HSBC Raises AI TAM Forecast: B2B Surges, B2C Slows

Overview

HSBC released its latest AI industry outlook, raising its cumulative global AI Total Addressable Market (TAM) revenue forecast for 2026-2030 by 38%. The report's core view highlights significant structural divergence: the B2B market has been substantially revised upward due to the rapid deployment of Agentic AI and coding tools, with Anthropic showing strong enterprise performance; conversely, the B2C market has been revised downward due to peaking user growth, ARPU dilution from low-cost plans, and monetization difficulties exceeding expectations. Despite facing challenges on the consumer side, OpenAI's estimated funding gap through 2030 has halved, supported by B2B momentum, a cap on Microsoft revenue sharing, and a new financing round. The report suggests the Western AI market is moving toward an oligopolistic structure driven by high compute costs.

Core views

The B2B market has become a new engine for AI growth, with HSBC raising its 2026-30 cumulative B2B AI industry revenue forecast by 74% to USD 1.98tn. This adjustment is primarily based on Agentic AI creating new use cases for enterprises and the explosive growth of coding tools. Anthropic has emerged as the clear leader in the B2B space, with its Annual Recurring Revenue (ARR) soaring from USD 900mn to USD 4.7bn in five months, far exceeding OpenAI's combined B2B+B2C ARR of approximately USD 3bn over the same period. The coding tools segment is particularly hot; Microsoft GitHub CoPilot usage is doubling monthly, while Anthropic Claude Code and OpenAI Codex have reached ARR levels of USD 780mn and USD 280mn, respectively. Enterprise CoPilot tool revenue reached an ARR of USD 3.2bn in May 2026, indicating accelerating enterprise AI adoption. The B2C market outlook has weakened, with the 2026-30 revenue forecast cut by 20%. Industry leader OpenAI faces a user growth bottleneck, failing to meet its 2025 target of 1bn users, and weekly active users plateaued in Q1 2026 as competitors Gemini and Claude gain share. On monetization, although OpenAI's launch of the USD 8/month ad-supported 'ChatGPT Go' plan improved paid conversion rates, it triggered downgrades from many USD 20/month Plus users, resulting in a nearly 50% downward revision to consumer ARPU forecasts. Furthermore, while the advertising business had a good start, long-term monetization efficiency remains lower than traditional social platforms due to shorter user session times and the ad-free preference of premium subscribers. OpenAI's financial pressure has eased somewhat, with the estimated 2030 funding gap decreasing from USD 154bn to USD 77bn. The improvement stems from three main factors: first, a revenue-sharing cap agreement with Microsoft (capped at USD 38bn), reducing future profit outflows; second, the completion of a new USD 122bn financing round; and third, increased value of its AMD holdings due to stock price appreciation. Although the cumulative free cash flow deficit forecast remains at USD 221bn, the narrowing funding gap buys time for an IPO or subsequent financing. The report notes that scaling back compute investment is not a viable option, as 'Scaling Laws' continue to dominate competition in model performance and market share. The global AI competitive landscape shows East-West divergence. Due to extremely high compute and infrastructure costs, the Western market has formed a 'sunk cost economy' and is building an LLM oligopoly structure where only a few giants can afford trillion-dollar investments and are likely to achieve long-term returns on capital above their weighted average cost of capital. In contrast, the Chinese market is more fragmented; independent labs have achieved rapid iteration and application deployment with lower budgets through model distillation and infrastructure optimization, though they remain constrained in accessing top-tier compute.

Analysis framework

The report employs a bottom-up sub-sector breakdown approach to calculate AI TAM, rather than a single macro penetration model. The institution precisely splits the AI market into B2B (Departmental AI, Horizontal AI, Vertical AI, Foundation Models) and B2C (Subscriptions, Advertising), tracking high-frequency indicators independently for each sub-sector. For example, in B2B coding tools, growth is validated by tracking GitHub CoPilot user counts, Claude Code ARR, and token consumption; in B2C, user activity and retention are cross-verified using Sensor Tower mobile data, Semrush web traffic, and third-party media reports. For the unlisted core target OpenAI, the report constructed a detailed Free Cash Flow bridge model to estimate the funding gap. Starting from revenue forecasts, the model deducts compute costs (based on GPU lease agreements with cloud providers and GW-level power capacity conversions), personnel costs, and Microsoft revenue sharing to derive the FCF deficit. It then adds existing financing and changes in the value of non-cash assets (such as AMD equity) to quantify external financing needs for the coming years. This method translates qualitative competitive dynamics into quantitative financial constraint analysis.

Methodology notes

• Industry Analysis FrameworkSupply-demand framework

  Sunk Cost Economics and Oligopoly Formation

  The report notes that the Western AI market has incurred massive sunk costs due to extremely high investment in compute infrastructure, creating natural high barriers to entry. Only a few players with trillion-dollar capital strength can sustain participation; this hard supply-side constraint dictates that the market will inevitably move toward oligopoly rather than perfect competition.

• Company Fundamentals & Financial FrameworkFree cash flow analysis

  Funding Gap Estimation for Unlisted Tech Companies

  For AI giants that are unprofitable and unlisted, the institution does not rely on traditional P/E valuation but calculates the future funding gap that must be filled via 'Cumulative FCF Deficit + Working Capital Needs - Existing Cash & Equivalents'. This is a core method for assessing survival pressure and IPO urgency for primary market unicorns.

