AMD, CoreWeave, and Upstart: Assessing the 10x AI Growth Thesis

The secular tailwind for AI investment is now a tidal wave, driven by a market that is projected to grow from $189 billion in 2023 to $4.8 trillion by 2033. That is a 25-fold increase in a decade, a scale that redefines entire industries. This isn't just incremental tech adoption; it's a fundamental shift in how enterprises allocate capital. The critical pivot is that AI budgets have graduated from being a line item for "innovation funds" to becoming core IT spending. As one analysis notes, AI has graduated from experiment to core operating expense.

This shift is already driving a stark market bifurcation. While overall IT budgets are rising just 1.8%, AI spending is increasing 5.7%. More telling is the math of budget reallocation: AI is capturing 30% of the total IT budget increase despite being a fraction of overall spend. Companies are stealing budget from elsewhere to fund AI, making it the only growth budget that matters. The result is a clear winner-take-most dynamic in the public markets. In 2025, AI-native or AI infrastructure companies like Palantir and CrowdStrike saw their valuations far outpace traditional SaaS firms. The divergence was brutal, with winners up over 140% and losers down more than 50%, a 4.7x difference in valuation multiple.

The bottom line is that the AI supercycle is now a structural force, not a speculative theme. The market's size and the enterprise budget shift create a powerful, self-reinforcing engine. For the bull market to sustain its momentum into 2026, this engine must continue to deliver. The risk is not a lack of investment, but a failure of that investment to translate into the promised earnings growth at the pace the market now demands.

AMD: The Challenger's Path to Market Share Capture

Advanced Micro Devices is positioning itself for a historic market share capture in the AI chip race, but the path is fraught with execution risk. The company's long-term growth plan is ambitious, forecasting a 30% compound annual growth rate for revenue, with its data center segment-the heart of its AI push-targeting an even faster 60% revenue CAGR. This aggressive scaling is built on a massive opportunity: AMD expects the total addressable market for data centers to reach $1 trillion by the end of the decade. The company's upcoming Instinct MI450 series of AI accelerators and its first rack-scale solution, Helios, slated for the second half of 2026, are seen as the catalysts that could shift the competitive narrative from distant second to serious challenger.

The core of AMD's strategy is a hardware-software ecosystem. While its new MI450 chips are described as "much more competitive" to Nvidia's offerings than previous products, the software platform is the critical differentiator. AMD's ROCm open-sourced AI software, now at version 7, is still viewed as "trailing Nvidia's functionality" in its parallel computing platform, CUDA. This software gap is a material risk; even powerful hardware can struggle to gain adoption if developers are entrenched in a more mature ecosystem. The company's success hinges on its ability to close this functionality gap and win developer mindshare.

Execution risk is amplified by the stock's volatility. Despite a 78% rally in 2025, AMD's shares have dropped 19% since peaking in late October. This pullback reflects the market's high expectations and sensitivity to any stumble in the AI buildout. The stock's recent performance underscores the challenge of being a challenger in a market dominated by a clear leader. Investors are betting that AMD's hardware roadmap and software improvements will be enough to capture a meaningful slice of the trillion-dollar data center pie, but the company must prove it can execute flawlessly on both fronts.

CoreWeave: Scaling the AI Infrastructure Stack

CoreWeave is executing a classic growth-at-any-price playbook, aggressively scaling its capacity to capture the explosive demand for AI compute. The numbers are staggering: revenue grew 204% year-over-year in the first nine months of 2025 to nearly $3.6 billion, with a backlog exceeding $55 billion. This isn't just growth; it's a massive, multi-year commitment to build out infrastructure, with the company targeting exceeding 850 megawatts of active power by year-end. The strategy is clear-scale now to lock in future demand and secure a dominant position in the AI cloud stack.

The competitive moat is built on performance and specialization. CoreWeave's purpose-built cloud for AI workloads offers a compelling value proposition, with the company claiming it can process tasks roughly 35 times faster and 80% more cost-effectively than broader cloud platforms. This performance edge, repeatedly recognized by industry analysts, is the foundation for its ability to attract and retain hyperscalers like Microsoft and OpenAI. The company has diversified beyond its crypto-mining roots, expanding its platform with strategic acquisitions and launching new services, aiming to become the "Essential Cloud for AI" for a wide range of customers.

Yet this aggressive expansion comes with a heavy price. The company is financing its massive capital expenditure with significant debt, a reality that has weighed on the stock. Guidance for 2025 revenue of $5.05 to $5.15 billion implies a torrid growth rate, but Wall Street's projections for 2026-averaging $12.01 billion, or around 134% growth-assume flawless execution. The path is fraught with execution risk, from securing power and hardware to managing the operational complexity of a global build-out. The recent shift of some 2025 revenue into 2026 due to third-party partner delays is a tangible example of the friction in this high-stakes race.

