure access service edge (SASE) and software and hardware firewalls all performed well. Customers are moving away from a point product model. They realized that relying on multiple vendors to individually defend network boundaries would leave gaps in the two critical aspects of clearly understanding network activity and implementing security strategies. We are also seeing a fundamental change in the nature of traffic carried by the network. Autonomous agents are continuously connected to applications, application application interfaces (APIs), and enterprise data. These interactions require real-time inspection. Agent traffic on the SASE platform has reached about 9 times that of the same period last year. The surge in traffic reflects that companies are putting AI into production environments. The need to protect this new attack surface also drove Prisma AIRS to achieve annual recurring revenue (ARR) of nearly $170 million in five quarters after official commercialization, making it the fastest-growing product in Pietuo Network's history.

Driven by XSIAM and observability business, Cortex grew by more than 50% in the first quarter. AI-driven attacks unfold at machine speed and rely on ex post facto remediation measures scattered across multiple data lakes, and do not constitute an effective defense solution. We need solutions that can perform real-time analysis, provide observability, dynamically correlate different signals, and provide correlation results and threat determination in real time. XSIAM currently accesses more than 20 petabytes (PB) of telemetry data every day, using AI to reduce response time from days to less than 10 minutes. Also, look at the identity security platform Idira. Based on test preparation standards, revenue increased 23% in the first quarter. Since completing the CyberArk acquisition earlier this year, we have been advancing our consolidation plans. In the first quarter, Idira had a net increase of more than 100 customers from Pietuo's existing customer base, making it the quarter with the best cross-selling performance between the two sides so far.

The rise of AI agents has brought a new class of non-human identities that require real-time protection. Effective monitoring of these identities requires full control over their activities and immediate revocation of access when suspicious activity is discovered. Precisely to solve this problem, we officially launched Idira's Agentic IdP (Agentic IdP) in the first quarter and integrated it directly into Prisma AIRS' AI gateway to execute agent permission policies in real time. Customers use our three platforms to solve the same core problem, which is to shorten the time between discovering safety hazards and eliminating them. Attacks advance at machine speed, and single-point products simply can't keep up. Businesses are integrating security architectures into a unified platform, and this transformation continues to drive platformization. Yes, that's another word, platformization. There was a net increase of about 170 platform-based deployments in the first quarter, setting a record, exceeding the first quarter of the past two fiscal years combined.

Customers using platform-based deployments now contribute two-thirds of our next-generation security business's annual recurring revenue. These customers are still rapidly expanding procurement and have a net revenue retention rate of over 120%. The quarter's biggest deals reflect the momentum of platformization. In the first quarter, we signed a $83 million agreement with a global financial institution. This customer integrated cybersecurity into our platform, adopted Frontier's virtual patch service, expanded identity security procurement, and signed the largest order ever for CyberArk. A global consulting firm is putting AI into production company-wide and has signed a $46 million deal with us, including Prisma AIRS's largest order ever. Finally, we signed a $20 million agreement with a global information technology services company to help it unify its security technology stack, reduce too many scattered point products, and upgrade its defense system. The customer purchased Prisma AIRS and SASE to protect its AI deployment and network access for employees around the world.

Customers choose us to make long-term decisions about their technology architecture. Every time a customer gives us an extra commitment, our responsibility becomes even greater. They want us to keep refining the products they operate on today, and they want us to respond to the next round of changes ahead of time. Six months ago, with the first release of models with cyber attack and defense capabilities such as Mythos, we predicted that equally powerful aggressive AI capabilities would rapidly spread through the open source ecosystem within three to six months, and now this prediction has come true. I want to make it clear that the current situation is being driven simultaneously by two different effects. First, there has been a surge in the number of vulnerabilities discovered and the number of patches that need to be deployed. The number of major and serious vulnerabilities discovered so far is close to 7 times the monthly average of last year. Vulnerabilities were discovered so quickly and so intensively, that IT teams were struggling.

We believe that many of the major service outages that have occurred in the past few weeks, including system outages for large enterprises and government agencies, were partly due to unintended consequences of installing patches and configuration changes. It's difficult for IT teams to fix these vulnerabilities quickly enough without impacting production environments. The second effect is the rise of AI-assisted cyber attacks. Attackers are using an open weighting model and a novel agent framework to discover and exploit vulnerabilities at an unprecedented speed and scale. New intelligence frameworks such as ARTEX allow attackers to turn simple instructions into self-executed multi-stage attacks, and only a few people can attack dozens of targets at the same time. A few days ago, our Unit 42 team published research on an attack. A small hacker gang linked to foreign forces invaded 19 institutions at the same time.

Looking ahead, we expect cyber threats to remain high, but we believe there are many ways for businesses to adapt to this change. First, companies will still need to deploy a large number of patches for at least the next 18 months, and the risk of collateral damage caused by this will continue. We expect this collateral damage to gradually decrease as customers become better at deciding what exactly to fix and when. Second, demand for virtual patches in all industries and types of software will grow. Enterprises can first establish reliable temporary protection for infrastructure, coordinate deployment arrangements, and then officially install patches. Demand for our advanced virtual patching service was strong in the first quarter, and we have close to 70,000 cybersecurity customers, so there is a great opportunity to provide them with this service.

Third, for all companies, there is no perfect way to prevent attacks, and no organization can assume that it will stay out of the situation. Attacks will happen, and defenders must be prepared to detect and dispose of threats faster than ever before; relying on piecemeal remediation measures after the fact is not enough. Therefore, we believe that the third most important of the three countermeasures is to upgrade traditional security and information technology infrastructure worth more than 1 trillion dollars to a real-time platform. Despite the growing threat, the market demand for AI is still strong. Every time the cost of inference drops, thousands of new agent workflows become viable. According to this trend, the industry's term revenue is expected to exceed $1 trillion by 2029. The market is focused on AI-themed transactions. We are focusing on the turning point of AI applications, that is, the industry is shifting from building AI infrastructure to actual use of AI by enterprises.

