AI "token path" becomes the main theme of the software, internet, and AI conference

Overview

This is Jefferies' review of the 2026 software, internet, and artificial intelligence conference, presented as a replay and a 20+ page deck of key takeaways. The report covers views from public companies, private companies, venture capital firms, and industry experts, with themes focused on AI-driven usage and revenue uplift, data center and compute supply, upgraded cybersecurity demand, and the divergence in application software under AI disruption.

Core views

The core views include: first, the AI "token path" is the industry's main theme, and recent results from SNOW and DDOG also reflect this usage-scaling logic; second, the report recommends staying bullish on hyperscale cloud vendors, cybersecurity, and infrastructure software, because the expansion of AI applications will increase demand for compute, observability, security, data lakes, and databases; third, application software requires greater caution, as the industry still faces uncertainty in pricing models, restructuring pressure, and AI talent competition, though vertical software such as PCOR and platforms that lower development barriers such as TEAM still show pockets of strength; fourth, management teams at multiple companies highlighted growth, margin, and monetization inflection points in 2H or around 2027, but execution risk is also rising in parallel.

Analysis framework

The report mainly uses a conference-notes style analytical approach, integrating management interviews, expert interviews, and company breakout-session takeaways into thematic judgments. The analytical framework spans several dimensions, including macro AI demand, data center supply, product monetization, GTM changes, margin inflection points, and company execution risk, and validates the industry theme through multiple company case studies.

Methodology notes

• Thematic investment frameworkAI token path

  A revenue path driven by expanding token usage

  The report believes AI winners are typically companies that can convert model calls, inference demand, and data usage into revenue growth, such as companies related to AI compute, observability, data lakes, and databases.

• Sector allocation frameworkBullish on cloud vendors, cybersecurity, and infrastructure software; selective on application software

  Layered AI beneficiary chain

  Infrastructure and security demand benefit more directly from growth in AI usage, while application software needs to prove its pricing, workflow transformation, and AI feature monetization capabilities.

• Company analysis frameworkCross-validation of management and expert interviews

  Conference-notes style validation

  The report combines views from company CFOs, CEOs, product leaders, and external experts to observe product adoption, customer budgets, channel changes, margins, and supply constraints.

• Risk assessment frameworkExecution, supply, and monetization risks

  Risk to growth realization

  The report flags areas requiring follow-up, including CRWV's data center supply, U's Vector execution, TEAM's FY27 revenue trough, WIX's Base44 investment, NOW's M&A integration, and CHKP's GTM recovery.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• CRWV

  Beneficiary of AI compute and inference infrastructure

  Strengths

  Inference demand has become the main source of demand; revenue and cost structure are locked in at contract signing; self-built data centers improve operational and strategic control; multiple new business lines are expected to exceed $100M ARR by the end of 2026.

  Weaknesses

  Supply, powered shell, and PP&E remain key constraints, and future bottlenecks may shift to components such as DRAM.

  Comparison

  The report believes rising prices for H100, A100, and L40 reflect strong returns on customers' AI projects; the company's cost-of-capital gap versus hyperscale cloud vendors has narrowed.

  Risks

  Data center delivery, financing costs, component supply, and demand durability are the main risks.

• U

  Game engine, ad monetization, and AI creation tools

  Strengths

  Vector revenue grew 80% y/y and is approaching a $1B run rate; low ad load, mobile engagement, and about 70% mobile runtime share provide long-term support; AI tools lower the barrier to game creation.

  Weaknesses

  Vector is critical to Grow revenue recovery, and scaled execution still needs to be validated; the Unity 6 adoption curve also affects revenue release.

  Comparison

  The 2-4% fee of the low-take-rate webstore can improve player LTV relative to traditional channel fees of 15-30%.

  Risks

  If improvements in ROAS and install quality cannot be maintained at scale, growth and margin recovery may be hindered.

• TEAM

  Enterprise collaboration, developer and non-developer workflow platform

  Strengths

  Teamwork Graph and Teamwork Collection can improve AI response quality while reducing token usage; there is significant room for expansion in large enterprise customers; 65-77% of core Jira, Confluence, and JSM users are non-developers.

  Weaknesses

  FY27 may see a revenue trough due to pull-forward of large deals; channel partners are somewhat negative because of incentive adjustments.

  Comparison

  Management believes Claude Code combined with TWG improves answer quality by 44% and reduces token usage to 48% compared with standalone Claude Code.

  Risks

  FY27 growth trough, channel adjustments, returns on AI investment, and execution of non-developer expansion are the main items to track.

• GOOGL

  Vertically integrated AI stack and Antigravity developer tools

  Strengths

  TPU, Gemini, Antigravity, and consumer entry points such as Search, Gmail, Docs, and Slides create structural advantages; Antigravity supports non-GOOGL models and extends to desktop, CLI, and SDK.

  Weaknesses

  Experts define the competitive focus as Google AI suite versus Anthropic AI suite, and adoption still needs to keep improving.