• Industry Analysis FrameworkVolume-price decomposition

  Revenue Driver Breakdown: Users × ARPU

  When forecasting B2C revenue, the report strictly decomposes revenue into 'Active Users × Paid Conversion Rate × ARPU'. Upon finding that low-price plans increased conversion rates but significantly dragged down ARPU, it could accurately determine that the net impact of the 'volume-for-price' strategy on total revenue was negative, avoiding misjudgments based solely on user growth.

• Competition & Strategy FrameworkMoat / competitive advantage

  Scaling Laws as a Dynamic Moat

  The report emphasizes the flywheel effect of 'More Compute → Smarter Models → More Users → More Revenue'. In the AI industry, technological leadership is not a static advantage but a dynamic moat dependent on continuously expanding compute investment. This means stopping cash burn equals forfeiting competitive qualification, explaining why giants dare not cut capex despite massive losses.

Asset mapping & comparison

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

• Advanced Micro Devices (AMD.US)

  Beneficiary: OpenAI holds AMD shares and leases 6GW of AMD GPU compute; AMD stock appreciation directly narrows OAI's funding gap

  Strengths

  Secured large-scale compute orders and equity binding with OpenAI; breaking NVIDIA's monopoly in the AI data center market

  Comparison

  Compared to NVIDIA, AMD locks in long-term demand from top clients through an innovative 'compute-for-equity' model

  Risks

  Potential short-term stock price impact if OpenAI IPOs or reduces AMD holdings

• Microsoft (MSFT.US)

  Dual Role: Both OpenAI's largest compute supplier/revenue share recipient and ChatGPT sales channel on Azure

  Strengths

  Revenue share cap agreement locks in max USD 38bn benefit; M365 CoPilot penetration at low single digits leaves huge upside

  Weaknesses

  No longer shares revenue from Azure ChatGPT sales; must bear impact of OpenAI financial deterioration on its RPO

  Comparison

  Compared to Amazon and Google, Microsoft has deepest ties with OpenAI but also faces highest related-party transaction risk

  Risks

  Persistent financial pressure on OpenAI could affect Microsoft Cloud RPO and investment returns

• OpenAI (Unlisted)

  Core Analysis Target: B2C under pressure but B2B gaining momentum; narrowing funding gap creates conditions for IPO

  Strengths

  #1 brand awareness in B2C; rapid growth in B2B tools like Codex; Microsoft revenue share cap improves cash flow

  Weaknesses

  Stalling user growth; low-price plans diluting ARPU; losing B2B foundation model share to Anthropic

  Comparison

  Compared to Anthropic, OpenAI has first-mover advantage in B2C but temporarily lags in B2B enterprise integration and coding capabilities

  Risks

  IPO progress below expectations; persistent B2C monetization misses; obstacles to compute expansion

Key data

• 2026-30 Cumulative AI Industry Revenue ForecastUSD 2.55tnRaised by 38% from previous forecast; B2B contributes USD 1.98tn, B2C contributes USD 565bn
• Anthropic B2B ARR (May 2026)USD 4.7bnGrew 5x in 5 months, significantly surpassing OpenAI's level in the same period
• OpenAI 2030 Funding GapUSD 77bnHalved from previous forecast of USD 154bn, benefiting from improved FCF, new financing, and appreciation of AMD holdings
• OpenAI Consumer ARPU AdjustmentCut by ~50%Due to USD 8 Go plan causing mass downgrades from USD 20 Plus users; 2026-30 consumer revenue forecast cut by 33%
• OpenAI 2030 Weekly Active User Forecast1.96bnRevised down from previous forecast of 2.65bn, reflecting slowing growth amid intensified competition
• 2026-30 Cumulative OpenAI Compute CostsUSD 687bnUnchanged; corresponds to approx. 11GW compute capacity by 2030; Microsoft/Oracle/Amazon as primary suppliers

Impact & implications

For the AI supply chain, the B2B explosion implies a higher-certainty growth window for enterprise application developers, cloud service providers, and the coding tools ecosystem, especially partners deeply integrated with Anthropic or OpenAI. For the semiconductor industry, OpenAI maintaining its USD 687bn compute spend plan and the surge in AMD share value confirms rigid demand for high-end GPUs/AI chips; even if downstream application monetization fluctuates, the upstream compute arms race will not cool down in the short term. For investors, caution is needed regarding B2C AI monetization traps; sheer user scale no longer equates to commercial success, and balancing ARPU with user retention is more critical than acquisition. Meanwhile, the narrowing of OpenAI's funding gap may accelerate its IPO process, making related primary-secondary market linkage opportunities worth watching.

Risks

• B2C AI monetization difficulty continues to exceed expectations, with further decline in user willingness to pay
• Tightening AI regulation may limit oligopolists' pricing power and data usage
• Compute infrastructure construction falling behind schedule or power supply constraints
• Delay in OpenAI IPO process or significant valuation correction
• Non-Western models breaking containment via technologies like distillation, weakening Western oligopoly pricing power

What to watch

• Specific timing of OpenAI filing draft IPO papers with SEC
• Progress of B2B deployment and ARR growth in JVs between Anthropic and Blackstone et al.
• Actual paid conversion rate of ChatGPT Go plan and downgrade ratio of Plus users
• ARPU changes after Microsoft GitHub CoPilot migrates to consumption-based billing
• Effectiveness of new OpenAI agreements with Amazon and Google Cloud in offsetting Azure revenue loss