Until recently, AI activities were mainly focused on general applications across industries such as consumer scenarios and programming. In these scenarios, there was almost no security protection for related terms. Looking ahead, we believe that the AI infrastructure that has already been deployed will also begin to reveal vulnerabilities. Recently, we have seen the vulnerabilities of AI infrastructure providers being exploited. As programming tools increasingly perform tasks autonomously, inadequate security controls can also bring agent risk into the enterprise environment. AI applications and AI security must advance simultaneously. We're already using Prisma AIRS to help customers bridge this gap. Over the past five quarters, the product has had more than 1,000 customers, and annual recurring revenue is close to $170 million. Today, Prisma AIRS provides comprehensive AI security capabilities in five areas, allowing enterprises to grasp, evaluate, and control all of their AI deployments on a single platform. The first is AI asset discovery and AI risk assessment.

First, can you find every AI asset within the company? Prisma AIRS can discover models and autonomous agents operating in any location, and then continuously assess their risks in terms of supply chain, security posture, and automated AI Red Team testing. The next step is to execute policies in real time through AI protection and AI governance. Can you focus all relevant traffic on the same policy execution point? With collaborative run-time security, identity security, and AI gateways, customers can check for prompts, prevent sensitive data leaks, protect agent identities, and execute policies in real time with every interaction. This is the industry's first AI security platform. What links these links in tandem is AI observability. It collects and monitors every reminder, every response, and every log, so companies can ensure that all actions are traceable, even when using AI on a large scale. Prisma AIRS currently covers an important aspect of AI safety, but we expect that as the model becomes more intelligent, the manually set static control methods will eventually fail.

Therefore, AI security models require advanced reasoning capabilities that are comparable to cutting-edge models. For people to trust big language models, we must be able to independently verify, interpret, and audit the actual deployment of AI. We believe that trust and control are critical to driving the adoption of AI by enterprises. We've seen that companies' demands for autonomy and control are driving them to adopt open weighting models and open source models. There are clear early signs of this transformation. Over the past year, the share of the open weight model in term traffic has risen from less than 5% to more than 60%. When companies place AI at the core of their operations, they want to keep proprietary data, the organization's accumulated judgment ability, and competitive advantage firmly internal. We anticipate that as open weighting models and open source models become popular in enterprise applications, the market opportunities for model inference services will reach hundreds of billions of dollars in the future.

Enterprises can both trust AI and control how AI is deployed. They can train models themselves and run models locally or on a sovereign technology stack controlled by themselves, which will speed up the adoption of AI by enterprises. This means that models must be delivered safely, in a way that guarantees the company's autonomy when needed, and that control must be completely in the hands of the enterprise. As we study how to train our own customized open weighting models, we are looking for partners across the AI ecosystem, hoping to find teams with excellent engineering capabilities and proven track records to solve enterprise problems. Applied Compute was ultimately selected because no other team could match its technical accumulation in training after opening up the weighting model. We are incorporating a world-class cutting-edge AI research team into the company and expect to have 100 researchers by the end of the year. Every company has decades of accumulated proprietary data, internal operating procedures, and organizational judgment experience, none of which can be disclosed to the public model.

Until now, however, these companies have lacked trusted partners to help them safely and autonomously post-train, operate, and monitor open weighting models. We believe Applied Compute is the right partner for us to seize this opportunity. The team has already won the favor of large enterprises and AI-native companies. After only 18 months of establishment, the annualized revenue has exceeded 100 million US dollars, and there are only a few companies that can achieve this. Applied Compute will help us move in two directions. First, embed AI into all security capabilities. We are training models for over 35 specific scenarios and providing these capabilities to customers so they can use AI against AI. Second, we will help enterprises train open weight models and provide model inference services in a safe and controlled manner.

Once the open weighting model is deployed, we can deliver a secure open weighting model to our customers, so that enterprises can use AI safely. To support this work, we plan to build diversified global reasoning capabilities over 100 megawatts over the next three to five years. To this end, the cumulative capital expenditure we plan to invest as of fiscal year 2030 is relatively limited, less than $3 billion, which is about equivalent to the free cash flow we have generated in half a year. Combining their platform with our global market reach and the trust we've earned among CIOs around the world, we expect Applied Compute to continue to rapidly scale up, contributing over $1 billion in annual recurring revenue to our next-generation security business by fiscal year 2029. Finally, the first quarter was a decisive turning point for the cybersecurity industry.

AI-driven attacks have left companies no room for gradual improvement in a relaxed manner. Facts have proven that using decade-old, isolated tools cannot stop autonomous attacks, and traditional infrastructure worth 1 trillion dollars is in urgent need of upgrading. At the same time, for the entire enterprise to use AI with confidence, AI security must evolve from clearly understanding activities and implementing strategies to enabling enterprises to trust the model itself. Without confidence that models behave reliably in production environments, it is difficult for companies to confidently hand over mission-critical workflows to AI. Finally, as autonomy and control become increasingly important in businesses, helping businesses safely adopt an open weighting model will be a huge opportunity. Over the next few years, we will deliver this basic capability to customers along with the core security platform, and play a central role in the customer's security system upgrade and AI transformation.

Let's ask Deepak to provide more in-depth financial details and explain the updated performance outlook.