  Comparison

  Experts believe AWS may have missed the early AI rhythm, Kiro is not viewed as a serious alternative, and MSFT PowerApps is not a major competitive focus in customer discussions.

  Risks

  Developer tool retention, enterprise adoption, and competition with the Anthropic ecosystem are the core risks.

• PCOR

  Construction vertical software and system-of-record platform

  Strengths

  The new management team comes from ANSS and has scaling experience; the platform carries about half of customers' total construction volume, with a strong data moat and stickiness; there is still room for expansion in international business and channels.

  Weaknesses

  The end market has not yet clearly improved, and weakness in areas such as manufacturing offsets strength in data center construction; new GTM and management changes may bring short-term volatility.

  Comparison

  Current growth is about 15 percentage points faster than the 0-1% end market, and historically it has typically been 10-20 percentage points faster than the end market.

  Risks

  Macro recovery, GTM adjustments, international expansion, and AI sales initiatives still need validation.

• WIX

  AI-first website, application, and workflow building platform

  Strengths

  Harmony combines vibe coding with drag-and-drop design, while Base44 expands into apps, dashboards, CRM, and workflow tools; its self-developed LLM reduces inference costs to about 5% of third-party models.

  Weaknesses

  Base44 is not yet profitable, and sales and marketing investment is causing short-term FCF to decline; the market worries that AI commoditization will pressure the core site-building value proposition.

  Comparison

  The report believes Harmony has a more structural advantage relative to pure AI tools and WordPress because it does not require re-prompting for every modification and has a clearer AI strategy.

  Risks

  It remains unclear whether Base44 is creating incremental TAM or cannibalizing core drag-and-drop users, and short-term cash flow and the reinvestment cycle also need monitoring.

• NOW

  Enterprise workflow, AI SKU, data layer, and security governance platform

  Strengths

  AI credit monetization can boost three-year ARR through early renewals; AI Control Tower combined with Veza's permissions graph and Armis can address AI agent governance pain points; RaptorDB Pro reached $100M ACV in its first year.

  Weaknesses

  Its win rate in entering the CRM market remains debated, as incumbents are also strengthening platform infrastructure and agentic AI capabilities.

  Comparison

  RaptorDB Pro and Analytics are expected to reach $1B+ ACV and support the FY30 revenue target of $30B+.

  Risks

  Multiple acquisitions completed in a short period increase integration execution risk; CRM competition may raise sales costs.

• CHKP

  Cybersecurity, SASE, and emerging security products

  Strengths

  Email Security, CTEM, and SASE together delivered billings growth above 45% and ARR growth above 40% y/y; subscription revenue continues to be driven by emerging products.

  Weaknesses

  The 1Q GTM adjustment was more disruptive than expected and weighed on revenue; SASE still needs 12-18 months before it can meet the needs of very large customers.

  Comparison

  Current SASE can serve about 50,000 users, but continued investment is needed to support large customers at the scale of hundreds of thousands of users.

  Risks

  Pressure on 2Q appliance sales, the pace of GTM recovery, SASE capacity expansion, and product recovery in 2H are the core risks.

• GEN

  Consumer cybersecurity, identity protection, and fintech cross-sell

  Strengths

  MoneyLion grew 44% in FY26, with potential in cross-sell and financial wellness markets; the cyber safety business has about a 60% operating margin; AI tools can also improve internal efficiency.

  Weaknesses

  Current AI concerns bring more stability than a growth step-up; agentic products have not yet become a clear source of incremental monetization.

  Comparison

  Management believes there are cross-sell opportunities between lending users and anti-fraud demand, with the potential to form $100M+ incremental ARR starting in F2H27.

  Risks

  MoneyLion integration, financial market expansion, ARPU improvement, and actual conversion of AI protection demand still need to be observed.

• SONO

  Consumer audio product cycle and execution improvement

  Strengths

  Management's confidence in 2H comes from the product pipeline, double-digit constant-currency growth in EMEA/APAC, marketing changes, reliability improvements, and future AI capabilities.

  Weaknesses

  The report input retains only limited excerpts from the main text, and details are incomplete; reliability repair after prior app issues still needs continued proof.

  Comparison

  No sufficient peer comparison was provided.

  Risks

  Product launch cadence, reliability improvements, regional growth, and AI capability rollout are the main watch points.

