Google targets tabular AI with new foundation model TabFM
The open-source model performs predictions on unseen tabular data in a single forward pass, eliminating weeks of pipeline engineering for trading, DeFi, and compliance datasets.
+15.1% (FY ended 2025-12-31 vs FY ended 2024-12-31) 10-K
Annual net income
132,170,000,000 USD (FY ended 2025-12-31, attributable to the company) 10-K
Annual operating margin
32.0% (FY ended 2025-12-31)
Forward P/E
Unavailable: no dated forecast source
Gloria daily close for session Oct 28, 2026. Market cap and trailing P/E use this close with the SEC share total and trailing diluted earnings, each dated next to its figure. Annual revenue, revenue growth, net income and margin show their own fiscal period and link the 10-K. Current context may postdate this article.
Alphabet Inc. and its peers
Company
Price
Market cap
P/E (trailing)
Forward P/E
Annual revenue (USD)
Annual revenue growth
Annual net income (USD)
Operating margin
Alphabet Inc.
GOOGL · Price dated Oct 28, 2026 · tokenized GOOGLx
3,333,751,000FY ended 2025-12-31, attributable to the company · 10-K
75.8%FY ended 2025-12-31
Missing figures stay unavailable. Forward P/E requires a dated source. Annual revenue, revenue growth, net income and margin come from each company's own 10-K, in US dollars, and each row shows that fiscal year. Revenue growth compares the fiscal year with the prior year stated in the same 10-K. Net income is the full year amount attributable to the company, or available to common stockholders where the filing states only that, or consolidated net income. Each row names its basis. Trailing P/E reads "Not meaningful" when the dated trailing diluted EPS is a loss or zero. An operating margin appears only when the filing states both the operating income and the revenue for the same year.
Sources: SEC filings, Gloria daily closes and CoinGecko. Oldest record refresh: Oct 28, 2026, 4:51 PM EDT. Refresh is not a price or filing date. Each row shows its price date and its annual fiscal period. Peers use industry code and company size, or editorial selection. Market cap multiplies that row price by the company share total the latest SEC filing states for every share class together. Trailing P/E divides it by trailing diluted earnings from SEC filings. Rolled earnings per share are approximate because share counts can change between periods. Both appear only when the SEC data carries its date, so they are estimates at one price date, not synchronized filing valuations. Not investment advice.
Token vs underlying stock
Current comparison for GOOGL. Historical observations may postdate this article.
Split-adjusted regular-session stock closes from the existing Gloria shared history, paired only with an observed token price within 45 minutes of 16:00 New York time. Up to 60 matched sessions. No interpolation or missing-date fill. Weekends and known early closes are omitted. Token units are not adjusted for issuer multipliers or dividend treatment. This compares quoted prices, not fair value, executable premiums or arbitrage returns. Solid line: token. Dashed line: stock. Exact quote times and values appear below. Historical observations. Retrieval refers to the token source. Token quote times use UTC. The fixed cohort shows its oldest member fetch.
Alphabet token vs Alphabet stock: exact source values
Quoted token and share units can have different rights, multipliers and dividend treatment. This is not an executable premium or arbitrage signal.
Foundation models have transformed text, images and time-series forecasting, but tabular data has remained dominated by traditional machine learning algorithms that must be trained separately for every new dataset.
Google is seeking to change that with TabFM, a foundation model that brings zero-shot, in-context learning to structured enterprise data, allowing users to generate predictions without manual model development.
The company said the model is intended for enterprise use cases ranging from fraud detection and customer analytics to financial forecasting and other structured data applications that have historically relied on algorithms such as XGBoost and random forests.
Instead of fitting a model to each new dataset, TabFM treats an entire table as context for in-context learning. Historical examples and unlabeled target rows are processed together, enabling the pretrained model to learn task-specific relationships during inference rather than through additional training. This allows predictions to be generated immediately in a single forward pass on previously unseen datasets.
The model uses a hybrid architecture that combines alternating attention across rows and columns with compressed row embeddings before applying transformer-based inference.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Google said this design captures complex feature interactions while reducing computational costs compared with applying attention directly across the entire table.
To overcome the shortage of large public tabular datasets, TabFM was pretrained exclusively on hundreds of millions of synthetic datasets generated using structural causal models.
Google said benchmark results on the TabArena evaluation suite showed TabFM consistently outperforming leading supervised machine learning models, while an ensemble version incorporating additional engineered features achieved even stronger performance.
The company plans to integrate TabFM directly into BigQuery, enabling users to run regression and classification models using the AI.PREDICT SQL function without specialized machine learning knowledge.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