Deepak Golecha, Chief Financial Officer of Pietuo Networks: Thank you, Nikish. Good afternoon everyone. We delivered record results for the first fiscal quarter, and all key metrics for measuring revenue and future revenue are growing at an accelerated pace. Remaining performance obligations (RPO) increased 35% year over year to reach 21 billion US dollars. Of these, the portion of revenue confirmed over the next 12 months, that is, current RPO, increased 36% year over year to reach 9.4 billion US dollars. The contract period remained the same as the same period last year. The next-generation security business's annual recurring revenue (NGS ARR) reached $9.73 billion, up 66% year over year, and the net increase in NGS ARR was more than double that of the same period last year. NGS ARR significantly exceeded guidelines, ranking among the highest in the company's history in every quarter. Total revenue for the first fiscal quarter increased 39% year over year to reach US$3.44 billion. As announced last quarter, we are now disclosing revenue in three businesses: Network and AI Security, Cortex, and Idira. Revenue growth in the first fiscal quarter was indeed quite extensive, and the growth rate of all three businesses accelerated. The revenue volume of the network and AI security business is already quite impressive. Following strong performance in the fourth fiscal quarter, growth accelerated further in the first fiscal quarter.

The business's revenue increased 20% year over year to $2.3 billion. Cybersecurity products have seen strong growth in all forms of deployment. Let's take a look at SASE. Its ARR increased 30% year over year, contributing more than 100 million US dollars to the net addition of NGS ARR. Demand for software-defined wide area networks (SD-WAN) and secure browsers is particularly strong, and customers continue to prefer our comprehensive SASE platform over vendors that only offer a single product. In terms of firewalls, both hardware and software performed well, and the number of signed orders over the past 12 months increased 28% year over year. Customers want to uniformly monitor network activity and implement protection, so they are integrating the vendors they use, and our market share continues to rise. Software firewall ARR accelerated growth for the third consecutive quarter, with a year-on-year increase of 34%. The main driving force comes from Prisma SASE, which is still our fastest growing product, and ARR is close to $170 million.

Let's take a look at our secure operation and observability platform Cortex. Revenue for the first fiscal quarter increased 53% year over year, and increased 40% year over year according to exam preparation standards. AI has exacerbated security threats, and customers have prioritized the modernization of security operations centers (SOC), driving XSIAM's ARR growth of more than 65% year over year to around $800 million. Furthermore, our observability platform was recently relaunched under the name Cortex XCOR and continues to grow strongly. ARR is approximately four times that of the same period last year. We mentioned last quarter that included a net increase in ARR due to the migration of a large cutting-edge AI lab customer from the original vendor to Chronosphere, which amounts to about nine digits of dollars. Identity security platform Idira's revenue for the first fiscal quarter increased 23% year over year in line with exam preparation standards. Our core sales team continues to work closely with the Idira team in developing markets.

In the first fiscal quarter, the partnership brought Idira a net addition of more than 100 customers, all from the existing customer base of our core business. This was the quarter with the highest number of customers gained through cross-selling since the acquisition was completed. We've also seen significant synergy effects on the profit side. As a reminder, the goal we set when announcing the acquisition was for Idira's operating margin to match our core business within 18 to 24 months after the transaction was completed. At current pace, the time required to achieve this goal is about half of what was originally planned. Regional performance was also generally strong, with revenue growth of 42% year-on-year in the Americas, 40% in Europe, the Middle East and Africa, and 26% in Japan and the rest of the Asia-Pacific region. Various revenue indicators for the first fiscal quarter set records, while we continue to focus on profitable growth. Adhering to this principle made the operating leverage effect more evident compared to the same period last year, and also strengthened our confidence in achieving our profit margin guidelines for the whole year.

Our scale and diversified business model helped ease the pressure on gross margins, which continued to remain around 75%. The overall gross margin for the first fiscal quarter was 75.2%, a year-on-year decrease of 170 basis points. As stated last quarter, we continue to actively address multiple factors affecting short- and medium-term gross margins. First, the cost of hardware components in the entire industry remains high, and we are taking the initiative to take advantage of our scale and use relationships with strategic suppliers to deal with it. Hardware accounts for about 10% of total revenue, but it is expected that throughout the fiscal year, higher component costs will affect gross profit margins. Second, the revenue share of some software-as-a-service (SaaS) products, which are still in the early stages of development and growing rapidly, has also affected gross profit margins.

This is particularly evident in Cortex's business. Both XSIAM and observability businesses are growing rapidly as we focus on increasing our market share. Additionally, the customer's migration from Prisma to Cortex Cloud is still ongoing, temporarily dragging down the business's profit margins. Considering platform migration and product portfolio changes, we expect Cortex's gross margin to remain below the company's overall average in the medium term and rise above 70% in the long term. In terms of operating profit margins, as the scale of our business expands and operational efficiency improves, we continue to offset the impact of rising sales costs by diluting operating expenses. Despite increased costs associated with mergers and acquisitions, the non-GAAP operating margin reached 30.7% in the first fiscal quarter, an increase of 50 basis points over the previous year. We are also using AI to improve efficiency in various business areas of the company in a targeted manner.

For example, we are using AI to assist in quoting to shorten the time required to sign an order and improve sales efficiency. Furthermore, we are also using AI to improve procurement efficiency, and have discovered significant cost savings by integrating suppliers. In terms of profit, non-GAAP diluted earnings per share were $1.05, up 13% year over year, above the guideline of $0.08. In accordance with US GAAP, earnings per share were $0.57, up 21% year over year. Increased profitability also boosted cash generation in the first fiscal quarter. Adjusted free cash flow was US$1.9 billion, and the adjusted free cash flow margin was 56%. In terms of capital allocation, we did not buy back shares this quarter, and we will still decide whether to do so depending on the opportunity. Our current share repurchase authorization is up to $1 billion and is valid until December 2027. The company's overall performance is directly due to our continued focus on innovation.

Through internal investment and targeted mergers and acquisitions, we strive to anticipate customer needs years in advance before technological transitions. As Niche said earlier, we're excited to announce our intention to acquire Applied Compute, and we hope to take the lead in providing secure AI to our customers. The transaction has a total consideration of $4.5 billion, all paid in shares, and is expected to close in the third quarter of fiscal year 2027. The acquisition covers Applied Compute's commitments to purchase cash and prepaid computing power, which together amount to approximately US$500 million. The company has grown quite well, reaching an annualized revenue of US$100 million within 18 months of establishment. Mergers and acquisitions will increase our revenue growth rate from the first day the deal is completed. We expect the business to expand to a new NGS ARR that can contribute $1 billion by fiscal year 2029.