Key data

• Report date2026-06-02The report is Jefferies' review of the 2026 software, internet, and AI conference.
• Conference materials20+页deckThe main text states that a deck with more than 20 pages of key conference takeaways was attached after the call.
• CRWV revenue guidance$18-19B revenue run rateCRWV guided to reach an $18-19B revenue run rate by period-end, with new networking, CPU, and software business lines each expected to exceed $100M ARR by the end of 2026.
• CRWV margin inflection pointLow double-digit EBIT margin by the end of 2026Management expects EBIT margin to inflect from Q2 to Q3 as revenue scales.
• CRWV cost of capitalWACC down about 600bpsThe company said its cost of capital has improved significantly in recent years, narrowing the gap versus hyperscale cloud vendors.
• U Vector growth80% y/y, approaching a $1B run rateVector remains a key driver of Unity's growth and margin recovery.
• U margin25-27%Management believes margins can continue expanding from the current range, although gross margin and R&D investment still show some near-term volatility.
• TEAM AI efficiency caseAnswer quality up 44%, token usage at 48%Management said Claude Code combined with Teamwork Graph can deliver higher-quality results with lower token consumption than standalone Claude Code.
• TEAM enterprise penetration85% of the Fortune 500 are customers but contribute only 10% of revenueThe company believes there is still ample room for expansion among large enterprises.
• GOOGL Antigravity adoption rateCurrently 5-10%, about 25-30% by year-endA SADA expert said Antigravity is gaining traction within the customer base and has advantages in supporting non-GOOGL models, desktop, CLI, and SDK.
• PCOR relative market growthAbout 15 percentage points faster than the end marketPCOR is still significantly outperforming in an end-market environment of 0-1% today, and historically it has typically grown 10-20 percentage points faster than the end market.
• WIX Base44 shareAbout 40% share in the United StatesBase44 usage is growing rapidly, expanding into apps, dashboards, CRM, and workflow tools.
• WIX cash flow changeFCF down from about $650M to about $400MManagement is increasing sales and marketing investment in Base44, putting short-term pressure on free cash flow.
• NOW RaptorDB$100M ACV in year one, targeting $1B+ ACVRaptorDB Pro is viewed as an important part of NOW's platform infrastructure layer and supports its FY30 revenue target of $30B+.
• NOW security and risk businessRevenue above $1BThe acquisitions of Armis and Veza add capabilities to AI Control Tower and the permissions graph.
• CHKP emerging productsBillings growth >45%, ARR growth >40% y/yEmail Security, CTEM, and SASE are the main drivers of subscription growth.
• GEN MoneyLionFY26 growth 44%Management is optimistic about MoneyLion cross-sell and financial wellness use cases, expecting them to bring $100M+ incremental ARR starting in F2H27.
• GEN cybersecurity marginAbout 60% operating marginManagement emphasized maintaining balance between growth and profitability.

Impact & implications

In terms of investment implications, the AI theme is no longer just competition at the model layer, but has spread to compute, data centers, observability, security, databases, developer tools, and enterprise workflow governance. The assets benefiting more directly are companies with control points in compute supply, the data layer, cybersecurity, and infrastructure; application software companies need to prove that AI features can generate incremental ARR, higher ARPU, or efficiency gains, rather than being eroded by low-cost generative tools.

Risks

• Application software companies still face unclear AI pricing models, business transformation pressure, and a disadvantage in AI talent competition.
• Growth in AI infrastructure is constrained by supply limits in data centers, powered shell, PP&E, and future components such as DRAM.
• Agent-driven attacks and zero-day vulnerabilities may raise security budgets, but they also increase the complexity of enterprise IT environments.
• CRWV needs to keep proving inference demand, capacity delivery, and stable margin contribution.
• If U's Vector cannot maintain improvements in ROAS and install quality after scaling, growth recovery may slow.
• TEAM is under pressure on short-term revenue and ecosystem sentiment due to FY27 pull-forward of large deals and channel incentive adjustments.
• WIX faces FCF decline from Base44 investment, risk of core product cannibalization, and commoditization risk in AI site building.
• NOW faces integration risk from multiple acquisitions completed in a short time, and CRM expansion may also encounter stronger competition and higher sales costs.
• CHKP's GTM adjustment previously caused disruption, and recovery in 2H26 products still requires validation through pipeline conversion.
• GEN's AI and MoneyLion synergies are still at an early stage and have not yet formed a clear growth inflection point.

What to watch

• Usage, revenue growth, and AI-related customer expansion at "token path" representatives such as SNOW and DDOG.
• Whether data center construction backlog remains strong through 2027 as experts suggest, and whether supply constraints shift from power and shell to components such as DRAM.
• Whether GOOGL Antigravity adoption can rise from the current 5-10% to about 25-30% by year-end.
• The ramp-up cadence of CRWV's self-built data centers in 2026 and 2027, and whether its low double-digit EBIT margin target is achieved.
• NOW AI credit monetization, AI Control Tower adoption, RaptorDB Pro ACV growth, and progress on M&A integration.
• Whether CHKP's 2H26 product recovery, subscription growth, and SASE expansion can offset 1H GTM disruption.
• Whether WIX Base44 brings incremental TAM rather than mainly cannibalizing core drag-and-drop site-building users.
• Whether TEAM's non-developer users, large enterprise customers, and Teamwork Graph-related AI usage can continue to expand.
• The new management team's GTM adjustments at PCOR, international market expansion, and path to the Rule of 40.
• AI product and enterprise demand signals related to KVYO, UPWK, CRM, HUBS, and AVPT in the day-two agenda, which require further confirmation against the full text.