The bottom line is a company in a winner-take-most race. CoreWeave has the scale, the backlog, and the performance credentials to be a major beneficiary of the AI infrastructure boom. Its valuation, trading at a multiple of projected 2026 sales, appears reasonable given its growth trajectory. But the investment thesis is binary: it hinges on the company's ability to execute its capex-heavy expansion plan without a stumble. For a value investor, the narrow moat of a specialized platform is compelling, but the high debt load and execution risk mean this is a bet on a single, high-stakes outcome.

Upstart: Disrupting a $1 Trillion Market

Upstart Holdings presents a classic value investing opportunity: a company with a clear technological moat, returning to profitability, trading at a steep discount to its historical highs. The core thesis is straightforward. The credit underwriting market is dominated by a decades-old model, Fair Isaac's FICO score. Upstart aims to disrupt this $1 trillion industry with an AI-driven alternative. Its model analyzes more than 2,500 variables, a vast expansion from traditional tools, and can approve 101% more applicants due to superior risk separation. This isn't just incremental improvement; it's a potential paradigm shift in how lenders assess creditworthiness.

The company's recent financials show the model is translating into business results. Revenue for the first nine months of 2025 surged 57% year-over-year to $685 million, and it has returned to profitability, earning $35 million during that period. The operational efficiency is also notable, with 91% of loans fully automated by its AI. This combination of growth and profitability is the foundation for any long-term investment case.

Yet the stock's 88% decline from its post-pandemic high reflects a market pricing in macroeconomic fears. The company's fortunes are tied to the broader economy's health, as credit demand and default rates fluctuate with cycles. This creates a classic tension: the valuation offers a wide margin of safety, but the execution risk is high. The company must successfully scale its newer products-auto, home, and small-dollar loans-which now account for a growing share of originations, while navigating potential headwinds in the loan evaluation market.

The valuation itself is compelling. With a market cap of roughly $4.3 billion, Upstart trades at a price-to-sales ratio of about 5. This low multiple, juxtaposed with its 57% revenue growth and return to profit, suggests the market is discounting the full potential of its AI moat. For a value investor, this is the setup: a durable competitive advantage in a massive market, trading at a fraction of its historical peak, with the risk that macroeconomic pressures could slow its growth trajectory. The path forward hinges on execution, but the foundation for a long-term compounding machine appears to be in place.

Catalysts, Risks, and the 10x Scenario

The AI narrative promises transformative growth, but translating that potential into a 10x return by 2036 requires navigating a path of immense execution risk. For three companies at the forefront-AMD, CoreWeave, and Upstart-the near-term catalysts are clear, yet the risks to their balance sheets and competitive positions are equally material.

For AMD, the pivotal catalyst is the launch of its Instinct MI450 AI accelerators and Helios solution in late 2026. This is seen as a potential inflection point, with analysts arguing it could bring the company closer to competing with Nvidia in the data center market. The company's long-term vision is massive, with CEO Lisa Su forecasting a $1 trillion total addressable market for data centers by the end of the decade. However, the path is fraught with execution risk. While the MI450 is considered more competitive than past products, AMD's ROCm software platform still trails Nvidia's CUDA in functionality. The company's stock has rallied strongly in 2025, but any stumble in software adoption or a failure to capture meaningful market share could quickly reverse gains.

CoreWeave's catalyst is its ability to execute on a staggering capital expenditure plan to meet a revenue backlog exceeding $55 billion. The company's growth is projected to accelerate to around 134% in 2026, supported by long-term contracts with giants like OpenAI and Microsoft. Yet this growth is funded by a ballooning balance sheet. The company's heavy debt and soaring costs have already led to a sell-off in AI stocks, with CoreWeave trading at a steep discount from its highs. The risk is that its aggressive capex, expected to more than double in 2026, strains its financials if demand softens or buildout timelines slip. Success hinges on flawless execution and sustained demand.

Upstart's catalyst is the continued expansion of its AI lending platform, which it has positioned as a $1 trillion opportunity. The company has already demonstrated its model's power, with AI making 91% of loan assessments without human intervention and approving significantly more applicants than traditional tools. It has returned to profitability, with net income of $35 million in the first nine months of 2025. However, its business is exposed to macroeconomic cycles and interest rate volatility. The stock remains down 88% from its post-pandemic high, a stark reminder of its sensitivity to broader financial conditions.

The question of a 10x return by 2036 is a test of superior execution within a massive market. The total addressable market for AI infrastructure is forecast to grow at a 31% compound annual rate through 2033, a powerful tailwind. But achieving a 10x return requires not just riding that wave, but capturing a disproportionate share of it. For AMD, that means winning the software war as much as the hardware race. For CoreWeave, it means converting its massive backlog into cash flow without breaking its balance sheet. For Upstart, it means scaling its platform profitably through multiple economic cycles. The growth rates are there, but the path to a 10x return is narrow and demanding.