In the longer term, we believe this will be another multi-billion dollar ARR business for the company, and it will also put us at the cutting edge of AI security. To provide customers with secure post-model training and inference services, we expect our computing power requirements to grow, but we still plan to adopt an asset-light model to maintain strong free cash flow margins throughout the capital expenditure cycle and beyond. In order to meet computing power requirements, we will continue to purchase from cloud partners and gradually increase our own dedicated computing power. The majority of this investment will go to graphics processing units (GPUs) and servers, and a smaller portion will be used to host the facilities that host these devices. We intend to expand computing power in stages based on already foreseeable needs. Overall, by fiscal year 2030, we expect related cumulative capital expenditure to be less than $3 billion, which is about half of the anticipated free cash flow for the current fiscal year.

Assuming the transaction is completed in the third quarter of FY2027, we expect the transaction to dilute operating profit margins by up to 100 basis points in FY2027, up to 200 basis points in FY2028, and then return to the current level of 29% to 30% by FY2029. Finally, from a free cash flow perspective, after excluding data center capital expenses, we expect the free cash flow margin to remain above 38% in fiscal year 2028 and beyond. Once the transaction is complete, we will provide more details, as well as more accurate modeling parameters. Next, let's talk about performance guidelines. The guidelines only cover the existing business of Pietuo Network and do not include the pending Applied Compute acquisition. For the second quarter of fiscal year 2027, we expect NGS ARR to be $10.2 billion to $10.22 billion, an increase of 61%. The RPO was $21.5 billion to $21.6 billion, an increase of 34% to 35%. Revenue was $3.53 billion to $3.55 billion, up 36% to 37%.

The number of fully diluted shares is expected to be between $837 million and $842 million, with non-GAAP earnings per diluted share of $1.01 to $1.03. For the full fiscal year 2027, we expect NGS ARR of $11.475 billion to $11.52 billion, an increase of 26% to 27%. The RPO was $25.3 billion to $25.5 billion, an increase of 19% to 20%. Revenue was $14.52 billion to $14.57 billion, up 26% to 27%. The estimated operating margin is 29.5%, non-GAAP diluted earnings per share of $4.26 to $4.30, fully diluted shares of 844 million to 847 million shares, and an adjusted free cash flow margin of 38%. Now hand it back to Hamza and move on to the question and answer session.

The following is the performance conference Q&A session (AI-assisted summary and translation):

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you, Deepak. The first question was asked by Saket Kalya of Barclays, followed by Brian Essex of J.P. Morgan Chase.

Barclays Analyst Saket Kalya: OK, great. Hello everyone, thank you for the opportunity to ask questions and congratulate you on starting the new fiscal year on such a strong start. Glad to see such results. Nikish, you've been talking about that trillion-dollar technology debt for a while. The issue seems even more pertinent in light of recent service outages. Based on your recent conversations with customers, where will CEOs and CIOs focus on modernization in 2027 as the new wave of cyber threats is clearly more active?

Nikish Arora, Chairman and CEO of Petro Networks: Thank you for your question, Saket. First, we see that customers are very interested in Frontier AI Defense's cutting-edge AI defense services. Almost every company is carrying out penetration tests and red team tests on their infrastructure to ensure that they use these new AI capabilities to test their defenses before attackers use them. As mentioned earlier, the difficulty is that new vulnerabilities are discovered every day, so many new bugs are discovered that companies simply can't fix them all. I think some companies are still busy dealing with this short-term crisis. Others are beginning to consider how this will affect its long-term structure. You can also see this from the platform integration we talked about.

We've completed 170 platform integrations this quarter. In other words, customers are saying, “Listen, I can't manage so much fragmented infrastructure, how can I integrate them?” The difficulty is that no matter what plan is adopted today, it will take three to six months, or even nine months, to implement, depending on how complex and old the current environment is. Everyone is starting to change, and in the foreseeable future, we'll see more companies doing this.

Barclays Analyst Saket Kalya: Very helpful, congratulations again.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pietuo Networks: OK, thank you, Saket. Next, ask questions from J.P. Morgan Chase's Brian Essex, followed by Rob Owens of Peggy Securities.

J.P. Morgan Analyst Brian Essex: Hello, thank you Hamza, and thank you for giving me the opportunity to ask questions. Niche, I think this acquisition is very interesting. What communication between you and your customers has sparked your interest in sovereign AI? Can you talk about the details of the exchange and how are you preparing to make AI autonomous and controllable?

Nikish Arora, Chairman and CEO of Pietro Networks: Brian, that's a great question. We started by developing Prisma AIRS. It allows you to clearly see what AI assets are within the company, and they are presented in front of you one by one. This step is easy to do; it's basically a matter of certainty. Assets are individual objects and can be found. Next, we wonder if we can gather all traffic and set up a node that uniformly enforces security policies to stop these malicious activities? We have identified common methods of attacking big language models using hint injection and other means, and developed corresponding protective measures. This step also went well.

After the Hugging Face incident occurred, we realized that in order to intercept AI misbehavior, methods other than deterministic control were needed, because such control could not anticipate every action the intelligence would take, nor every method of deception it would try. We realized that “Using AI to fight AI” can't just be a slogan. We really need to train a big language model that can interpret, understand, and stop AI misbehavior. Following this direction, we began to think, how can we train our own models? We looked at a number of companies to see who was competent, and compared the capabilities of various companies in the market. In the end, Applied Compute stood out.

Later, we came to the conclusion that while applying the training model for enterprises, it is also necessary to train an AI model that plays a supervisory role. This is almost equivalent to having AI build AI and then build an AI to oversee the previous AI. These two things cannot be separated. I don't understand how someone else's app is supposed to work, so I can't step in and determine what it's doing.

