Tech Budget Scrutiny Surges as Token Marketing Fades
In a pointed critique of how big tech has spent on artificial intelligence, Cognition CEO Scott Wu argues that token-based leaderboards and token budgets have drifted from practical value. He says the industry should shift focus from how many tokens teams burn to how much real output those efforts produce. The remarks arrive as U.S. and global markets absorb a cooling of feverish AI bets and as firms reassess where to allocate shrinking or stagnating budgets.
The conversation resurfaced during Wu’s appearance on a recent episode of the Founders podcast, where he emphasized that ROI must be defined in clear terms: revenue growth, efficiency gains, and cost savings tied to AI deployment. “We need to anchor spending to measurable outcomes,” Wu said, using a phrasing that echoed a broader industry push toward accountability in AI investments. cognition says tech companies are rethinking how they evaluate talent and progress as they recalibrate token usage against practical results.
Cognition’s Devin: A Case Study in AI Efficiency
Cognition has built a reputation around Devin, a coding assistant powered by artificial intelligence. The company frames Devin as a scalable way to increase engineering capacity, and it has already found traction with major financial institutions and automakers. Goldman Sachs has integrated Devin as part of its software engineering toolkit, while Mercedes-Benz and Rivian have turned to the platform for research and development tasks. Wu positions Devin as a benchmark for what real AI-driven efficiency can look like when tied to concrete outputs rather than token counts.
Officials at Cognition say the company’s business model centers on boosting productive capacity in engineering and product teams. The focus, they argue, is the bottom line—how much additional work gets completed and how much faster teams can move from concept to market. Wu notes that a realistic measure of success for AI tools should track engineering throughput, cycle times, and the ability to deploy features that generate revenue or reduce costs.
Industry Trends: From Leaderboards to ROI
Wu’s remarks come as industry insiders recount how some firms experimented with internal incentives tied to token usage. Reports have indicated that large players like Meta and Amazon explored leaderboards to spur teams to find new AI use cases. The aim, executives said, was to accelerate adoption and surface practical applications. However, several accounts describe an overemphasis on token tallies that didn’t always translate into meaningful business outcomes.
Around the same time, leadership at one e-commerce giant warned staff not to adopt AI for its own sake. A senior vice president reportedly reminded teams that clever bot tricks should serve real needs rather than chase points on a leaderboard. The shift away from token-centric incentives reflects a broader market impulse: hedge against runaway spending while steering AI work toward tangible return on investment.
What the Numbers Tell Investors
As markets gauge the trajectory of AI investments in 2026, investors are watching for signals about ROI in tech spending. The debate over token budgets vs. demonstrated output has real implications for earnings projections and investment theses around AI-centric companies. If companies successfully demonstrate productivity gains and clear revenue uplifts from AI initiatives, the case for sustained or expanded AI budgets could strengthen. If not, the market could reward more disciplined spending and ROI-focused roadmaps.
Key data points under discussion include:
- How AI tools shift engineering capacity and delivery speed in product cycles.
- The correlation between AI-enabled automation and cost reductions across manufacturing, logistics, and back-office operations.
- Evidence that AI-driven innovations translate into measurable revenue growth or margin expansion.
- The risk of misaligned incentives when token-based metrics overshadow real, market-facing outcomes.
Industry Voices and the Market Pulse
Analysts say the debate is timely as tech earnings season approaches and as CIOs weigh how to allocate capital in a slower-growth environment. Cognition’s stance aligns with a broader push to tie technology investments to outcomes that investors can quantify. In July 2026, market participants are wary of speculative AI bets and favor disclosures that clearly tie spend to productivity signals and competitive differentiation.
Wu notes that the best AI stories are those that demonstrate repeatable, scalable value. He points to real-world deployments where AI accelerates product development, reduces manual debugging, or shortens cycle times, thereby translating into improved margins or faster revenue recognition. cognition says tech companies should not reward activity for its own sake; instead, they should reward progress that moves the needle for customers and shareholders alike.
Implications for Workers and Compensation
Beyond the C-suite, the token debate touches workers and compensation structures. If firms begin to measure output more directly, performance reviews may pivot toward concrete deliverables rather than token tallies. That could influence project-based bonuses, equity compensation, and career advancement tracks, particularly for engineers and data scientists who are at the center of AI-driven efficiency gains.
Some employees might welcome a more outcome-driven system, especially when AI tools demonstrably cut bottlenecks and enable more creative and valuable work. Others could resist if the new metrics feel opaque or if AI bathwater—token-based incentives, quick wins, or gamified scores—no longer rewards day-to-day efforts. The risk for executives is to balance transparency with motivation, ensuring that metrics reflect both individual contribution and team collaboration.
What Bosses Should Do Now
For leaders weighing AI budgets in a volatile market, several actions emerge from the current discourse:
- Define a clear ROI framework for AI initiatives, linking spending to revenue growth, efficiency gains, and cost savings.
- Shift incentives from token expenditure toward measurable output, quality of work, and customer impact.
- Assess the pipeline of AI-driven features that can be deployed quickly and monetized or used to reduce churn.
- Communicate expectations transparently to teams about how progress will be evaluated and rewarded.
- Monitor compliance and governance around AI usage, ensuring that tools are used for meaningful business purposes rather than vanity metrics.
Conclusion: A Timely Recalibration
The case for cognition says tech companies adopting a more disciplined approach to AI spending is gaining momentum. As the market calibrates what constitutes real value in AI initiatives, the emphasis is shifting from token budgets to tangible output and ROI. Wu’s call to measure actual productivity—the true engine of growth—could reshape how tech firms recruit, retain, and compensate talent in an era where AI is ubiquitous but its business impact remains the ultimate test.
In the coming months, investors and workers alike will watch for indicators that AI investments are translating into durable competitive advantages. If leaders can demonstrate clear, repeatable value from AI—and tie that value to earnings and shareholder returns—the industry could emerge from the token era with a clearer path to sustainable growth. cognition says tech companies are at a crossroads, and the decision to emphasize output over tokens may define the next phase of AI in corporate America.
Discussion