The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Swift Ventures launched an AI Index for publicly traded companies on December 9, 2024, to identify businesses making measurable AI investments rather than simply mentioning the technology. Its framework combines a fine-tuned large language model with evidence from filings, earnings calls, hiring, workforce composition, research, open-source work and AI-linked revenue. The result is best treated as a research filter—not an audited benchmark, guaranteed source of excess returns or confirmed ETF.
What Swift Ventures launched
Swift described the product as an index covering about 90 public companies at launch. It was designed to rank or identify companies showing substantive AI activity across three broad dimensions: technical talent, research and open-source contribution, and revenue materially connected to AI operations. VentureBeat’s December 9, 2024 coverage reported that Swift was considering an ETF for early 2025, but the available sources do not confirm that such a fund launched.
The current Swift website has evolved into a company-level research interface. Pages for Nvidia, Broadcom, Meta, Alphabet, Accenture, Teradyne, CoreWeave and other companies provide AI-related business descriptions, market information, executive commentary and lists of similar companies. Current pages should not automatically be assumed to use exactly the same rules as the original 2024 launch.
Why a separate AI-investment measure is needed
Corporate use of the word “AI” has expanded much faster than the disclosure of comparable investment data. Swift told VentureBeat that its analysis counted more than 16,000 AI mentions in earnings calls during a recent quarter. A mention can describe a product, a future plan, a customer experiment or a marketing message; it does not by itself show hiring, research spending, product adoption or revenue.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
The index therefore tries to distinguish several different states that are often collapsed into one label:
- Talking about AI in an earnings call.
- Hiring people for machine-learning, data, robotics or other AI-specific work.
- Publishing research, releasing models or contributing software.
- Launching products that depend materially on AI.
- Generating revenue that management can reasonably connect to AI operations.
How the scoring system is intended to work
Swift cofounder Brett Wilson described a fine-tuned large language model that analyzes filings and other external data. His description includes earnings-call transcripts, regulatory filings, hiring and team-composition data, research publications and open-source contributions. The description appears in his LinkedIn post about the benchmark.
An LLM can classify a large volume of text and connect evidence across sources, but automation does not remove judgment. Results depend on the training labels, taxonomy, data freshness, entity matching, treatment of contradictory disclosures and any human review. Companies can also change terminology once they know what a scoring system rewards.
AI talent density
The framework reportedly examines the share of a company’s workforce in AI-specific roles. Swift said only about 200 public companies had more than 1% of their workforce in such roles. That is a Swift-derived statistic, not a universal definition of an AI-heavy company.
The signal raises important measurement questions: whether data scientists, machine-learning engineers, chip designers, robotics specialists and AI product managers are all counted; whether contractors are included; whether the denominator is global employees or job postings; and whether a percentage measure favors small companies. A large company may employ thousands of AI specialists while recording a lower percentage than a small firm with a handful of hires.
Research and open-source contribution
Publishing papers, releasing models, maintaining developer tools and contributing code can demonstrate technical capability and ecosystem influence. Those activities are especially relevant to model developers, infrastructure companies and firms with advanced engineering teams.
They are not universal requirements for commercial success. A company may keep valuable work proprietary for competitive, security or regulatory reasons. Internal use of open-source software is also different from releasing code or models to the public.
AI-linked revenue
The third signal asks whether AI is connected to operating revenue rather than merely to a project or announcement. The answer can vary dramatically by business model:
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall| Business type | What “AI revenue” might mean | Key question |
|---|---|---|
| Semiconductor supplier | Sales of accelerators, networking or memory used in AI systems | How much demand is incremental, and how concentrated are the customers? |
| Cloud provider | GPU rentals, model services or AI workloads | Do revenue gains cover power, equipment and capital-expenditure costs? |
| Software company | AI-native products or paid features added to existing software | Are customers paying more, or is AI mainly a cost of serving them? |
| Consultancy | AI transformation projects, bookings or implementation work | Have bookings converted into recognized, profitable revenue? |
| Incumbent operator | Products improved by AI or savings from automation | Is the effect separately disclosed and durable? |
Swift’s current pages illustrate this variety. Its Broadcom page discusses AI semiconductor revenue alongside infrastructure software, while the CoreWeave page describes AI infrastructure as central to revenue and backlog. Those descriptions are useful leads, but company-reported “AI revenue” is not a standardized accounting category.
Which companies can the index surface?
Launch coverage highlighted less-obvious examples including Doximity, associated with AI-powered medical-writing applications, and Leidos, associated with defense-oriented autonomous systems. Swift reportedly said these companies were growing more than 50% annually, but the retrieved material does not specify enough detail about the exact metric and period to treat that figure as a general investment fact.