Because I didn't train that app. Just imagine if you could train your own app and have the same team train a model to oversee the app, I think this is the future of AI security.

From this perspective, to do a good job in model supervision, we must enter the model training business. This is the direction we're betting on, and I'm glad the Applied Compute team shares this line of thought. We basically negotiated and announced this deal today. Everyone is very excited, and I think the opportunity is very great. In fact, just as I was sitting here, a CEO sent me a text message. He just met them this morning and is very excited. Their original concern was how can large enterprise applications be entrusted to a small team of only 100 people? Now that there is support from Pietuo Network, they feel they can rest assured that this is a long-term partnership. I'm really excited about it.

Hope we bet on the right direction this time. If successful, I think its potential will far exceed what our current cybersecurity business can achieve, and it may be larger than our business today.

J.P. Morgan Analyst Brian Essex: Great. I have follow-up questions; I'll leave them for later. Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Piotr Networks: Thank you, Brian. Next, ask Peggy's Rob Owens to ask questions, followed by Morgan Stanley's Mehta Marshall. Rob, are you online? OK, let's ask the next one first. Ask Morgan Stanley's Mehta Marshall, followed by Scotiabank's Patrick Colville.

Morgan Stanley Analyst Mehta Marshall: OK, thanks for giving me the opportunity to ask questions. Nickish, I'd like to ask for advice. There were some errors when the customer launched these measures, and some virtual patching or speedy patching practices apparently caused more service outages. How do you think customers are responding? Are they slowing down? Is there some kind of batch processing approach? Is the service team getting more involved? How do you think the increase in outages will change the patch deployment process?

Lee Claridge, Chief Product and Technology Officer at Pioneer Networks: Thank you, Nickish. Thanks for the question too. First, let's differentiate between the two things you just mentioned. Patching too fast can present challenges and affect the operation of the system. Virtual patching is the solution to these problems. Looking back, AI discovered a large number of bugs, triggering a new wave of patches. Next, the operation and maintenance problem faced by enterprises is how quickly can all relevant systems be patched? This puts pressure on the operations team. Some of the outages we've seen recently are due to companies looking to be happy when actually installing the patch.

Virtual patching allows enterprises to safely deploy temporary protective measures, freeing up time for operation and maintenance teams to install official patches. At the end of the day, we can provide the real-time protection needed as soon as vulnerabilities are discovered, giving businesses time to install patches. For example, our service now includes thousands of virtual patches. We expect this number to reach tens of thousands by the end of this year, and is currently growing very fast to keep up with the speed at which vulnerabilities are discovered.

Morgan Stanley Analyst Mehta Marshall: OK, thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Piotr Networks: Thank you, Mehta. Next, ask questions from Scotiabank's Patrick Colville, followed by TD Cowan's Shaul Eyal.

Patrick Colville, Scotiabank Analyst: Thank you, Hamza. Nikish, maybe ask Deepak to talk with you. I'd like to continue talking about the actual impact that increased cyber risk has had. When we spoke with some of your company's biggest customers recently, it was clear that their demand for Frontier AI Defense is extremely strong. I'd like to hear how widely you think this product can be used. Does it work for all medium and large businesses? Can this category gradually develop into a fairly large market? Or is this a pain point for the specific period of November 2026. It currently brings expenses, and it is likely to stabilize after a while? Thank you.

Nikish Arora, Chairman and CEO of Pietuo Networks: The concept of vulnerability management has been around for a long time, and it itself forms a market. In the past, it was generally a group of professionals who discovered bugs, organized the findings in a systematic manner, and sold the flow of vulnerability intelligence data to other agencies, so that these agencies could check against their own infrastructure. Over the past few months, as cutting-edge big language models have demonstrated these capabilities, companies can incorporate vulnerability detection into their routine work, and can check their own infrastructure and specific configurations without outsourcing. Unfortunately, malicious attackers have the same capabilities. Every business is now racing against time to fortify network boundaries and infrastructure to ensure that it cannot be breached. Vendors are also racing against the clock to close possible loopholes.

The model has improved rapidly in the past six months, and the latest model can now find far more bugs than six months ago. I don't think model capability has peaked yet, and I don't expect it to peak in the next six to nine months. In other words, better models will continue to appear in these six to nine months. Smarter models can perform deeper simulations to identify possible paths attackers may take. I think in the next 12 months, or roughly 12 to 18 months, companies will need to continue testing using cutting-edge AI. This is why we call our products “continuous cutting-edge AI defense,” and we don't think this kind of demand will subside in the short term.

Judging from the industry landscape, I believe that continued cutting-edge AI defense will eventually become a capability provided by platform vendors to every enterprise. As far as we are concerned, it can be delivered through the XSIAM platform or through the border security platform, because our platform must continuously inspect the customer's infrastructure, stress test it, and gradually become self-healing. We can no longer expect customers to find problems first and then discuss with us how to fix them. As you can see, manufacturers in the industry will integrate this capability into their products, but I think it will take another two years. Until then, customers will always want or need to test their infrastructure using an open weighting model, open source model, or cutting-edge big language model.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you Patrick. Next, ask TD Cowen's Saul Eyal questions, followed by Citi's Fatima Blaney. Shaul?

TD Cowen Analyst Shaul Eyal: Yes. Good afternoon, and congratulations. Over the past 12 months, firewall orders increased 28% year over year. How much of this comes from AI-related requirements and how much from equipment update cycles? What factors can support continued growth? Niche, put it another way, this is a cyclical and structural problem. I'm referring to which of the requirements are cyclical and which are structural.

Nikish Arora, Chairman and CEO of Pioneer Networks: First, you need to understand that hardware only accounts for a very small part of our business, right? Our product business data includes both software and hardware, and we cannot understand all of them as hardware. AI security requirements are driving the growth of software firewalls, so keep this in mind when judging how much hardware and software each account for in the product business. Having said that, like the entire industry, we are indeed seeing a rise in demand for hardware. I think one of the reasons is that the increase in component prices pushes up product prices and encourages customers to purchase in advance. Everyone knows that due to the demand for AI, parts prices in the entire industry are high. Demand is indeed increasing, and a number of new cloud service providers need to gradually consolidate their infrastructure.

As can be seen, we are benefiting from the infrastructure construction cycle of new cloud service providers. I think there are consolidation factors at play, and the customer wants to start demolishing and replacing some of the old facilities. All three factors should have played a role because we've never seen the hardware business grow at this rate before. We believe this momentum will continue for at least 12 months, or even longer.

TD Cowan Analyst Shaul Eyal: OK, thank you so much.

Deepak Golecha, Chief Financial Officer of Pietuo Networks: Let me add one more thing. Some customers may have said before, “I want two vendors.” What we're hearing now is: “With just one, I can handle all the patches and other things more easily.” There are a number of factors at play, but overall, we think demand is really strong, and the momentum will continue.

TD Cowan Analyst Shaul Eyal: Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you, Shaul. Next, ask Citi's Fatima Blarney to ask questions, followed by William Blair's Jonathan Ho.

Citibank Analyst Fatima Blaney: Thank you. Good afternoon everyone, and thanks for giving me the opportunity to ask questions. Nikish, I'd like to talk more about the motivations and reasons for the acquisition of Applied Compute. This strategy, at least in my opinion, is already very clear. But what I want to know is how the unique value you provide will evolve as a cybersecurity platform that remains neutral like Switzerland in the AI infrastructure platform ecosystem? Will it weaken as a result, or will it change in some other way? After all, your relationships with certain customers and partners may develop into relationships that are both competitive and mutually beneficial. There is another related question, I might ask Lee to answer. To what extent can Applied Compute's results feed back into your existing business, help you launch new product sales units, new capabilities, and new products, and further enhance your platform, safety products, and security capabilities?

Nikish Arora, Chairman and CEO of Pietro Networks: I'm wondering what you mean by “Swiss style” and “mutual aid.” You mean “competition,” right? OK, got it. Let me talk about it separately.

Citibank Analyst Fatima Blaney: Yes, that's what I meant.

Nikish Arora, Chairman and CEO of Pioneer Networks: Fatima, first, there are very few talents on the market that can reach the level of cutting-edge artificial intelligence laboratories; none of us have enough such talents. Just saying that they can bring about changes as soon as they join, this group of people will significantly enhance the artificial intelligence capabilities of our entire company. They've been working at the cutting edge of the lab and brought the way of thinking there. We've been working hard to use big language models to train our products, and I think this team will greatly speed up our progress so we can deliver these artificial intelligence capabilities to our customers. This is the first point. Even if you just do that, it's already worth it. Second, as we said, when we went looking for this type of service ourselves, we discovered that there was indeed a market behind it. We looked at nine vendors and determined that this team is the best at training models.

Even the best teams among them are still in the early stages of growth as a company and as an organization. Years of experience in corporate customer service tell us that customers are looking for reliable partners who are fully equipped in all areas, not only responsible for results, but also able to deliver quickly. As long as we combine our capabilities in sales promotion, operational resilience, and cybersecurity with them, their existing business will expand even if they don't do anything else. That in itself is another gain. What excites me the most is the third point. As we go deeper and deeper in the field of AI security, I gradually came to the conclusion that before starting to build a product that is neutral and adaptable to various companies like Switzerland, we must first ensure that we can make a product that is end-to-end integrated and works well like Apple, right? Because there is currently no such end-to-end capability in the industry.

You need the same vendor to discover your AI assets, gather relevant traffic to a unified control point, and let this vendor understand what the big language model is doing to determine what actions to release and what actions to block. Imagine someone who built an artificial intelligence system himself. I don't know anything about what's going on inside, but they claim to be able to provide him with a security product, figure out what the artificial intelligence wants to do, and stop it if necessary. I think this idea is just too far-fetched to do. As you can see, the field of artificial intelligence is forming a complete set of research methods around model evaluation and execution trajectory analysis.

Just two days ago, OpenAI announced that it had reached an agreement with four companies to give them early access to their models and see how they would perform. Once the model behaves abnormally, they can tell OpenAI how to improve training. To stop the abnormal behavior of artificial intelligence, we have to have such a deep understanding of it. If I can't even determine when the model performs normally or when it deviates from expectations, how can I stop it? This requires cutting-edge big language models — people who have mastered cutting-edge big language model technology. This makes sense, and explains why we're doing it. I think you'll eventually see the entire industry moving in this direction, and many will try to keep up with us.

At that time, we'll also develop big language models, so don't worry.

Citibank Analyst Fatima Blaney: Understood, thank you.

Nikish Arora, Chairman and CEO of Pietro Networks: Hmm--

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pietuo Networks: OK, thank you, Fatima. Next, ask William Blair's Jonathan Ho to ask questions, followed by Wells Fargo's Michael Thurin.

William Blair Analyst Jonathan Ho: Great. Good afternoon, and congratulations on your impressive results. Prisma AIRS is now the fastest product for your company to reach $170 million in annual recurring revenue. How can we determine what stage artificial intelligence applications have reached? As customers move from developer scenarios to true enterprise-level intelligence and inference scenarios, is 2027 likely to become an inflection point for further acceleration in popularity? Can you talk about why as application scenarios change, so does the growth trajectory of demand? Thank you.

Nikish Arora, Chairman and CEO of Pietro Networks: OK, but I think it's still early. Like I said, go ask any company you've worked with: “Do you use artificial intelligence programming tools?” They'll say it's in use. Then ask, “Are these tools safe?” Most of them will ask the opposite question, “What do you mean?” Everyone still hasn't figured out how to secure Claude Code, Codex, or Cursor. I think that's the starting point. You have to consider that once these programming tools have the ability to act autonomously, they will spawn new agents. How do you control them? I think most companies haven't solved this problem yet, but are already busy using these programming tools. I expect you to see that some of the intelligence experiences generated by programming tools deviate from expectations. Let's start here.