The current site spans several distinct categories:
- Infrastructure and components: Nvidia, Broadcom and CoreWeave.
- Platforms and consumer technology: Meta and Alphabet.
- Services and implementation: Accenture and EPAM.
- Industrial and automation exposure: Teradyne.
- Industry-specific or data applications: TransUnion and PDF Solutions.
These are AI companies, AI suppliers and AI beneficiaries in different senses. A high score does not mean that each has the same economics, competitive position or exposure to end-user demand.
What the reported performance means—and does not mean
VentureBeat reported Swift’s claim that the index produced 37% annualized growth over the preceding three years, compared with approximately 12% for the Nasdaq and 19% for the S&P 500. The available account does not establish whether this was live performance or a backtest, the exact start and end dates, dividend treatment, rebalancing rules, transaction costs, taxes, slippage, weighting or the treatment of delisted companies.
| Reported figure | How to interpret it |
|---|---|
| Swift AI Index: 37% annualized growth over three years | Swift’s reported index or backtest result; independent reproduction is not established. |
| Nasdaq: approximately 12% | Benchmark comparison reported in the same coverage; index definition and dates require verification. |
| S&P 500: approximately 19% | Benchmark comparison reported in the same coverage; risk and return conventions are not fully specified. |
A credible comparison would disclose inclusion and exclusion rules, rebalance timing, concentration, survivorship and look-ahead controls, and whether a small number of semiconductor or mega-cap winners drove most of the result. Without that information, “outperformed” is a reported historical result, not evidence that the strategy will outperform after it becomes known or at today’s valuations.
Research contribution and profitability
Swift reportedly told VentureBeat that companies regularly contributing to AI research and open-source models had average gross profit of about 55%, versus 25% for comparable technology companies that did not. Gross profit is not net income, free cash flow or shareholder return, and the comparison group, sector controls and time period are not disclosed in the retrieved material.
The relationship may reflect selection rather than causation. Larger, better-funded or technically stronger companies may both contribute more research and earn higher margins. Hardware, cloud, software, consulting and biotechnology businesses also have structurally different economics.
Recommended Free Tools
Where the methodology can fail
Uneven disclosure
Companies report AI investment in different levels of detail. A transparent company can score better than a capable company whose work is proprietary or buried in broad operating categories.
Sector and size effects
The signals naturally favor semiconductors, cloud providers, software platforms and research-heavy technology firms. Talent percentages can favor small employers, while revenue attribution can favor businesses whose products are easy to label as AI.
Hiring-data noise
Job postings can be recycled, aspirational or generated by recruiting systems. They show intent more reliably than completed hiring, deployed systems or profitable products.
Classification and gaming risk
A fine-tuned model may misread ambiguous language, duplicate entities or fail to distinguish a customer’s AI use from the company’s own capability. Strategic use of AI terminology could inflate a score unless the system tests claims against filings, workforce evidence and operating results.
Best Value
Backtest and valuation risk
A historical basket can benefit from survivorship, hindsight, concentration and favorable rebalancing assumptions. Even a company with genuine AI execution can be an unattractive stock if its valuation already discounts years of success. The index’s reported signals do not, by themselves, measure price paid, downside risk or drawdown.
How investors should use the index
Use Swift as a candidate-generation layer, then perform independent due diligence:
- Read the latest annual and quarterly filings and investor-relations materials.
- Identify whether AI revenue, costs, bookings or margins are separately disclosed.
- Check whether hiring, research and product claims correspond with deployed products and customer evidence.
- Assess gross margin, free cash flow, capital expenditure, dilution and balance-sheet risk.
- Separate an AI supplier, infrastructure beneficiary, software vendor and AI-native developer instead of treating them as one trade.
- Compare valuation with realistic growth, competitive threats and dependence on third-party models, chips or cloud capacity.
- Review whether the apparent AI advantage creates a durable moat or merely reflects a capital-spending cycle.
The Swift pages can provide a starting point for this work, including Nvidia, Meta, Alphabet, Accenture, Teradyne, Alibaba, EPAM, TransUnion and PDF Solutions.
What remains unknown
- The exact weights assigned to talent, research, open source and revenue.
- Formal inclusion, exclusion and rebalancing rules.
- Whether the reported return is live, simulated or reconstructed.
- Independent audit or reproducibility of the scoring and performance history.
- Whether the proposed ETF ever launched.
- How valuation, concentration and risk enter the ranking.
Those gaps matter because a useful measurement philosophy is not automatically a usable investment product. Swift’s strongest contribution is the insistence that AI exposure should be tested against actions and operating evidence. Its claims about returns and profitability still require the transparency investors would expect from a benchmark or investable strategy.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