This is the first application scenario for Prisma AIRS to prevent programming agents from getting out of control within the enterprise. As for future application scenarios and other fields, I asked Li to discuss it.

Lee Claridge, Chief Product and Technology Officer at Pioneer Networks: Yes. First of all, I agree with Niche. This is still--

Nikish Arora, Chairman and CEO of Pietro Networks: Good strategy.

Lee Claridge, Chief Product and Technology Officer at Pioneer Networks: We're still in the early stages. Business adoption of artificial intelligence has only just begun. Artificial intelligence programming is the first scenario where it is actually more commonly used in production environments. The vast majority of other artificial intelligence and smart applications are still being tested, piloted, or developed. Often there are two issues that prevent these applications from being fully produced. First, can they be kept safe? Second, are the results of their missions accurate? I think both of these issues can be solved. Our work with Applied Compute will go a long way in helping us find answers to enable enterprises to adopt agents and artificial intelligence in a wider range of business scenarios.

William Blair Analyst Jonathan Ho: Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Piotr Networks: Thank you for the question, Jonathan. Next, we asked Wells Fargo's Michael Thurin to ask questions, followed by Josh Tilton of Wolf Research.

Wells Fargo Analyst Michael Turin: Thank you, Hamza. Nikish, some of the risks you warned about before seem to have come to light now. Can you talk about where the money for companies to deal with these risks comes from? How much does the integration that the platform-based strategy emphasizes help drive customers to make budget decisions? Is the company prepared if demand continues to surge? How will your company and the industry respond in the future? Thank you.

Nikish Arora, Chairman and CEO of Pioneer Networks: I think the money for the company's purchase of Frontier AI Defense's cutting-edge AI defense services is currently being taken from emergency funds. I don't think any management team would sit there and say, “We don't want to do this, we can't do it.” They said, “We have to do it.” I don't know where they've squeezed out the budget, but they did find the money to make sure they can check for flaws. Depending on what happened in the past two weeks, even people who have been blind to events before will probably say, “Damn, if we haven't done it yet, we'd better check ourselves quickly.” I'm not worried that customers will pay. What is really unresolved is the longer-term question.

A lot of CIOs will come to us and say, “Listen, how can we solve this problem in the long run?” I think that's why they started integrating security tools and vendors. What they thought was, “How can we make every dollar make a greater difference? How can we ensure that the money we spend today is in line with our long-term direction?” I believe that in the next two years, you will see an increase in cybersecurity budgets and a certain degree of integration, because in the past, you spent money everywhere and hoped that there would always be a method that worked, and it didn't work in the field of cybersecurity.

Wells Fargo Analyst Michael Turin: Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you, Michael. Next, ask questions from Josh Tilton of Wolf Research, followed by Gabriella Borges of Goldman Sachs.

Wolf Research Analyst Josh Tilton: Hello everyone, can you hear me?

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Piotr Networks: I can hear you, Josh.

WOLF STUDY ANALYST JOSH TILTON: Great. I would like to address one point that was just mentioned in the management statement. You said that this quarter, or in recent times, there was a sharp increase in smart traffic on secure access service edge platforms. When we talked to investors, we found that they didn't seem to see SASE as a market that could benefit from the development of artificial intelligence. Can you be specific, as far as SASE is concerned, what opportunities do you see intelligence bring? It's clear that changes are already taking place on the platform. Looking a little deeper, have you started charging for smart traffic on the SASE platform now? Is this traffic generated in increments based on human user traffic? How can we determine how big of an opportunity this actually means for SASE?

Nikish Arora, Chairman and CEO of Pietuo Networks: I thought you wanted to know about the industry; it turns out that what you really care about is how to make money. OK. In the past, SASE always charged per user, right? Because there is an upper limit on the amount of traffic each user can generate, after all, the amount of traffic that one person can generate in a month or year is limited, so SASE sells per user and per seat. There's a similar example in the industry. Today, many agents are moving around and initiating visits on the public network, and traffic passing through network application firewalls (WAF) has increased as a result. Imagine the same situation happening within an enterprise. Agents will also begin to move around within the enterprise, putting an additional load on the SASE network. To deal with this change, I think we've already begun to act, and other companies in the industry are probably doing the same.

In the past, when we sold SASE, we included a certain amount of free network capacity with the license. In the future, we will measure actual usage in more detail, because the more traffic flows through our SASE network, the higher our operating costs. You'll see that we're starting to earn revenue from traffic. At present, smart device traffic has soared to 9 times that of the same period last year. Even so, looking at the entire customer organization, the total usage still did not exceed the existing license limit. But I expect that soon there will be customers whose total SASE traffic significantly exceeds the capacity they are licensed to use. As the number of agents within the enterprise increases, we'll see SASE start generating some pay-per-traffic revenue. I'm guessing it will take 6 to 9 months, but this will happen.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you, Josh. Next, ask questions from Goldman Sachs's Gabriela Borges, followed by Guggenheim's John Diffucci.

Goldman Sachs Analyst Gabriela Borges: Good afternoon, thank you. Nikesh, you said a while ago that you hope the company's business can participate more deeply in the word flow. I'd like to ask about this. The Chronosphere and observability business clearly fits this line of thought, and I think the same goes for businesses like Applied Compute. Could you please talk about it from a more macroscopic perspective? As AI applications expand, do you think more business areas will intersect with security? Can you talk specifically about how security and infrastructure will be integrated?

Nikish Arora, Chairman and CEO of Pietuo Networks: Looking back at our past cybersecurity business model, we mainly sold licenses to companies and also used a per-user fee model. As you know, as smart entities join, the pay-per-user model will change. The enterprise licensing model is in some cases inherently about usage. For example, software firewalls are configured according to processing capacity. The more traffic that passes through the firewall, the more capacity is needed, and customers will have to buy more capacity. Prisma AIRS is the most typical example; its core is to check terms. The more terms an enterprise uses, the more test terms are needed, and test terms are costly. Over the long term, customers may spend 2% to 5% of their term spend on Prisma AIRS.

If you believe that the word sales value will reach 1 trillion dollars in the next few years, then according to this logic, the entire industry's security expenses around these terms should be equivalent to 3% to 5% of the word spend. I think the same logic applies to Applied Compute; this isn't meant to be a word game. As you can see, and as I mentioned earlier, Applied Compute is essentially a model training business. The more models you train, the more models you put into inference operations. Based on this, Applied Compute will also provide security capabilities in the future. Customers will have to pay not only for training large language models for everyday applications, but also for security. If you think about our various products, where traffic increases and word usage increases, we will participate in the word flow there.

Take a look at our three businesses, from cybersecurity and Cortex to identity security and observability, then take a look at the new business of Applied Compute. Observability and Applied Compute both depend on traffic. The more traffic in the AI ecosystem, the more the observability business benefits. The more traffic the AI ecosystem needs to handle, the more Applied Compute benefits. The larger the amount of data that circulates within the enterprise, the more Cortex benefits because it needs to ingest this data. As traffic increases, Prisma AIRS and cybersecurity businesses also benefit, as customers require more processing capacity whether using software firewalls, hardware firewalls, or SASE. Leaving aside the corporate licensing factor, the only thing I don't see directly related to traffic is Idira, but that's why we're developing agent-related capabilities for Idira. The more intelligence agents, the more non-human identities Idira will need to manage.

As a result, we're reorienting our direction to tie all of our business to term traffic. Because in the long run, as AI becomes more popular and used more, we believe that the business model of Pioneer Network will also echo the actual usage of AI.

Goldman Sachs Analyst Gabriela Borges: I love this perspective. Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Piotr Networks: OK, thank you, Gabriela. The last question came from Guggenheim's John Diffucci.

Guggenheim Analyst John Diffucci: Thank you, Hamza. You have done a great job, and everyone can see that this is indeed a very strong performance. However, in many ways, this acquisition is very different from everyone's current business. After listening to Deepak's preliminary remarks, this is obvious. Nikish, considering your experience before joining Pietuo Network, can we assume that you also have personal experience in the field of computing power? Also, Deepak, how should we see the financial characteristics of Pietuo Network evolve, especially if this business reaches the scale you expect?

Nikish Arora, Chairman and CEO of Pietro Networks: Jonathan, you know that my career has been based on doing things I've never done before.

John Diffucci, Analyst at Guggenheim: Okay.

Nikish Arora, Chairman and CEO of Pietuo Networks: You can count on me to win in the field of computing power due to my lack of experience in computing power. Sorry, I was kidding because he just asked about my experience with computing power. Computing power isn't that hard to understand, right? Find a large plot of land, make sure there is electricity, make sure there is water, you can air cool the chip... I mean water cooling. If you find a way to get some chips from a friend, you can start building a large data center. The good news is that many people are now building such facilities. Applied Compute currently has several computing power contracts with these major operators. Furthermore, we ourselves are major computing power customers of Amazon Web Services (AWS) and Google Cloud Platform (GCP). Until now, we have developed our own business capabilities without first becoming a computing power expert.

Maybe we don't need this experience in the future either. As the business grows, we may gain more computing power through inference business cooperation. Anyway, when it comes to the Applied Compute business, the first thing I think about is creating demand. To create demand, we need to convince companies that they need to choose Pietro Networks and Applied Compute because they need secure inference services and secure model training. We can use pay-as-you-go computing power to provide these services, or we can use the three-year computing power contract that Applied Compute has already signed. It doesn't require us to be experts in computing power. Of course, when demand comes up, if we want to obtain higher profit margins from the computing power business, we can integrate computing power resources ourselves while taking corresponding risks; that is another story. Once again, we said we wanted to master 100 to 150 megawatts of computing power.

I can do this by cooperating with new cloud service providers, and I can also cooperate with various cloud service providers in the market. I don't have to be an expert on this right now. If I go further and start talking about gigawatts of computing power, then yes, I'll have to consider how to become a computing power expert. But until then, I think we can handle it. I think the key is to drive further revenue growth, and that growth will come from the secure delivery of inference and training services to businesses.

Guggenheim Analyst John Diffucci: OK, thank you. Deepak, how will this affect the company's financial characteristics in the long run?

Deepak Golecha, Chief Financial Officer of Pietuo Networks: John, I think the most important thing is to understand that we are adopting an asset-light model, right? We do have plenty of options. In my preliminary remarks, I talked about how we expect this to affect the company's short- and medium-term financial performance, and also talked more broadly about the long-term impact. But the direction we have always determined is to provide companies with sufficient security to act with confidence. We believe this is where we can make money. This field will continue to have good profit margins, and we will add another highly profitable business.

Guggenheim Analyst John Diffucci: You have made a lot of bold moves. At least as far as my observations are concerned, these decisions have been quite accurate so far. We'll wait and see what comes next. Thank you.

Hamza Fordwala, Senior Vice President and Head of Investor Relations at Pioneer Networks: OK, thank you, John. This concludes the question and answer session of this conference call. I would now like to give the floor to Niche to conclude.

Nikish Arora, Chairman and CEO of Pietro Networks: I would like to thank all of our employees, partners, and customers for their trust in us, and for their hard work to help us get to where we are today. We are off to a good start to this fiscal year, and I hope everyone will continue to work hard and achieve great results. Thank you all, and thank you all for participating in this conference call.

Disclaimer:Webull uses external vendor Google Translation Service for news translations where we endeavour to ensure these are correct, however, we recommend that you please double-check this information accordingly. Webull is not responsible for translation errors or issues.
What's Trending